Reply
 
Thread Tools Display Modes
  #1   Report Post  
lazyadm1n
 
Posts: n/a
Default CRC vs CLC?

Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power supply?

Asides from the higher cost and larger size of a choke, is there any reason
to choose one over the other?

One more question about chokes has to do with sizing. How does one determine
what size choke to use?
_________________
Rod

www.dortoh.ca


  #2   Report Post  
Ronald
 
Posts: n/a
Default

Just one thing : Duncans PSU designer ......
http://www.duncanamps.com/software.html



"lazyadm1n" schreef in bericht
...
Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power

supply?

Asides from the higher cost and larger size of a choke, is there any

reason
to choose one over the other?

One more question about chokes has to do with sizing. How does one

determine
what size choke to use?
_________________
Rod

www.dortoh.ca




  #3   Report Post  
Patrick Turner
 
Posts: n/a
Default



Ronald wrote:

Just one thing : Duncans PSU designer ......
http://www.duncanamps.com/software.html


I tried the free program.

It was very slow to use, and I dunno if it even gives the correct results.

For example, I have a solid state amp I am rebuilding,
and I have 65 vrms from the mains tranny,
and I plan to use solid state rectifiers, 250 mH chokes, and 10,000 uF
caps on the rails.
The idle load is about 200 ohms.

Anyway, I typed all that info into program OK, and
I hit 'simulate' and the voltage at the 10,000 uF ( 10 milli farads ) cap
and 200 ohm load
is shown soaring to +85v just after switch on then settling down to +57
volts, which was about
what I calculated on the back of an envelope.

I am not so sure I have seen the soaring voltage in the second or two
after turn on with an LC input filter.

The program also warned me thatbthe PIV of the diodes had been exceeded by
hundreds of volts,
but hey, I only got 65 volts at the sec....

Any comments ?

Patrick Turner.

  #4   Report Post  
Tim Williams
 
Posts: n/a
Default

"Ronald" wrote in message
...
Just one thing : Duncans PSU designer ......
http://www.duncanamps.com/software.html


Bah... automation ruins the learning process. It's good tool but doesn't
solve errors between the keyboard and operator...

In a word: regulation.

Anything from 10mH for high amperage supplies to 10H on thinner (say, down
to 10-20mA) loads is acceptable. Actually the load doesn't matter, it's the
capacitors - but it's absurd to use 4,700uF and 20mH filtering 10mA of
+300V!
10mH feeding 20uF is useless, because the attenuation is nil.

Typically, the first cap is chosen to be around 500uF per ampere (less for
tube rectifiers), the second cap a bit larger (up to 4 times) to store
energy to supply momentary demands of the amplifier. The choke is typically
chosen to give between 10 and 200 times attenuation (i.e., around -20
to -50dBV) of the ripple on the first cap.

For CRC filters, the resistor is chosen similarly, or to drop a certain
voltage. If it drops too much for your requirements, you might have to use
more capacitance after it to get sufficiently low hum; likewise, if you need
to drop a lot of, and a certain amount of voltage, you might take that route
instead.

Tim

--
"I've got more trophies than Wayne Gretsky and the Pope combined!"
- Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms


  #5   Report Post  
Fabio Berutti
 
Posts: n/a
Default

Advantages:

1 - The choke has a much lower DC resistance for the same hum attenuation,
hence it doesn't drop voltage unnecessarily
2 - The choke filters high frequencies (diodes commutation "spikes" and/or
radio noise & other HF garbage coming in thru the AC line) very effectively,
'cause its impedance increases with frequency
3 - The choke STORES energy, this is why it smoothens DC much more than a
simple R
4 - The choke dissipates less heat than a R giving comparable filtering
5 - Since the L draws a constant current, it allows tube rectifiers to give
much more in LC filter arrangement than in CLC (the tube is never forced to
give violent current pulses as with a C connected to cathode)

Disadvantages:

1 - the choke is big, heavy and expensive
2 - the choke is a source of electromagnetic noise; if used in a phono
preamp it should be screened (potted), and in any case it should be
installed "rotated" with respect to other magnetic components in order not
to "mix" stray magnetic fields, which is not always easy

Briefly:

Any "serious" PS needs to use a choke. Its value is a matter of $, weight
and requirements. Ordinary power amps are happy with anything between 2 and
10H, the PP being less demanding. The more the current, the smaller the L
value for the same pounds of iron. For preamps a larger value is good, I
used a 40H choke to supply less than 20 mA to my line preamp (a single ECC81
cathode-follower). There's a diode bridge, a 47uF, the choke, another 47uF,
then 2 resistors 1k5 each feeding two 10uF capacitors, finally feeding the
ECC81 anodes. Cap values are small, but the 40H smoothens the power so
effectively that when I put the 'scope probe on the 10uF cap I did not
manage to see anything like a sine wave.

Ciao

Fabio


"lazyadm1n" ha scritto nel messaggio
...
Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power
supply?

Asides from the higher cost and larger size of a choke, is there any
reason to choose one over the other?

One more question about chokes has to do with sizing. How does one
determine what size choke to use?
_________________
Rod

www.dortoh.ca





  #6   Report Post  
Mark
 
Posts: n/a
Default

"lazyadm1n" wrote in message ...
Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power supply?

Asides from the higher cost and larger size of a choke, is there any reason
to choose one over the other?

One more question about chokes has to do with sizing. How does one determine
what size choke to use?
_________________
Rod

www.dortoh.ca


Fabio gave the best advantages and disadvantages of a reactor. I am
one who always uses a reacor in designs. I tolerate the additional
cost and weight. I want a job done right so I do not have to go back
to it.

The easiest way to determine the amount of henries you need is to
divide the mA into the B+ and that gives the minimum amount of
inductance needed. This is done at static currents. The formulas to
get the exact inductance vs. the easier way are a few percent
difference. It is best to go higher in inductance than lower.
  #7   Report Post  
Patrick Turner
 
Posts: n/a
Default



Fabio Berutti wrote:

Advantages:

1 - The choke has a much lower DC resistance for the same hum attenuation,
hence it doesn't drop voltage unnecessarily


I like chokes, and use them in all my power amps.


2 - The choke filters high frequencies (diodes commutation "spikes" and/or
radio noise & other HF garbage coming in thru the AC line) very effectively,
'cause its impedance increases with frequency


True, but a lot of the diode noise is caused by stray inductive and capacitive
pick
up by input leads of the pulses of resonant F created by the leakage L of the
power tranny
and the shunt C available. Often a 0.05 across the HT seconday will
move this F down to one which won't travel.

Where you have a CLC, one can often have a small C + R across the L,
so that this C and the L are resonant at twice the mains F,
and so the L+C form a parallel resonant circuit which
rejects the ripple F much better.
To stop higher harmonics getting thru via the small C, the R in series
is about 20 ohms, and forms a damping R to the tuned circuit.
Such a $2 solution increases the ripple rejection about 15 dB,
so its like using a choke 5 times the value.




3 - The choke STORES energy, this is why it smoothens DC much more than a
simple R


Well, not really. If the impedance of a 1H choke is 628 ohms at 100 Hz,
the choke will filter only as well as a 628 ohm resistance at 100 Hz.
But the choke forms a second order LC filter with C2 of the CLC,
abd *this* is the big advantage with the choke.

However where you have CRCRC, the LF ripple caused by mains voltage
undulations tend to be better filtered than with CLCLC,
so for preamps, I may have a small choke in a CLC to start with,
but then its all CRCRC down to the MC amp.
And I use lots of 470 uF electros with 2 uF plastic caps to bypass them close to
where the
tubes are connected, because one don't want stray RF finding its way around,
and the cascode MC amps I use were prone to RF oscillations until
I made all the leads less than 20mm, ie, treated the circuit as if it was an
RF circuit.
I used choke feeds from the DC supply for the heaters to the 6EJ7
cascoded MC amp tube, with ceramic C bypasses to 0V, just to make sure
RF noise couldn't find its way around the heater circuit.

4 - The choke dissipates less heat than a R giving comparable filtering


This is a major advantage in a power amp, but not so major
in a preamp.


5 - Since the L draws a constant current, it allows tube rectifiers to give
much more in LC filter arrangement than in CLC (the tube is never forced to
give violent current pulses as with a C connected to cathode)


The current in chokes varies.

But what you refer to is the LC filter, not CLC filter.
The LC filter, known as a choke input supply does have a major
benefit; the flow of DC is constant from the rectifiers, and there
are no pulse charges to the C1 of a CLC or CRC type of
capacitor input supply.
So a choke allows tube rectifiers to be used, and tube rectifiers have very
restricted peak current handling, so C1 has to be limited to say
47 uF max in a CLC or CRC supply.

The disadvantage of the LC input is the cost of the input choke,
which has to be very carefully made, to prevent
mecanical hum, and radiated fields and vibration, and
interaction with steel chassis.
And getting the choke to just the right value, gap size,
and DC resistance is a bother.

And the power tranny winding has to have a lot higher voltage to get the same
value of B+.




Disadvantages:

1 - the choke is big, heavy and expensive
2 - the choke is a source of electromagnetic noise; if used in a phono
preamp it should be screened (potted), and in any case it should be
installed "rotated" with respect to other magnetic components in order not
to "mix" stray magnetic fields, which is not always easy


Potting is almost mandatory with LC filters.
But not with CLC filters because the amount of AC flow in the CLC filter
is usually a tiny fraction of the AC in an LC filter.



Briefly:

Any "serious" PS needs to use a choke. Its value is a matter of $, weight
and requirements. Ordinary power amps are happy with anything between 2 and
10H, the PP being less demanding. The more the current, the smaller the L
value for the same pounds of iron. For preamps a larger value is good, I
used a 40H choke to supply less than 20 mA to my line preamp (a single ECC81
cathode-follower). There's a diode bridge, a 47uF, the choke, another 47uF,
then 2 resistors 1k5 each feeding two 10uF capacitors, finally feeding the
ECC81 anodes. Cap values are small, but the 40H smoothens the power so
effectively that when I put the 'scope probe on the 10uF cap I did not
manage to see anything like a sine wave.


High value chokes are not so easy to come by,
so for wherever I can, I prefer CRCRC for preamps, and use a high
B+ to start with to allow for filering down with R.
And if I use a 47 uF off a tube rectifier, I will
have say at least 340 ohms before the next C.
The 47 uF has 34 ohms of reactance at 100 Hz.
To make sure the tube rectifier isn't loaded by even more reactance of a second
C, R should be at least 10 x ZC, so 340 ohms ( or more, whatever suits..)
Then I don't muck around, I use 470 uF for the next C.
Then another R, but its value doesn't have to be large, say
330 ohms, and the DC voltage drop in a preamp won't be large,
and then another 470 uF cap.
470 uF has only 3.4 ohms of reactance, so the 330 ohm
plus 470 uF gives a hum attenuation of 100 times or 40 dB.

So if you had 5 volts of ripple at the 47 uF off the rectifier,
then after two 330 ohms and two 470 uF, the hum
will be 10,000 times lower, or -80 dB, so 0.5 mV.

If the preamp current was only 30 mA, the voltage drop
in 660 ohms is only 20v.

I like to start with a HT winding of 240v on the power tranny, giving
me 340v off the rectifier, then I can afford to
drop down to 280v for the preamp stage supplies.

Caps are cheaper and easier to source than chokes.

If you want more attenuation of the hum, and I would,
then use say 47 uF, 330 ohms, 470 uF, 470 ohms, 470 uF,
and then perhaps have 4k7 and 100 uF to each tubes anode supply point.
This should stop all the stages talking to each other.


Sometimes the use of chokes in LC inputs can solve a problem.

I recently made a supply for a rebuilt solid state amp which had 62vrms
windings.
I didn't want the possible +/- 87 volt rails with a cap input.

With an LC filter, the B+ rail is 0.88 x the rms value of the winding voltage,
if the DC resistances of the choke is low, and the R of the rectifiers is low.

So I figured about +/- 54v for the rails would be just right.
So one has to design the choke to suit the conditions and the pocket.
I wanted 220 mA at idle for the amp, and up to 4 amps at a couple of hundred
watts of output,
using the 54 volt rails.

The smallest current to be drawn by a circuit is where you start,
and the circuit is a load on the supply, and it's highest ohmic value is found
by the wanted B+ divided by the DC current at idle.
This supply is for an AB amp where the current increases, but we must
find the L for the highest value of RL we will have.

Where the mains is 50 Hz, L = RL / 940,
so in my case V = 55v, I = 0.22amps DC, so RL = 55/0.22 = 250 ohms.

So L wanted = 250 / 940 = 0.266 Henrys.

I want the choke to handle up to 4 amps DC,
so I want to have only 4 volts maximum drop in DV due to the choke's DCR,
so after checking out what my little yellow book which is a guide to choke
design,
and doing a run with the Hanna method,
I came up with a choke with wasteless pattern iron, 50 stack, 25mm tongue,
and 345 turns of 1.25mm wire.
This gave me a DC resistance just under one ohm, and hopefully enough L.
One never knows until one is done.

After 2 hrs of careful layer winding and getting all sticky with varnish as I
wound ans assembled it,
I tried this choke in the circuit. with 250 ohms as the load, and 10,000 uF as
the cap
from which the collector circuit would operate.
I arranged a 32 ohm load to simulate a higher load current test.

The air gap has to be set for optimum.

A choke in this situation is called a swinging choke, because
when Idc = 0.22 amps, we want L to be high, and when
its say 2.2 amps, we allow it to reduce due to the increasing DC flow
causing the iron to have a lower U factor, and thus less L.

If you have the gap too large, the amount of L will be too low
when Idc is low, and the input voltage will tend to saturate the core,
and you will get nasty step in the 100 Hz wave form at the
diodes. But if the choke has a gap too small, it might saturate
with DC, at high current, and the smoothing action is minimised.

So I added strips of paper to each side of the Is in the E&I assembly,
and recorded the ripple voltage at the load as the gap was adjusted.

With no gap, the ripple was lowest with low current, and highest with high dc
current.

With more gap, the ripple slightly increased with low current, but fell
considerably with high current.
Then with 4 sheets of paper the value of L fell below the
calculated **critical value** for maintaining B+ = 0.89 v winding vrms,
and it began to saturate, and the wave form at the output of the diodes
looked awful, instead of a nice series of arches.

So I backed off with gapping, and settled for 3 sheets of paper.

This was enough gapping to stop the choke saturating due to too much DC.

Then I measured the value of inductance at 0.22 amps, and 1.9 amp.

This may seem difficult, but it isn't, and approximate measure will do.

The ripple voltage at 0.22 amps DC at the 10,000 uF was measured to be 0.02
vrms, 100 Hz.
The 100 Hz voltage across the choke was 27vrms, so
the reactance of the capacitor is much smaller then the chokes, when you
consider
equal ac current flows through each to ground from the rectifier.

The ratio of voltages is 27 / 0.02 = 1,350.
Now the reactance of the cap is only 0.16 ohms at 100 Hz, so
the reactance of the choke must be 1,350 x 0.16 = 216 ohms,
and since there is 628 ohms per Henry at 100 Hz, the choke's value must be
216 / 628 = 0.34 Henries, which is slightly above
the critical value we were aiming for at the beginning.

At 1.9 amps of dc, the ripple voltage at the 10,000 uF was 0.06 vrms,
and since the voltage across the choke is substantially constant at 27 vrms,
the choke's reactance was calculated at 73 ohms, so
its value has fallen to 0.12 Henrys at the 1.9 amp dc.

Between 0.22 amps and 1.9 amps, the B+ fell from +55v to +53v,
only about 4%, which is quite satisfactory for a class B amp.

With tube amps the same procedure can be followed, and for the same
**power** involved, the choke will have more turns, the cap will be
less uF, the load will be more ohms, but the choke will remain about the same
size.

The RDH4 has the Hanna method spelled out fairly clearly, but you really
have to guess about what you need for size to begin, and that comes from
experience.
The Hanna method allows the turns to be precisely calculated for the least
DC drop, and suitablity for the current change.

As the value of inductance is reduced below the critical value for a given DC
current,
the B+ will rise, so that when L = 0.0, it will have risen to
a max of 1.41 x the rms value of the winding voltage.

With swinging choke situations like the one I have exampled,
one has to allow for at least 10% of the maximum current to be drawn
by the AB amp to prevent the B+ from reaching the unwanted value
of a cap input filter, which would explode the electrolytics
if they were not rated well.
Often a "bleeder" resistance is used, like the 250 ohms in the example above.
At idle, such an R dissipates 15 watts, so it needs to be rated for 50,
to prevent it fatiguing, going open, and allowing B+ to soar too high.

Although I have given a solid state example, choke inputs are best in tube amps.

The B+ regulation for a class AB amp is poorer for a given
tube rectifier and C than when the same rectifier is used with a choke
in front of the C and the winding voltage is made higher to
get the same B+.

The lower the DCR of the choke, the less it heats up.

The AC dissipates very little heat in the choke.
The DC current is the main cause of heat.
Dissipation = I x I x R and so if I is doubled, the heat increases 4 times.

In a class A amp, the current to the amp never varies, so there is no point
in seeking the better regulation of a choke input, and so
the CLC can be used, and the choke need only be a fraction of the size
as used for the LC input.

Now if you all followed all that, we give out medals...

Lodge your complaints about electro-magnetism with
God Of Triodes,
69 Gold Street,
Heaven.

Patrick Turner.



  #8   Report Post  
N. Thornton
 
Posts: n/a
Default

"Fabio Berutti" wrote in message ...

Any "serious" PS needs to use a choke.


While your list of pros was very good, I'm not personally convinced
about the above claim. There are various ways to catch a rat, and CLC
supplies arent the only one. Modern amps almost never use them...
because they are not essential IMHO.

Regards, NT
  #9   Report Post  
N. Thornton
 
Posts: n/a
Default

Patrick Turner wrote in message ...
Ronald wrote:


Just one thing : Duncans PSU designer ......
http://www.duncanamps.com/software.html


I tried the free program.

It was very slow to use, and I dunno if it even gives the correct results.

For example, I have a solid state amp I am rebuilding,
and I have 65 vrms from the mains tranny,
and I plan to use solid state rectifiers, 250 mH chokes, and 10,000 uF
caps on the rails.
The idle load is about 200 ohms.

Anyway, I typed all that info into program OK, and
I hit 'simulate' and the voltage at the 10,000 uF ( 10 milli farads ) cap
and 200 ohm load
is shown soaring to +85v just after switch on then settling down to +57
volts, which was about
what I calculated on the back of an envelope.

I am not so sure I have seen the soaring voltage in the second or two
after turn on with an LC input filter.

The program also warned me thatbthe PIV of the diodes had been exceeded by
hundreds of volts,
but hey, I only got 65 volts at the sec....

Any comments ?

Patrick Turner.


Yes: you're learning something.

NT
  #10   Report Post  
Fabio Berutti
 
Posts: n/a
Default

I admit, I used a categoric statement which is not necessarily true. There
are stabilised PS and many more "engineered" designs to do the job. I was
thinking only about a comparison among CLC and CRC "plain" filters. As per
the modern amps, I suppose that size, weight, appearence and $ have
something to do with the reduction in the kg of Si-iron used in these
designs (sand-state devices are excluded: I guess that, because of their
high current and low voltage requirements, a choke would need to be some 50
pounds to do its job, while a 317 or the like will cost a song and work
perfectly).

Ciao

Fabio


"N. Thornton" ha scritto nel messaggio
om...
"Fabio Berutti" wrote in message
...

Any "serious" PS needs to use a choke.


While your list of pros was very good, I'm not personally convinced
about the above claim. There are various ways to catch a rat, and CLC
supplies arent the only one. Modern amps almost never use them...
because they are not essential IMHO.

Regards, NT





  #11   Report Post  
Fabio Berutti
 
Posts: n/a
Default


Lodge your complaints about electro-magnetism with
God Of Triodes,
69 Gold Street,
Heaven.

Patrick Turner.



Inter-continental post fare applied, isnt'it?

Fabio


  #12   Report Post  
Patrick Turner
 
Posts: n/a
Default



"N. Thornton" wrote:

"Fabio Berutti" wrote in message ...

Any "serious" PS needs to use a choke.


While your list of pros was very good, I'm not personally convinced
about the above claim. There are various ways to catch a rat, and CLC
supplies arent the only one. Modern amps almost never use them...
because they are not essential IMHO.


The reason they are not used today is cost,
and the availability of reliable large value electros.
But I still use chokes, and they are still very effective,
and reliable, in comparison to active methods of ripple filtering.

They had more favour when electros were not so reliable,
expensive and large.

But I even got a choke to work well in an LC input filter in a solid state amp supply.
Nobody is supposed to that these days, but I find its works very well.

I did try a sodium lamp ballast but although the ballast was large,
its DCR was more than a smaller one that I wound myself, and the gap was way to big,
to make sure the choke never saturated, and the value of inductance
wouldn't "swing" down from a critical high value to a lower value with increasing DC.

Patrick Turner.




Regards, NT


  #13   Report Post  
Patrick Turner
 
Posts: n/a
Default



"N. Thornton" wrote:

Patrick Turner wrote in message ...
Ronald wrote:


Just one thing : Duncans PSU designer ......
http://www.duncanamps.com/software.html


I tried the free program.

It was very slow to use, and I dunno if it even gives the correct results.

For example, I have a solid state amp I am rebuilding,
and I have 65 vrms from the mains tranny,
and I plan to use solid state rectifiers, 250 mH chokes, and 10,000 uF
caps on the rails.
The idle load is about 200 ohms.

Anyway, I typed all that info into program OK, and
I hit 'simulate' and the voltage at the 10,000 uF ( 10 milli farads ) cap
and 200 ohm load
is shown soaring to +85v just after switch on then settling down to +57
volts, which was about
what I calculated on the back of an envelope.

I am not so sure I have seen the soaring voltage in the second or two
after turn on with an LC input filter.

The program also warned me thatbthe PIV of the diodes had been exceeded by
hundreds of volts,
but hey, I only got 65 volts at the sec....

Any comments ?

Patrick Turner.


Yes: you're learning something.


So what do you think I'm learning?

Patrick Turner.



NT


  #14   Report Post  
Patrick Turner
 
Posts: n/a
Default



Fabio Berutti wrote:

Lodge your complaints about electro-magnetism with
God Of Triodes,
69 Gold Street,
Heaven.

Patrick Turner.


Inter-continental post fare applied, isnt'it?


Yeah, and they apply HST as well ( HST is like a goods and services tax,

or VAT, only in this case is a Heaven Sent Tax..)

Economic rationalism is going on up there too ;-)

Patrick Turner.



Fabio


  #15   Report Post  
N. Thornton
 
Posts: n/a
Default

Patrick Turner wrote in message ...
"N. Thornton" wrote:

Patrick Turner wrote in message ...
Ronald wrote:


Just one thing : Duncans PSU designer ......
http://www.duncanamps.com/software.html

I tried the free program.

It was very slow to use, and I dunno if it even gives the correct results.

For example, I have a solid state amp I am rebuilding,
and I have 65 vrms from the mains tranny,
and I plan to use solid state rectifiers, 250 mH chokes, and 10,000 uF
caps on the rails.
The idle load is about 200 ohms.

Anyway, I typed all that info into program OK, and
I hit 'simulate' and the voltage at the 10,000 uF ( 10 milli farads ) cap
and 200 ohm load
is shown soaring to +85v just after switch on then settling down to +57
volts, which was about
what I calculated on the back of an envelope.

I am not so sure I have seen the soaring voltage in the second or two
after turn on with an LC input filter.

The program also warned me thatbthe PIV of the diodes had been exceeded by
hundreds of volts,
but hey, I only got 65 volts at the sec....

Any comments ?

Patrick Turner.


Yes: you're learning something.


So what do you think I'm learning?



I guess that there are complications that you might have not
appreciated before.

Take overshoot: the L acts like a flywheel. The switch on current
surge into C2 puts a fair bit of energy into the choke, and it tries
to keep that current going when C2 has reached working V, so you get
overshoot.

Diode voltages: Firstly a 65v secondary will give you much more than
65v. 65v rms is ballpark 100v peak. If the transformer has say 10%
regulation, then when unloaded that will actually be 110v peak. It
will be unloaded during V overshoot. Now add to that mains borne
noise, spikes etc. Now add the fact that each diode will see that peak
transformer's output when the cap V is of opposite polarity, so that
adds another V_overshoot onto your diodes V_it_handles.

A small cap across the TF secondary can reduce noise and spikes. It
acts as an RC filter.


NT


  #16   Report Post  
TubeGarden
 
Posts: n/a
Default

Hi RATs!

I have used PSUD2 to tune lots of PSs.

It seems very good.

It is a bit reactionary about topology

Did you use a slide rule or an abacus for your back of the envelope
calculations? :^)

Happy Ears!
Al


Alan J. Marcy
Phoenix, AZ

PWC/mystic/Earhead
  #17   Report Post  
Patrick Turner
 
Posts: n/a
Default



"N. Thornton" wrote:

Patrick Turner wrote in message ...
"N. Thornton" wrote:

Patrick Turner wrote in message ...
Ronald wrote:


Just one thing : Duncans PSU designer ......
http://www.duncanamps.com/software.html

I tried the free program.

It was very slow to use, and I dunno if it even gives the correct results.

For example, I have a solid state amp I am rebuilding,
and I have 65 vrms from the mains tranny,
and I plan to use solid state rectifiers, 250 mH chokes, and 10,000 uF
caps on the rails.
The idle load is about 200 ohms.

Anyway, I typed all that info into program OK, and
I hit 'simulate' and the voltage at the 10,000 uF ( 10 milli farads ) cap
and 200 ohm load
is shown soaring to +85v just after switch on then settling down to +57
volts, which was about
what I calculated on the back of an envelope.

I am not so sure I have seen the soaring voltage in the second or two
after turn on with an LC input filter.

The program also warned me thatbthe PIV of the diodes had been exceeded by
hundreds of volts,
but hey, I only got 65 volts at the sec....

Any comments ?

Patrick Turner.


Yes: you're learning something.


So what do you think I'm learning?


I guess that there are complications that you might have not
appreciated before.

Take overshoot: the L acts like a flywheel. The switch on current
surge into C2 puts a fair bit of energy into the choke, and it tries
to keep that current going when C2 has reached working V, so you get
overshoot.


There actually isn't any overshoot beyond the peak voltage of the rectifier.

What I do see on the CRO with only the bleeder R in place,
is a set of arches you expect to see with a full wave rectifier feeding a choke.
But imposed on each positive going arch half there is a vertical section up strating
near the bottom of the +ve arch, and then a level
which joins to the rest of the arch about 2/3 the way up the arch curve.
There is ringing at the up then across angle.

Methinks that while the charge up of the cap takes place,
the L is reduced to below the critical value * for the value of RL connected*

When a much lower RL of even below 16 ohms instead of 250 ohms is connected,
no such phenomena is encountered.
suddenly connecting 16 ohms, then releasing it also causes the distorted waves between the
diodes and choke.
I am using SS diodes.

And the rise in voltage at the C2 of the choke input isn't instantaneous,
it takes 3 seconds to get up to the right voltage, and my meters don't
indicate a soar in the voltage like Duncan says; I see the voltage just ramping up slowly.

But sure, there is more current during the charge cycles until it gets up to working voltage.


Diode voltages: Firstly a 65v secondary will give you much more than
65v. 65v rms is ballpark 100v peak.


65 x 1.414 = 91.9v, and less 0.7v for the diode drop,
I saw 91.7 volts maximum with no choke, and no load.

If the transformer has say 10%
regulation, then when unloaded that will actually be 110v peak.


You are 20 volts out.

The 940 VA tranny I have is very well regged, and I measured
only about 2v drop with a load change from 220 mA to 1.9 A.
Some of that is the mains drop.

It
will be unloaded during V overshoot. Now add to that mains borne
noise, spikes etc. Now add the fact that each diode will see that peak
transformer's output when the cap V is of opposite polarity, so that
adds another V_overshoot onto your diodes V_it_handles.

A small cap across the TF secondary can reduce noise and spikes. It
acts as an RC filter.


Correct. I use such a cap routinely, and this amp will have one.

But I think the phenomena is to do with the critical L
falling below the right value for the low current bleeder R.

Reducing the bleeder R pproduces the distorted wave permanently.

None such phenomena is mentioned in RDH4, and I have never
heard anyone suggesting this ****e happens before.

However, in some old circuits I see a snubber RC across the L,
even with tube rectifiers,
which should work OK to absorb some switching energy
caused by what I think is a back emf function.

The switching could be a noise problem for the amp.

I wound the second choke, since there are two rails,
and I have yet to bend up some sheet metal boxes,
pop rivent them together with the choke inside,
and pour in some molten pitch.
That should get the chokes to shut the FU
with mechanical noise, and prevent radiated
spuriae, and I probably will use some coax leads to the chokes from where the diodes are,
plus snubbers across the diodes.

Even SS amps are prone to stray noise pick up.

The amp used to be a Phase Linear, and with +/-90 volt rails.
It had the reputation for being a Flame Linear, so itsa gonna be Turnerized,
with lots of MJL21193/94 massive output transistors,
and sensible fusing and active protection circuitry.

It will operate from +/-55v, and be good for 225 watts into 6 ohms,
or a ****eload more when bridged for a sub.

Originally, with 90 volt rails, the theoretical
instantaneous max po into 4 ohms was 945 watts,
with 1,890 watts into 8 ohms bridged.
No need for that, and I prefer the easier working with
the lower rails, more class A before crossover to AB.
With 5 x output transistors in emitter follower each side of the circuit,
its thd without loop fedback should be low, and I am expecting 0.005%
at 200 watts with FB; this is needed with SS, since a small %
of thd sounds crook...
The EF outputs are powered by darlington EF drivers,
a pair of MJE15030/31,
then these driven by a complementary pair VAS,
driven by a symetrical pair of NPN and PNP diff pairs,
which are effectively in parallel.
I might use BF469/70 for the pairs and VAS, since they are fast, linear
and I've done it before, and all the bjts are commonly available.

I had to add 100 mm to the depth of the case to make room
for the chokes, and have plenty of access for service;
not so darn crammed like the original.

When these amps had bad fall, all the king's men
were no help.
The 940 VA mains tranny is so heavy, that if it falls right,
it can easily bend the 1.2 mm thick Al casing, resulting in a mangled mess.
Really bad original case engineering indeed, but the power tranny
and the electros seem OK.
The original diode bridge shunted itself as soon as I tried the choke input.

The alternative to the chokes would have been to replace the mains
tranny, or instal a 120v to 42-0-42 toroidal to make about +/-56v rails.
Then I could have had a CLC supply yo each rail.

I have two of these old amps, maybe the next one
I will do differently.

But I gotta try this, and see if the pundits are right about the
better sound.

It feels like a holiday working on a big BJT amp.
Not a hot cranky tube in sight.
But a fukkin bus driver's holiday it is!


Patrick Turner.





NT


  #18   Report Post  
N. Thornton
 
Posts: n/a
Default

Patrick Turner wrote in message ...
"N. Thornton" wrote:


I guess that there are complications that you might have not
appreciated before.

Take overshoot: the L acts like a flywheel. The switch on current
surge into C2 puts a fair bit of energy into the choke, and it tries
to keep that current going when C2 has reached working V, so you get
overshoot.


There actually isn't any overshoot beyond the peak voltage of the rectifier.


What happens re overshoot in the real world depends on the Rs and
load. Transformer, rectifier and choke all have R. The applied load
will also absorb energy.


What I do see on the CRO with only the bleeder R in place,
is a set of arches you expect to see with a full wave rectifier feeding a choke.
But imposed on each positive going arch half there is a vertical section up strating
near the bottom of the +ve arch, and then a level
which joins to the rest of the arch about 2/3 the way up the arch curve.
There is ringing at the up then across angle.


I didnt follow that, too multi-interpretable for my few remaining
cells.


And the rise in voltage at the C2 of the choke input isn't instantaneous,
it takes 3 seconds to get up to the right voltage, and my meters don't
indicate a soar in the voltage like Duncan says; I see the voltage just ramping up slowly.


did you see that on scope or meter? I'd be surprised if it took as
long as 3s.


But sure, there is more current during the charge cycles until it gets up to working voltage.


Diode voltages: Firstly a 65v secondary will give you much more than
65v. 65v rms is ballpark 100v peak.


65 x 1.414 = 91.9v, and less 0.7v for the diode drop,
I saw 91.7 volts maximum with no choke, and no load.


you dont get 0.7v diode drop, thats a famous myth. 0.6v is your diode
knee voltage, not the actual working V drop. Power rectifiers are more
likely to give around 1v to 1.5v drop on load, maybe 2v, down to 0.6v
off load.


If the transformer has say 10%
regulation, then when unloaded that will actually be 110v peak.


You are 20 volts out.


Using your figures, 91.9 / .9 = 102v. But thats only if its a 10%
transformer and its unloaded, neither of which I know.


The 940 VA tranny I have is very well regged, and I measured
only about 2v drop with a load change from 220 mA to 1.9 A.


ok so delta i there is 1.9 - .22 = 1.68A, delta v 2v, apx.
65v 940W = 14.46A.
delta V over 14.5A will be 2v x 14.5/1.68 = 17.26v

As a %age of 65v thats 26%
So according to your rough figs your TF has around 25% regulation. Not
very good for a 1kw device, but quite workable. If it has 2x 65v
outputs the figs will be a bit different, but either way, real world
transformers do have regulation that is not particularly good.

If its really 25% that would explain the slow start up.


Some of that is the mains drop.


Mains supply impedance is going to be trivial compared to transformer
regulation. Assuming youve not got it on a small unregulated generator
or something.


It
will be unloaded during V overshoot. Now add to that mains borne
noise, spikes etc. Now add the fact that each diode will see that peak
transformer's output when the cap V is of opposite polarity, so that
adds another V_overshoot onto your diodes V_it_handles.

A small cap across the TF secondary can reduce noise and spikes. It
acts as an RC filter.


Correct. I use such a cap routinely, and this amp will have one.

But I think the phenomena is to do with the critical L
falling below the right value for the low current bleeder R.


?

Reducing the bleeder R pproduces the distorted wave permanently.

None such phenomena is mentioned in RDH4, and I have never
heard anyone suggesting this ****e happens before.


maybe a diagram of it? ascii art? have little idea what you mean
otherwise.


However, in some old circuits I see a snubber RC across the L,
even with tube rectifiers,
which should work OK to absorb some switching energy
caused by what I think is a back emf function.


switching energy? back emf function?


plus snubbers across the diodes.


that sure does reduce crap, plus reduce the D PIV.


Originally, with 90 volt rails, the theoretical
instantaneous max po into 4 ohms was 945 watts,


I thought 945w was the power rating of the mains tf?
The 940 VA mains tranny is so heavy, that if it falls right,

Coincidence?


With 5 x output transistors in emitter follower each side of the circuit,
its thd without loop fedback should be low, and I am expecting 0.005%
at 200 watts with FB; this is needed with SS, since a small %
of thd sounds crook...


yep - and mfr's figs are unrealistic anyway. Typical tr amp thd is
specified at 1kHz, by the time you get up to 20kHz it will have gone
up massively. I think that may be the basic reason why a 0.1% valve
amp can beat the pants off a 0.003% tranny amp.


The 940 VA mains tranny is so heavy, that if it falls right,
it can easily bend the 1.2 mm thick Al casing, resulting in a mangled mess.
Really bad original case engineering indeed, but the power tranny
and the electros seem OK.


Yeah, though to be fair, housing a 1kw transformer is never gonig to
be a doddle. Could sure do better than thin ali though: maybe it was
designed for portability first.


The alternative to the chokes would have been to replace the mains
tranny, or instal a 120v to 42-0-42 toroidal to make about +/-56v rails.


or possibly wind a bit more on the primary - if theres enough space.
Of course there more often isnt. Very occasionally I've seen italian
mains TFs also come with a 160v primary tap, but those are rarer than
hens' teeth now.


NT
  #19   Report Post  
John Stewart
 
Posts: n/a
Default



lazyadm1n wrote:

Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power supply?

Asides from the higher cost and larger size of a choke, is there any reason
to choose one over the other?

One more question about chokes has to do with sizing. How does one determine
what size choke to use?
_________________
Rod

www.dortoh.ca


With the CLC filter there will be a resonance, possibly in the AF range.
I used Patrick's proposed values in this simulation. See the results at ABSE.

The resonance appears at about 4.5 Hz. This will have some effect
on the amp's sound. Quite common in most amps using this configuration,
whether tubed or SS.

On the other hand the CRC filter response is smooth but the power
losses are high.

Take your pick. There are no free rides while using simpler topologies.
A relatively easy way out would be a SS filter/regulator. I favor things
like the Int. Rectifier FETs to do the work.

Good Luck with your project, John Stewart


  #20   Report Post  
N. Thornton
 
Posts: n/a
Default

Patrick Turner wrote in message ...
"N. Thornton" wrote:
"Fabio Berutti" wrote in message ...


Any "serious" PS needs to use a choke.


While your list of pros was very good, I'm not personally convinced
about the above claim. There are various ways to catch a rat, and CLC
supplies arent the only one. Modern amps almost never use them...
because they are not essential IMHO.


The reason they are not used today is cost,
and the availability of reliable large value electros.
But I still use chokes, and they are still very effective,
and reliable, in comparison to active methods of ripple filtering.

They had more favour when electros were not so reliable,
expensive and large.

But I even got a choke to work well in an LC input filter in a solid state amp supply.
Nobody is supposed to that these days, but I find its works very well.

I did try a sodium lamp ballast but although the ballast was large,
its DCR was more than a smaller one that I wound myself, and the gap was way to big,
to make sure the choke never saturated, and the value of inductance
wouldn't "swing" down from a critical high value to a lower value with increasing DC.

Patrick Turner.



Amps with choke psus can work well: after all, the Quad II used them.
But things have moved on, and you can now get much better performance
from a solid state regulated psu, and at lower cost and weight. I
would say if youre really serious about quality one would go with a
proper regulated psu, not the cruder lower performance choke psu.

The downsides of choke psus compared to regulated ones a
very poor regulation
significant impedance
size
weight
cost
resonance
and patchy rejection

In the past reg psus were not much used simply because they were too
pricey. They typically used neon tubes as the Vref, and the pass valve
would have to drop substantial voltage, making the whole setup very
inefficient. Much cruder partial-regulating schemes were sometimes
used like swinging chokes, bleed resistors, baretters, etc.


NT


  #21   Report Post  
TubeGarden
 
Posts: n/a
Default

H RATs!

If you are really serious about quality, you will keep building PS circuits
until you find one that pleases your ears

Unfortunately, no physical dimension correlates perfectly with "I like the
sound."

From a personal standpoint, I have discovered that the joys of audio far
outweigh the efficiencies of retail engineering.

Yes, there have been some nice amps marketed.

No, none of them went very deep into any of the applicable technologies.

I am disabled, and gravel crazy, but sometimes a great notion passes between
these ears.

It is OK for some folks to create realistic designs and workaday
implementations. It seems harmless if some of us just hook stuff up and Listen


Quest into the Unknown!

oh, yeah, and ...

Happy Ears!
Al


Alan J. Marcy
Phoenix, AZ

PWC/mystic/Earhead
  #22   Report Post  
Patrick Turner
 
Posts: n/a
Default



John Stewart wrote:

lazyadm1n wrote:

Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power supply?

Asides from the higher cost and larger size of a choke, is there any reason
to choose one over the other?

One more question about chokes has to do with sizing. How does one determine
what size choke to use?
_________________
Rod

www.dortoh.ca


With the CLC filter there will be a resonance, possibly in the AF range.
I used Patrick's proposed values in this simulation. See the results at ABSE.


Nowhere did I ever suggest all the values for the CLC filter shown in the
test schematic at your post at ABSE.

The values you posted to gain the results for resonance were :-

PS generator resistance = 1kOhm,
C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.
I assume your model used the questionable Duncan Amps freely
downloadable power supply designer program which I found gave me some
incomprehensible results.

The schematic you posted is misleading to those who
may think that 10mF = 10 uF!!!!!!


I recall the discussions revolved around a
very low generator impedance, ie, a large SS amp
power tranny feeding through SS diodes into an LC filter,
with approx 250 mH and 10,000 uF.

Fo = 3.18 Hz.

The 200 ohms is not low enough to damp the LC resonance,
but for critical damping, 7 ohms is required, so the 1.0 ohm dcR of the choke
effectively in series with the L will still leave the Q of the filter somewhat
higher than desirable,
but I found the voltage stability at the 10,000 uF didn't bounce around
when intermittently shunting the 200 ohm bleeder R with 32 ohms.

In the case of an LC filter, any resonance of the LC circuit
is damped when the value of the inductance changes considerably
with a varying charge rate into the C.
In my case with the LC filters for this SS amp,
the tested L for a 250 ohm load with a +55v rail was 0.35 H, or 350 mH.
When 32 ohms was added to the 250 ohms for total 28 ohms,
Idc became 1.9 amps dc, and L became 0.12 H, or 120 mH.

But the test circuit showed no sign of wild resonance voltage swings.
if there were, the choke goes from being higher L to a low value and which
saturates
for part of the rectifier wave forms, and becomes a very much lower
value of L when it does, so resonance seems damped
since L does not stay at a constant value.

Its not just simple filter theory which is applicable here in this case of
LC power supply filter.


The amp in question will be used as a sub amp, but will still have
input filtering to give a pole at 10 Hz.

I still don't believe the resonance will affect the sound one iota.

The resonance appears at about 4.5 Hz. This will have some effect
on the amp's sound.


There may be some amps which have a resonance in the CLC filter,
or LC filter ot CLCLC filter which may be at just under 5 Hz.

In none of them have I found there is any audible artifact.
In a PP amp, any 5 Hz ripples are applied to the CT and common mode
rejection prevents their appearance in the output signal.
But even in SE amps, I sure don't get such problems in my amps.

I have always recommended the use of very large value C2 in a CLC
filter of a tube amp, 470 uF being typical.

If L = 2H, then the Fo between the 470 uF and the 2H is 5.2 Hz.
At 5.2 Hz, ZC and ZL = 62.8 ohms.
The RL may typically be 1600 ohms, which won't be low enough
to damp the LC resonance critically.
But the L will have typically 50 Ohms of dcR, which is effectively in with the L
value,
and approaching the value of 87 ohms required for a -3 dB maximally flat filter
response with a pole
at 5.2 Hz.
So the resonance Q of the filter will be very low, and I have never found any
LF instability of audible effects using the values just quoted, not even when L =
1 H.
In the last pair of SE amps I sold last month I used
C1 = 470 uF,
L = 1H,
C2 = 940 uF.
The dcR of the choke was only 30 ohms,
and although the LC filter was underdamped, and should have
given a moderately high Q resonance at 5.2 Hz, the amps showed no
resonance problems. The value of critical damping R
would be ZL x l.41, or 32.6 ohms x 1.41, or 46 ohms, and
the dcR of the choke is 30 ohms, so almost enough for
perfect damping.
Should anyone want to counter the effects of resonance better than this,
then add more series R as well as retain the choke.
I didn't want to do this because the B+ would have had to be derived from a higher
tap
on the HT winding, and placed the B+ value at C1 perilously close to the voltage
rating of C1, only +450v.



Quite common in most amps using this configuration,
whether tubed or SS.


Very, very few SS amps have CLC filtering to their rails because
most are designed with stupendous values of C, and whatever ripple voltage
appears at the rails does not get into the circuit because of the higgh collector
resistance and the use of typically 50 dB of loop NFB, along
with typically 30 dB of local feedback in the emitter follower config
of most output stages.

I have used a pair of 100,000 uF caps in my own mosfet amp for
300 watts per channel.
No need for any CLC, but others are welcome to try them.

In another SS 50/50 stereo class A amp, I have used C1 = 9,400 uF,
L = about 150mH, C2 = 45,000 uF.
Fo = 1.93 Hz, and ZC = ZL = 1.8 ohms.
The RL = 6.6 ohms, not enough R to damp the resonance,
since the DCR of the choke is 0.5 ohms.
This class A amp has an OPT, and CR coupling, like a tube amp,
and the circuit has a total of 20 dB of NFB,
and there is no resonance problems at LF whatever,
or stability problems because
the circuit open loop gain is well below unity where the LF phase shift
is 180 degrees.

On the other hand the CRC filter response is smooth but the power
losses are high.


The power losses are low, not high with a CLC filter.

Take your pick. There are no free rides while using simpler topologies.
A relatively easy way out would be a SS filter/regulator. I favor things
like the Int. Rectifier FETs to do the work.


People have been using CLC filters with no sonic problems for years without
resorting to solid state regulators, which always do result in a power
loss, because of the voltage drop x load current across the series
pass element, or the B+ voltage x shunt element in the case of a shunt regulator.

Regulators done with SS power fets or well rated bjt like the BU208 waste power,
CLC filters don't, because of the insignificant power lost
in the dcR of the choke, which totals Idc squared x R, in watts.


Good Luck with your project, John Stewart


I don't think the original poster's luck
will be improved by erroneous advice.

Patrick Turner.


  #23   Report Post  
BFoelsch
 
Posts: n/a
Default


"Patrick Turner" wrote in message
...


John Stewart wrote:

lazyadm1n wrote:

Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power
supply?

Asides from the higher cost and larger size of a choke, is there any
reason
to choose one over the other?

One more question about chokes has to do with sizing. How does one
determine
what size choke to use?
_________________
Rod

www.dortoh.ca


With the CLC filter there will be a resonance, possibly in the AF range.
I used Patrick's proposed values in this simulation. See the results at
ABSE.


Nowhere did I ever suggest all the values for the CLC filter shown in the
test schematic at your post at ABSE.





You mean like this,? posted by Patrick Turner on 10/20/2004?

For example, I have a solid state amp I am rebuilding,
and I have 65 vrms from the mains tranny,
and I plan to use solid state rectifiers, 250 mH chokes, and 10,000 uF
caps on the rails.
The idle load is about 200 ohms.






The values you posted to gain the results for resonance were :-

PS generator resistance = 1kOhm,
C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.
I assume your model used the questionable Duncan Amps freely
downloadable power supply designer program which I found gave me some
incomprehensible results.

The schematic you posted is misleading to those who
may think that 10mF = 10 uF!!!!!!


I recall the discussions revolved around a
very low generator impedance, ie, a large SS amp
power tranny feeding through SS diodes into an LC filter,
with approx 250 mH and 10,000 uF.

Fo = 3.18 Hz.

The 200 ohms is not low enough to damp the LC resonance,
but for critical damping, 7 ohms is required, so the 1.0 ohm dcR of the
choke
effectively in series with the L will still leave the Q of the filter
somewhat
higher than desirable,
but I found the voltage stability at the 10,000 uF didn't bounce around
when intermittently shunting the 200 ohm bleeder R with 32 ohms.

In the case of an LC filter, any resonance of the LC circuit
is damped when the value of the inductance changes considerably
with a varying charge rate into the C.
In my case with the LC filters for this SS amp,
the tested L for a 250 ohm load with a +55v rail was 0.35 H, or 350 mH.
When 32 ohms was added to the 250 ohms for total 28 ohms,
Idc became 1.9 amps dc, and L became 0.12 H, or 120 mH.

But the test circuit showed no sign of wild resonance voltage swings.
if there were, the choke goes from being higher L to a low value and which
saturates
for part of the rectifier wave forms, and becomes a very much lower
value of L when it does, so resonance seems damped
since L does not stay at a constant value.

Its not just simple filter theory which is applicable here in this case of
LC power supply filter.


The amp in question will be used as a sub amp, but will still have
input filtering to give a pole at 10 Hz.

I still don't believe the resonance will affect the sound one iota.

The resonance appears at about 4.5 Hz. This will have some effect
on the amp's sound.


There may be some amps which have a resonance in the CLC filter,
or LC filter ot CLCLC filter which may be at just under 5 Hz.

In none of them have I found there is any audible artifact.
In a PP amp, any 5 Hz ripples are applied to the CT and common mode
rejection prevents their appearance in the output signal.
But even in SE amps, I sure don't get such problems in my amps.

I have always recommended the use of very large value C2 in a CLC
filter of a tube amp, 470 uF being typical.

If L = 2H, then the Fo between the 470 uF and the 2H is 5.2 Hz.
At 5.2 Hz, ZC and ZL = 62.8 ohms.
The RL may typically be 1600 ohms, which won't be low enough
to damp the LC resonance critically.
But the L will have typically 50 Ohms of dcR, which is effectively in with
the L
value,
and approaching the value of 87 ohms required for a -3 dB maximally flat
filter
response with a pole
at 5.2 Hz.
So the resonance Q of the filter will be very low, and I have never found
any
LF instability of audible effects using the values just quoted, not even
when L =
1 H.
In the last pair of SE amps I sold last month I used
C1 = 470 uF,
L = 1H,
C2 = 940 uF.
The dcR of the choke was only 30 ohms,
and although the LC filter was underdamped, and should have
given a moderately high Q resonance at 5.2 Hz, the amps showed no
resonance problems. The value of critical damping R
would be ZL x l.41, or 32.6 ohms x 1.41, or 46 ohms, and
the dcR of the choke is 30 ohms, so almost enough for
perfect damping.
Should anyone want to counter the effects of resonance better than this,
then add more series R as well as retain the choke.
I didn't want to do this because the B+ would have had to be derived from
a higher
tap
on the HT winding, and placed the B+ value at C1 perilously close to the
voltage
rating of C1, only +450v.



Quite common in most amps using this configuration,
whether tubed or SS.


Very, very few SS amps have CLC filtering to their rails because
most are designed with stupendous values of C, and whatever ripple voltage
appears at the rails does not get into the circuit because of the higgh
collector
resistance and the use of typically 50 dB of loop NFB, along
with typically 30 dB of local feedback in the emitter follower config
of most output stages.

I have used a pair of 100,000 uF caps in my own mosfet amp for
300 watts per channel.
No need for any CLC, but others are welcome to try them.

In another SS 50/50 stereo class A amp, I have used C1 = 9,400 uF,
L = about 150mH, C2 = 45,000 uF.
Fo = 1.93 Hz, and ZC = ZL = 1.8 ohms.
The RL = 6.6 ohms, not enough R to damp the resonance,
since the DCR of the choke is 0.5 ohms.
This class A amp has an OPT, and CR coupling, like a tube amp,
and the circuit has a total of 20 dB of NFB,
and there is no resonance problems at LF whatever,
or stability problems because
the circuit open loop gain is well below unity where the LF phase shift
is 180 degrees.

On the other hand the CRC filter response is smooth but the power
losses are high.


The power losses are low, not high with a CLC filter.

Take your pick. There are no free rides while using simpler topologies.
A relatively easy way out would be a SS filter/regulator. I favor things
like the Int. Rectifier FETs to do the work.


People have been using CLC filters with no sonic problems for years
without
resorting to solid state regulators, which always do result in a power
loss, because of the voltage drop x load current across the series
pass element, or the B+ voltage x shunt element in the case of a shunt
regulator.

Regulators done with SS power fets or well rated bjt like the BU208 waste
power,
CLC filters don't, because of the insignificant power lost
in the dcR of the choke, which totals Idc squared x R, in watts.


Good Luck with your project, John Stewart


I don't think the original poster's luck
will be improved by erroneous advice.

Patrick Turner.




  #24   Report Post  
Patrick Turner
 
Posts: n/a
Default



TubeGarden wrote:

H RATs!

If you are really serious about quality, you will keep building PS circuits
until you find one that pleases your ears


And I find the best sound comes from the application of
engineering principles with power supplies;
ie, to give the tubes, or SS a chance, have a PS with
sufficiently low impedance, carefully measured out,
not guessed, at all.


Unfortunately, no physical dimension correlates perfectly with "I like the
sound."


I am not sure about that.

What is chosen to be used for a sound system is a list
of many physical characteristics of the active devices,
passive components, and the topology, along with the room,
its furnishings, size, speaker placement, speakers, etc.....

Experiment long enough, and you get a sound that is better than when you started
if you are lucky, but it don't always happen.
A bloke I know just built a pair of boxes for his full range Lowther PM8
drivers for which he paid a pretty penny.
One was a horn design he found on the web some place,
the other was a vented box.
The amps are a single 6BQ5 in triode, no NFB.
He complained to me that there was no bass, and the treble
sounded harsh and bright, compared to Paradigm speakers he aleady has,
but which are not sensitive.

I just measured the system.
The vented box gives a response that falls at 3 dB per octave below 2 kHz.
The horn showed the same response, but with a 4 dB hump in the F with
its centre at 130 Hz.

The sound is muddy as well as bright, and only listenable
with single instruments.

I will have to make him a passive RC filter which can be placed in series
with the source and power amps to make the speakers listenable.

He will never guess the RC values in a million years, and his ears
are totally useless to get him out of the hole he's in with his DIY efforts.

The Paradigm measured quite well by comparison,
and even sounded OK with the SE amps, used at low volume,
since 6BQ5 in triode only makes about 1.5 watts.

In his bedroom, he has a diy pair of speakers with a couple of cheap 4" full range
speakers with
what appear to have a low Qts, and suitable for use with about
20 litre vented boxes, which he has.
The amp is a 10 watt /channel SS mini system, with an all chromed plastic
front panel.
It sounds way better than the SETs and Lowthers; the sound has enough bass,
there is little distortion, and the sound at least some imaging,
is musical, open, and listenable, even using a cheapo DVD
recorded someplace in Thailand on probably what was not
too good a system....


From a personal standpoint, I have discovered that the joys of audio far
outweigh the efficiencies of retail engineering.

Yes, there have been some nice amps marketed.

No, none of them went very deep into any of the applicable technologies.


Some went a fair depth, imho, and produced far more listenable items than
any DIYer without any science in his thinking could ever hope to achieve.


I am disabled, and gravel crazy, but sometimes a great notion passes between
these ears.

It is OK for some folks to create realistic designs and workaday
implementations. It seems harmless if some of us just hook stuff up and Listen


Well harmless it may be, sure, but countless times I have had to listen to what a
bloke hooks up
and to what he things is a great sounding accomplishment,
and at times I have had to be ever so polite to suggest he keep trying....,
or that perhaps something is not fully optimized,
rather than say it all sounds like crap, unless he does, which just happened
twice, one being the example above, the other being a bloke
with a sub active pre-filter unit being quite faulty, like almost everything
people buy at a cheap price on ebay.

Even with a recording of the take off of a Space Shuttle mission, he could barely
rattle a window, with all the gain turned up
using a 1,200 watt amp and a huge 15" sub.
Measurements revealed all his problems.

Usually the crap sound is due to a poor F response in the tube amps with no FB,
and FB is something that *only* science will make right, lest clouds of smoke
be seen. DIYers often miss the fact that some transformers they think are real
cool
need resistive damping to make the reponse flat.
Then there are speakers. Speakers do take some rigously scientific
application to even start to get good sound. Without science, one has to be fond
of a perpetual
effort at the cut and try method.
The ears are still the final arbiter, and some engineers appear to have no ears;
which is sad, but the best engineers go to live concerts where only
accoustic instruments are used, and it trains them to
only be happy when their engineering gives them the best fidelity, without being
overly complex, eg, 49 items in a cross over when 5 would be fine.


Quest into the Unknown!


Ultimately, we are all headed into that quest.

oh, yeah, and ...


may you keep the home triodes burning for many many years.........



Happy Ears!
Al

Alan J. Marcy
Phoenix, AZ

PWC/mystic/Earhead


Patrick Turner.


  #25   Report Post  
Patrick Turner
 
Posts: n/a
Default



Nowhere did I ever suggest all the values for the CLC filter shown in the
test schematic at your post at ABSE.


You mean like this,? posted by Patrick Turner on 10/20/2004?

For example, I have a solid state amp I am rebuilding,
and I have 65 vrms from the mains tranny,
and I plan to use solid state rectifiers, 250 mH chokes, and 10,000 uF
caps on the rails.
The idle load is about 200 ohms.


I believe this is what JH was describing, but a little inaccurately.

We have some way to go before reaching the 20th month of 2004.

The God Of Triodes may issue all of us a spare extra 8 mths,
which I need to get through the work I have, but only if we don't misbehave.

Patrick Turner.



  #26   Report Post  
Patrick Turner
 
Posts: n/a
Default



"N. Thornton" wrote:

Patrick Turner wrote in message ...
"N. Thornton" wrote:
"Fabio Berutti" wrote in message ...


Any "serious" PS needs to use a choke.


While your list of pros was very good, I'm not personally convinced
about the above claim. There are various ways to catch a rat, and CLC
supplies arent the only one. Modern amps almost never use them...
because they are not essential IMHO.


The reason they are not used today is cost,
and the availability of reliable large value electros.
But I still use chokes, and they are still very effective,
and reliable, in comparison to active methods of ripple filtering.

They had more favour when electros were not so reliable,
expensive and large.

But I even got a choke to work well in an LC input filter in a solid state amp supply.
Nobody is supposed to that these days, but I find its works very well.

I did try a sodium lamp ballast but although the ballast was large,
its DCR was more than a smaller one that I wound myself, and the gap was way to big,
to make sure the choke never saturated, and the value of inductance
wouldn't "swing" down from a critical high value to a lower value with increasing DC.

Patrick Turner.


Amps with choke psus can work well: after all, the Quad II used them.


Indeed, Quad II does have a choke, but its for the screen supply.
meanwhile the anode supply has only 16 uF from the tube rectifier,
and there is about 17 vrms of sawtooth ripple voltage at the CT of the OPT.
Its applied to the circuit in common mode, so the IMD caused by
injection of all this noise is no more than the THD measured in thse amps,
which can vary by as much as 20 dB depending on whether the tubes are matched or not,
so I have found.

So with Quad II, I like to remove the existing large low value C, and
insert mdern 47 uF off the rectifier, then 1.7H lamp ballast choke, then 100 uF
at the CT, thus reducing the hum at the CT to utterly negligible levels.
The IMD caused by PS noise is eliminated.


But things have moved on, and you can now get much better performance
from a solid state regulated psu, and at lower cost and weight.


But not in a Quad II amp, without having a considerable V drop,
and the unreliablity of the SS regulator.



I
would say if youre really serious about quality one would go with a
proper regulated psu, not the cruder lower performance choke psu.


Just use the right values of L&C.
No need for all that SS regulation.
Use of SS diodes is fime though, and a voltage doubler
with SS and CLC will have better reg than tube diodes and LC input filter.

Regulation in tube amps is only needed for class AB amps where they are used hard,
and where the amp is mainly class B.
if you don't believe me, try clipping a meter on the B+
of an average class AB amp B+, and take the music up to a level where
clipping only just starts to be visible on the CRO.
Usually, a supply of +450v moves only few volts.
With a sine wave input, sure, the B+ will move maybe 10%,
but that is still OK.





The downsides of choke psus compared to regulated ones a
very poor regulation


Not if the PT is low R, and diode R is low, and DCR of L is low.

significant impedance


Not if the L is followed by a large enough C.


size


A regulator may use a heatsink, itself needing to be insulated.


weight


OK, but a choke is only one simple element.


cost


Depends how one is set up.

I wind my own, re-cycling old tranny cores...


resonance


Eliminated by those who know what they are doing.


and patchy rejection


Of what?



In the past reg psus were not much used simply because they were too
pricey. They typically used neon tubes as the Vref, and the pass valve
would have to drop substantial voltage, making the whole setup very
inefficient. Much cruder partial-regulating schemes were sometimes
used like swinging chokes, bleed resistors, baretters, etc.


Indeed.

I have only used one regged PS in a power amp,
see in the image at http://www.turneraudio.com.au/webpic...ab400w317h.jpg

Its been going well for about 6 years, but all the power amps since then have chokes,
after I got some partial failures in a preamp. Leakage and spikes
cause death in SS regs all too easily.

I used to have an SS regged bench top test PS, but the BU208 pass element
blew up one evening when a storm was passing.
I wasted several BU208 when doing R&D for this PS.

I switched to a pair of 6AS7G, and a 6BX6 pentode gain tube,
and a switchable LC input filter to vary
the input voltage so there would not be excessive voltage across the 6AS7G
when using high current and low voltage output.

This test supply has never missed a beat in the last 7 years.

But I never would place a plate supply regulator in anything I sold;
it simply is not needed.

Patrick Turner.


NT


  #27   Report Post  
John Stewart
 
Posts: n/a
Default

Patrick Turner wrote:

John Stewart wrote:

lazyadm1n wrote:

Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power supply?

Asides from the higher cost and larger size of a choke, is there any reason
to choose one over the other?

One more question about chokes has to do with sizing. How does one determine
what size choke to use?
_________________
Rod

www.dortoh.ca


With the CLC filter there will be a resonance, possibly in the AF range.
I used Patrick's proposed values in this simulation. See the results at ABSE.


Nowhere did I ever suggest all the values for the CLC filter shown in the
test schematic at your post at ABSE.


The post was not meant to criticize your choice of components for the filter
or to embarrass you. I simply used the values which happened to be in the
discussion. I could just as easily chosen any number of other sets of values.

The values you posted to gain the results for resonance were :-

PS generator resistance = 1kOhm,
C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.
I assume your model used the questionable Duncan Amps freely
downloadable power supply designer program which I found gave me some
incomprehensible results.


Not at all. I have not tried the software you refer too, nor have I bothered to
download it from the net. I have been using a version of Electronic Workbench
for a few years now & find it to be quite accurate.

The schematic you posted is misleading to those who
may think that 10mF = 10 uF!!!!!!


I am certainly aware of the difference between a milli & a micro.
The simulation uses caps of 10 millifarads.

I recall the discussions revolved around a
very low generator impedance, ie, a large SS amp
power tranny feeding through SS diodes into an LC filter,
with approx 250 mH and 10,000 uF.

Fo = 3.18 Hz.


I have redrawn the simple schema so that anyone can see that
the circuit is parallel resonant at 4.505 Hz & included the formula.
The caps are in series so far as the circuit is concerned. If there
are any doubters it is easy to hook up a test circuit to try on
the bench. The resonance resulting in each case is quite obvious.

The 200 ohms is not low enough to damp the LC resonance,
but for critical damping, 7 ohms is required, so the 1.0 ohm dcR of the choke
effectively in series with the L will still leave the Q of the filter somewhat
higher than desirable,
but I found the voltage stability at the 10,000 uF didn't bounce around
when intermittently shunting the 200 ohm bleeder R with 32 ohms.


If the DCR of the choke is One Ohm I can add that in later. If you have
the ESR of the caps we can add that into the simulation as well.

For now go to ABSE to see the filter redrawn.

Cheers, John Stewart

  #28   Report Post  
Patrick Turner
 
Posts: n/a
Default



John Stewart wrote:

Patrick Turner wrote:

John Stewart wrote:

lazyadm1n wrote:

Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power supply?

Asides from the higher cost and larger size of a choke, is there any reason
to choose one over the other?

One more question about chokes has to do with sizing. How does one determine
what size choke to use?
_________________
Rod

www.dortoh.ca

With the CLC filter there will be a resonance, possibly in the AF range.
I used Patrick's proposed values in this simulation. See the results at ABSE.


Nowhere did I ever suggest all the values for the CLC filter shown in the
test schematic at your post at ABSE.


The post was not meant to criticize your choice of components for the filter
or to embarrass you. I simply used the values which happened to be in the
discussion. I could just as easily chosen any number of other sets of values.


But the choice of values just happened to be extraordinarily similar to what I had
chosen.
Then you appeared to use this choice to support the general idea that CLC filters were

poor engineering.





The values you posted to gain the results for resonance were :-

PS generator resistance = 1kOhm,
C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.
I assume your model used the questionable Duncan Amps freely
downloadable power supply designer program which I found gave me some
incomprehensible results.


Not at all. I have not tried the software you refer too, nor have I bothered to
download it from the net.


Duncan amps presents the circuit in an almost identical fashion.

But in a power supply, who would ever have
such a filter?

Why on earth would there be 1 kohm of series R between a voltage source
and C1?

Why isn't the power supply represented by a low impedance with diodes and
providing a DC flow, which affects the way the choke works, and its inductance value.

To newbies, the post of yous could be very confusing.


I have been using a version of Electronic Workbench
for a few years now & find it to be quite accurate.


Fair enough.
I found Duncan's software a bit plain wrong in its
depictions of the C2 voltage after turn on with an LC input,
but never mind, that's a separate issue on which nobody has commented after I raised
it.



The schematic you posted is misleading to those who
may think that 10mF = 10 uF!!!!!!


I am certainly aware of the difference between a milli & a micro.
The simulation uses caps of 10 millifarads.


Yes, I just reminded everyone, lest they are bewildered from
the times when where 10 mF meant 10 uF.



I recall the discussions revolved around a
very low generator impedance, ie, a large SS amp
power tranny feeding through SS diodes into an LC filter,
with approx 250 mH and 10,000 uF.

Fo = 3.18 Hz.


I have redrawn the simple schema so that anyone can see that
the circuit is parallel resonant at 4.505 Hz & included the formula.
The caps are in series so far as the circuit is concerned.


Indeed, because of the large value of series R before C1,
the CLC behaves as a sub optimally terminated low pass filter,
about as useful as tits on a bull for anyone designing and building a tube or SS amp.

General conclusions about CLC in power supplies cannot be realized from the
circuit you have posted.

are any doubters it is easy to hook up a test circuit to try on
the bench. The resonance resulting in each case is quite obvious.


So what?

Nobody in their right mind would use your circuit, they'd just be confused by it.





The 200 ohms is not low enough to damp the LC resonance,
but for critical damping, 7 ohms is required, so the 1.0 ohm dcR of the choke
effectively in series with the L will still leave the Q of the filter somewhat
higher than desirable,
but I found the voltage stability at the 10,000 uF didn't bounce around
when intermittently shunting the 200 ohm bleeder R with 32 ohms.


If the DCR of the choke is One Ohm I can add that in later. If you have
the ESR of the caps we can add that into the simulation as well.


My post on the other hand specifically relates the CLC behaviour to real world
situations and conditions,
and is not theory with ill fitting conclusions and recomendations attached.

I went to some length to address the issue of the importance of resistance
damping of LC filter circuits to avoid the problems of unwanted peaks in their
voltage output.

An LC filter when driven from a low resistance source
but with no R component either across the C, or across the L,
or in series before the LC to the source,
will produce a very peaked response at the C, at Fo depending on the
dcR of the L.

To make the filter have no resonant peak, but achieve
a -3 dB attenuation at Fo, and a following attenuation slope of 12dB /octave,
you need to have Rsource = very low, and have dcR = very low,
and then have at least an R across the C ( etc) = 1.41 x ZC or ZL at Fo.

1.41 x ZC or ZL is regarded as the critical value of R for damping an LC
circuit for "maximum flatness" of the response.

At Fo, ZC = ZL.

Now if R = some lower value than Rcrit, then the response slope will be what is called

over damped, and the roundness of the shoulder of the attenuation less sharp,
with the -3 dB point at a lower F, but the filter will still have a 12 dB/octave slope

at 3 Fo and beyond.

This sort of over damped LC is not used in PS, but is useful in
speaker crossovers.

As the R is made higher than Rcrit, the response of the filter becomes non flat,
and a peak appears around Fo, becoming a high peak maybe 15 dB above the LF input
level to the filter.

Its important that students studying such things be fully aware of such behaviours,
and get off their bums, away from the PC, and into the workshop
to examine some real world LC behaviours.
For those wondering about the relevance of this discussion,
I suggest they study the behaviour of CLC low pass filters and their
interactions with R attatched at the input and output, and be prepared for
what seem to be some very odd non-common sense behaviour.


For now go to ABSE to see the filter redrawn.


Its precisely no different to the previous circuit, but yes,
you effectively have a 5,000 uF plus 250 mH in a circuit in which
the resonant F is 4.5 Hz.



A useful formula for resonance is

Fo = 5,035 / sq rt of ( C x L ),

where F is in Hertz,
C is in uF,
L is in milli Henrys.

From this we can derive any value of L or C to make a
a resonant if we know the value of one of the L&C components.

When I searched for a decent LC filter program on the Web
for all types of lpf, hpf, band stop f , band pass f, notch f etc, using passive
components,
the only freeware i could find was at

http://www-users.cs.york.ac.uk/~fisher/lcfilter/

This doesn't allow you to cobble anything you like to together and find out
what the response is; you must specify typical input and output impedances
which are generally similar.

So its not a very useful resource, because with many filter apps, we start with a low
Z signal source
such as a power tranny, or a cathode follower and don't want to have to have a low
impedance on the
output, or adversely load the source impedance.

afaik, there are not many free downloadable LC filter programs which are worth
the trouble of downloading.

Perhaps someone may correct me on that.



Cheers, John Stewart


  #29   Report Post  
N. Thornton
 
Posts: n/a
Default

Patrick Turner wrote in message ...
"N. Thornton" wrote:


would say if youre really serious about quality one would go with a
proper regulated psu, not the cruder lower performance choke psu.


Just use the right values of L&C.
No need for all that SS regulation.


tech specs of CLCs far poorer. Covered.


Regulation in tube amps is only needed for class AB amps where they are used hard,
and where the amp is mainly class B.
if you don't believe me, try clipping a meter on the B+
of an average class AB amp B+, and take the music up to a level where
clipping only just starts to be visible on the CRO.
Usually, a supply of +450v moves only few volts.
With a sine wave input, sure, the B+ will move maybe 10%,
but that is still OK.


it isnt needed at all, but use it and you can improve amp specs.
Regged PSUs also enable you to get a bit more power out of a given set
of output devices - though not a whole lot more.


The downsides of choke psus compared to regulated ones a
very poor regulation


Not if the PT is low R, and diode R is low, and DCR of L is low.


I dont think youre understanding where the (lack of) regulation comes
from. The prime R is in the mains TF. There is no such thing as a
mains TF that has low R, they simply arent like that. Read up on
transformer regulation if you want.


significant impedance


Not if the L is followed by a large enough C.


Reduces af impedance, but does not eliminate it at lower frequencies,
and does not stop the v changing. Reg psus just dont have these
problems.


size
weight
cost


Depends how one is set up.

I wind my own, re-cycling old tranny cores...


thats a b---- of a job! Time is money in my book.


resonance


Eliminated by those who know what they are doing.


You can drop it right down with series R, but you wont eliminate it,
and you get even more psu impedance then.


and patchy rejection


Of what?


LF mains V variations changes arent rejected by the CLC, nor is higher
frequency muck: chokes have interwinding C that passes it. Compare
rejection figures of a choke psu to a proper reg, different league.


after I got some partial failures in a preamp. Leakage and spikes
cause death in SS regs all too easily.


you do have to design the reg to survive what it will meet IRL.
National semi have some good data sheets on this on their site, plus 1
or 2 HT reg designs.


I used to have an SS regged bench top test PS, but the BU208 pass element
blew up one evening when a storm was passing.
I wasted several BU208 when doing R&D for this PS.


design problem.


Put a suitably rated light bulb on an HT line and watch what happens
as you crank it up. It may stimulate your thoughts!


NT
  #30   Report Post  
N. Thornton
 
Posts: n/a
Default

Patrick Turner wrote in message ...
John Stewart wrote:

lazyadm1n wrote:

Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power supply?


the worms are everywhere now.


The values you posted to gain the results for resonance were :-

PS generator resistance = 1kOhm,


whats a PS generator?

C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.
I assume your model used the questionable Duncan Amps freely
downloadable power supply designer program which I found gave me some
incomprehensible results.

The schematic you posted is misleading to those who
may think that 10mF = 10 uF!!!!!!


it is. Lots of uF caps are marked mF. Always puzzles newbies. If your
software treats mF as = 1000uF, its a bit odd.


In the case of an LC filter, any resonance of the LC circuit
is damped when the value of the inductance changes considerably
with a varying charge rate into the C.


really? how do you reach that conclusion?


In my case with the LC filters for this SS amp,
the tested L for a 250 ohm load with a +55v rail was 0.35 H, or 350 mH.
When 32 ohms was added to the 250 ohms for total 28 ohms,
Idc became 1.9 amps dc, and L became 0.12 H, or 120 mH.

But the test circuit showed no sign of wild resonance voltage swings.
if there were, the choke goes from being higher L to a low value and which
saturates
for part of the rectifier wave forms, and becomes a very much lower
value of L when it does, so resonance seems damped
since L does not stay at a constant value.


I'm wondering how that would damp resonance.


The amp in question will be used as a sub amp, but will still have
input filtering to give a pole at 10 Hz.

I still don't believe the resonance will affect the sound one iota.

The resonance appears at about 4.5 Hz. This will have some effect
on the amp's sound.


There may be some amps which have a resonance in the CLC filter,
or LC filter ot CLCLC filter which may be at just under 5 Hz.

In none of them have I found there is any audible artifact.


This is fundamental theory. No output stage has perfect PS rejection,
therefore psu gunk will cause atrifacts on the output. Inevitably.


In a PP amp, any 5 Hz ripples are applied to the CT and common mode
rejection prevents their appearance in the output signal.


Rejection is never perfect. Look at some transistor curves and see
what happens as you vary Vce.


But even in SE amps, I sure don't get such problems in my amps.


there is no way for you not to.


I have always recommended the use of very large value C2 in a CLC
filter of a tube amp, 470 uF being typical.


which will certainly help.


This class A amp has an OPT, and CR coupling, like a tube amp,
and the circuit has a total of 20 dB of NFB,
and there is no resonance problems at LF whatever,
or stability problems because
the circuit open loop gain is well below unity where the LF phase shift
is 180 degrees.


what happens at 90 degrees determines stability.


On the other hand the CRC filter response is smooth but the power
losses are high.


The power losses are low, not high with a CLC filter.

Take your pick. There are no free rides while using simpler topologies.
A relatively easy way out would be a SS filter/regulator. I favor things
like the Int. Rectifier FETs to do the work.


People have been using CLC filters with no sonic problems for years


incorrect.


without
resorting to solid state regulators, which always do result in a power
loss, because of the voltage drop x load current across the series
pass element, or the B+ voltage x shunt element in the case of a shunt regulator.


yes, thats the price of better kit. (I cant imagine any grown up
designer using a shunt reg though.)


Regulators done with SS power fets or well rated bjt like the BU208 waste power,
CLC filters don't, because of the insignificant power lost
in the dcR of the choke, which totals Idc squared x R, in watts.


Good Luck with your project, John Stewart


I don't think the original poster's luck
will be improved by erroneous advice.


You either cant work out what you dont know, or are determined to
maintain an Uberexpert image to sell your goods. I'm sure you know
enough to make some fairly nice amps, but they wont be the best by any
means. If you read up on power supply rejection, transformer
regulation and ss reg design you'd produce some much better kit.


NT


  #31   Report Post  
Patrick Turner
 
Posts: n/a
Default



"N. Thornton" wrote:

Patrick Turner wrote in message ...
"N. Thornton" wrote:


would say if youre really serious about quality one would go with a
proper regulated psu, not the cruder lower performance choke psu.


Just use the right values of L&C.
No need for all that SS regulation.


tech specs of CLCs far poorer. Covered.


I agree, but no need for regs with tube power or preamps
to get great sound.



Regulation in tube amps is only needed for class AB amps where they are used hard,
and where the amp is mainly class B.
if you don't believe me, try clipping a meter on the B+
of an average class AB amp B+, and take the music up to a level where
clipping only just starts to be visible on the CRO.
Usually, a supply of +450v moves only few volts.
With a sine wave input, sure, the B+ will move maybe 10%,
but that is still OK.


it isnt needed at all, but use it and you can improve amp specs.
Regged PSUs also enable you to get a bit more power out of a given set
of output devices - though not a whole lot more.

The downsides of choke psus compared to regulated ones a
very poor regulation


Not if the PT is low R, and diode R is low, and DCR of L is low.


I dont think youre understanding where the (lack of) regulation comes
from. The prime R is in the mains TF. There is no such thing as a
mains TF that has low R, they simply arent like that. Read up on
transformer regulation if you want.


I make mains trannies and OPTs for a living.

The transformer regulation is usually better than 5% between
idle and full power with the lowest usable load in a class AB amp.

With a music signal, the B+ voltage changes less than 1% during normal use.

With class A amps, the variation in B+ is less than 1%, since the input power
scarcely changes.

Same goes for preamps.

The largest cause of sagging B+ voltage is the tube rectifiers,
but even with these, the variation of B+ on a typical class AB amp
in normal use is negligible.



significant impedance


Not if the L is followed by a large enough C.


Reduces af impedance, but does not eliminate it at lower frequencies,
and does not stop the v changing. Reg psus just dont have these
problems.


The problems you see are imagined, and not real, imho.



size
weight
cost


Depends how one is set up.

I wind my own, re-cycling old tranny cores...


thats a b---- of a job! Time is money in my book.


Yes, but I work for a pittance.



resonance


Eliminated by those who know what they are doing.


You can drop it right down with series R, but you wont eliminate it,
and you get even more psu impedance then.


There are no psu resonance problems in any of my amps.



and patchy rejection


Of what?


LF mains V variations changes arent rejected by the CLC, nor is higher
frequency muck: chokes have interwinding C that passes it.


The shunt C of a typical choke is say 300pF.
When followed by a cap of 470 uF, there is an attenuation
of any HF by a factor of over -100 dB.
There is no HF noise to be measured on the rails of amps I build.

The variations in mains voltage is such a small variation, and at such LF,
that there is zero variation conveyed to the output of any of the amps I build.


Compare
rejection figures of a choke psu to a proper reg, different league.


They measure better, but are a pita.



after I got some partial failures in a preamp. Leakage and spikes
cause death in SS regs all too easily.


you do have to design the reg to survive what it will meet IRL.
National semi have some good data sheets on this on their site, plus 1
or 2 HT reg designs.


I have repaired too many failed regulators to want to ever use them.


I used to have an SS regged bench top test PS, but the BU208 pass element
blew up one evening when a storm was passing.
I wasted several BU208 when doing R&D for this PS.


design problem.


No, spike problem.

The tubes have done a far better job for a bench top PS,
where the occasional short caused no trouble, but immediately
fused a BJT.


Put a suitably rated light bulb on an HT line and watch what happens
as you crank it up. It may stimulate your thoughts!


I already use light bulbs in series with SS amps when they turn up here for
repair with their fused parts within.
Better a light comes on than another fuse blows.

Good designs still fail, no matter how well illuminated the
thoughts were during the design process.

Patrick Turner.





NT


  #32   Report Post  
Patrick Turner
 
Posts: n/a
Default



"N. Thornton" wrote:

Patrick Turner wrote in message ...
John Stewart wrote:

lazyadm1n wrote:

Not to open a bag of worms or anything, but what (if any) are the
advantages/disadvanteges of a CRC filter vs a CLC filter in a power supply?


the worms are everywhere now.

The values you posted to gain the results for resonance were :-

PS generator resistance = 1kOhm,


whats a PS generator?

C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.
I assume your model used the questionable Duncan Amps freely
downloadable power supply designer program which I found gave me some
incomprehensible results.

The schematic you posted is misleading to those who
may think that 10mF = 10 uF!!!!!!


it is. Lots of uF caps are marked mF. Always puzzles newbies. If your
software treats mF as = 1000uF, its a bit odd.

In the case of an LC filter, any resonance of the LC circuit
is damped when the value of the inductance changes considerably
with a varying charge rate into the C.


really? how do you reach that conclusion?


You obviously have not carefully observed the "swinging choke"
inductance value with varying power levels going into the amp circuit.
I described it in detail in my posts here.



In my case with the LC filters for this SS amp,
the tested L for a 250 ohm load with a +55v rail was 0.35 H, or 350 mH.
When 32 ohms was added to the 250 ohms for total 28 ohms,
Idc became 1.9 amps dc, and L became 0.12 H, or 120 mH.

But the test circuit showed no sign of wild resonance voltage swings.
if there were, the choke goes from being higher L to a low value and which
saturates
for part of the rectifier wave forms, and becomes a very much lower
value of L when it does, so resonance seems damped
since L does not stay at a constant value.


I'm wondering how that would damp resonance.


The Fo changes with a changing L value, so little resonance occurs.
For resonance, you need a fixed value for L and C, and the Q
of the LC circuit must be high....

I suggest you build a few PS, and observe, observe....



The amp in question will be used as a sub amp, but will still have
input filtering to give a pole at 10 Hz.

I still don't believe the resonance will affect the sound one iota.

The resonance appears at about 4.5 Hz. This will have some effect
on the amp's sound.


There may be some amps which have a resonance in the CLC filter,
or LC filter ot CLCLC filter which may be at just under 5 Hz.

In none of them have I found there is any audible artifact.


This is fundamental theory. No output stage has perfect PS rejection,
therefore psu gunk will cause atrifacts on the output. Inevitably.


But the unresolved gunk you speak of is a lot less than noise and distortions from
other sources.
Even with the Quad II attrociously inadequate PS, the IM distortion caused by
the PS is about the same as the THD levels of the amps.
With an improved CLC plate supply, the IM from PS is reduced about 30 dB at least.

And Quad II at 2 watt normal listening levels make only about 0.03%
thd when in stock standard form.



In a PP amp, any 5 Hz ripples are applied to the CT and common mode
rejection prevents their appearance in the output signal.


Rejection is never perfect. Look at some transistor curves and see
what happens as you vary Vce.


All the more reason to filter out the hum on transistor collector
circuit rails, achievable just LC, or CLC, where C1 is say 5,000
uF, and C2 is a far higher value, with a low dcR choke from C1 to C2.

There is then little dependance on NFB to remove noise in the SS amp output.
So less NFB need be used, or if a lot is used, its more effective.


But even in SE amps, I sure don't get such problems in my amps.

there is no way for you not to.


What bull****. The levels of PS caused spuriae
in at the output of all my SE amps is inaudible, and measures far less
than the natural thd/imd of the signal processing, at all levels of operation.




I have always recommended the use of very large value C2 in a CLC
filter of a tube amp, 470 uF being typical.


which will certainly help.

This class A amp has an OPT, and CR coupling, like a tube amp,
and the circuit has a total of 20 dB of NFB,
and there is no resonance problems at LF whatever,
or stability problems because
the circuit open loop gain is well below unity where the LF phase shift
is 180 degrees.


what happens at 90 degrees determines stability.


Er, you need 180 degrees of phase shift for instability.

But lets not argue about a few degrees.



On the other hand the CRC filter response is smooth but the power
losses are high.


The power losses are low, not high with a CLC filter.

Take your pick. There are no free rides while using simpler topologies.
A relatively easy way out would be a SS filter/regulator. I favor things
like the Int. Rectifier FETs to do the work.


People have been using CLC filters with no sonic problems for years


incorrect.


Well you are very inexperienced.



without
resorting to solid state regulators, which always do result in a power
loss, because of the voltage drop x load current across the series
pass element, or the B+ voltage x shunt element in the case of a shunt regulator.


yes, thats the price of better kit. (I cant imagine any grown up
designer using a shunt reg though.)


I can imagine a grown up using a shunt reg.



Regulators done with SS power fets or well rated bjt like the BU208 waste power,
CLC filters don't, because of the insignificant power lost
in the dcR of the choke, which totals Idc squared x R, in watts.


Good Luck with your project, John Stewart


I don't think the original poster's luck
will be improved by erroneous advice.


You either cant work out what you dont know, or are determined to
maintain an Uberexpert image to sell your goods. I'm sure you know
enough to make some fairly nice amps, but they wont be the best by any
means. If you read up on power supply rejection, transformer
regulation and ss reg design you'd produce some much better kit.


And what do you produce and sell?

Complex junk riddled with SS crap all through the circuit?

I have compared the sound of my gear at audiophile meetings
with examples of high end with all their complex and very unreliable
SS regs, and recieved very favourable reports.

You are one of the crowd who bases his judgements of "bestness"
on measured results.

That being the case, get off this group, you don't belong, you waste out time,
and your own, because well all know an SS amp can measure
at 0.0001% thd at 200 watts, but frankly, we don't hear any difference if we remove
two zeros from the figures.

To conclude, there is no evidence that well designed CLC
filters in tube gear will ever contribute more than 10% of the total measured
N&D of the amps concerned, and if that is say 0.03% as it is with most
of my amps at normal levels, then the N&D contribution from my PSUs is
a mere 0.003% of the signal.

If the signal level is 4 vrms at the 8 ohm speaker, the
level of the PSU caused N&D at 0.003% is only 0.012 mV.

If you wanna come back with some figures to support the idea that
well designed CLC are ****e in amplifiers, then do so,
but you better get your facts well sorted out.

Patrick Turner.



  #33   Report Post  
N. Thornton
 
Posts: n/a
Default

Patrick Turner wrote in message ...
"N. Thornton" wrote:


would say if youre really serious about quality one would go with a
proper regulated psu, not the cruder lower performance choke psu.

Just use the right values of L&C.
No need for all that SS regulation.


tech specs of CLCs far poorer. Covered.


I agree, but no need for regs with tube power or preamps
to get great sound.


I think we covered that one.


The transformer regulation is usually better than 5% between
idle and full power with the lowest usable load in a class AB amp.

With a music signal, the B+ voltage changes less than 1% during normal use.

With class A amps, the variation in B+ is less than 1%, since the input power
scarcely changes.


1% times op stage psrr, could do better.


significant impedance

Not if the L is followed by a large enough C.


Reduces af impedance, but does not eliminate it at lower frequencies,
and does not stop the v changing. Reg psus just dont have these
problems.


The problems you see are imagined, and not real, imho.


If youd taken the suggestion of the lightbulb trick then looked up
some curves, you might be thinking twice about that


I wind my own, re-cycling old tranny cores...


thats a b---- of a job! Time is money in my book.


Yes, but I work for a pittance.


smart


The variations in mains voltage is such a small variation, and at such LF,
that there is zero variation conveyed to the output of any of the amps I build.


check out mains specs, then add in v drops from deomestic wiring. Here
in Europe youre looking at a variation of around 20% max to min.


I used to have an SS regged bench top test PS, but the BU208 pass element
blew up one evening when a storm was passing.
I wasted several BU208 when doing R&D for this PS.


design problem.


No, spike problem.


Any psu designer knows you can properly and reliably protect a reg
against spikes. And that you need to: if youve had failures from
spikes, your designs were unsatisfactory.


Put a suitably rated light bulb on an HT line and watch what happens
as you crank it up. It may stimulate your thoughts!


I already use light bulbs in series with SS amps when they turn up here for
repair with their fused parts within.
Better a light comes on than another fuse blows.


So you wont try that.
  #34   Report Post  
N. Thornton
 
Posts: n/a
Default

Patrick Turner wrote in message ...
"N. Thornton" wrote:
John Stewart wrote:
lazyadm1n wrote:


PS generator resistance = 1kOhm,


whats a PS generator?

C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.
I assume your model used the questionable Duncan Amps freely
downloadable power supply designer program which I found gave me some
incomprehensible results.

The schematic you posted is misleading to those who
may think that 10mF = 10 uF!!!!!!


it is. Lots of uF caps are marked mF. Always puzzles newbies. If your
software treats mF as = 1000uF, its a bit odd.

In the case of an LC filter, any resonance of the LC circuit
is damped when the value of the inductance changes considerably
with a varying charge rate into the C.


really? how do you reach that conclusion?


You obviously have not carefully observed the "swinging choke"
inductance value with varying power levels going into the amp circuit.
I described it in detail in my posts here.


A silly comment. The changing L of a swinging choke does not kill
resonance, it merely moves it about frequency wise.


In my case with the LC filters for this SS amp,
the tested L for a 250 ohm load with a +55v rail was 0.35 H, or 350 mH.
When 32 ohms was added to the 250 ohms for total 28 ohms,
Idc became 1.9 amps dc, and L became 0.12 H, or 120 mH.

But the test circuit showed no sign of wild resonance voltage swings.
if there were, the choke goes from being higher L to a low value and which
saturates
for part of the rectifier wave forms, and becomes a very much lower
value of L when it does, so resonance seems damped
since L does not stay at a constant value.


I'm wondering how that would damp resonance.


The Fo changes with a changing L value, so little resonance occurs.


non sequitor

For resonance, you need a fixed value for L and C, and the Q
of the LC circuit must be high....


nope.


I suggest you build a few PS, and observe, observe....


posing


There may be some amps which have a resonance in the CLC filter,
or LC filter ot CLCLC filter which may be at just under 5 Hz.

In none of them have I found there is any audible artifact.


This is fundamental theory. No output stage has perfect PS rejection,
therefore psu gunk will cause atrifacts on the output. Inevitably.


But the unresolved gunk you speak of is a lot less than noise and distortions from
other sources.
Even with the Quad II attrociously inadequate PS, the IM distortion caused by
the PS is about the same as the THD levels of the amps.
With an improved CLC plate supply, the IM from PS is reduced about 30 dB at least.

And Quad II at 2 watt normal listening levels make only about 0.03%
thd when in stock standard form.



But even in SE amps, I sure don't get such problems in my amps.

there is no way for you not to.


What bull****. The levels of PS caused spuriae
in at the output of all my SE amps is inaudible, and measures far less
than the natural thd/imd of the signal processing, at all levels of operation.


you say inaudible, i say inevitable.


I have always recommended the use of very large value C2 in a CLC
filter of a tube amp, 470 uF being typical.


which will certainly help.

This class A amp has an OPT, and CR coupling, like a tube amp,
and the circuit has a total of 20 dB of NFB,
and there is no resonance problems at LF whatever,
or stability problems because
the circuit open loop gain is well below unity where the LF phase shift
is 180 degrees.


what happens at 90 degrees determines stability.


Er, you need 180 degrees of phase shift for instability.


wrong. Anything from 90 degs to 270 can do it with the right gain.
Think about this: lets say phase shift is 179 degs, and gain 2. You
draw those 2 cycles superimposed on paper, and try tell me it wont
oscillate.


But lets not argue about a few degrees.



without
resorting to solid state regulators, which always do result in a power
loss, because of the voltage drop x load current across the series
pass element, or the B+ voltage x shunt element in the case of a shunt regulator.


yes, thats the price of better kit. (I cant imagine any grown up
designer using a shunt reg though.)


I can imagine a grown up using a shunt reg.


in an amp psu? I cant. Senseless idea.


You either cant work out what you dont know, or are determined to
maintain an Uberexpert image to sell your goods. I'm sure you know
enough to make some fairly nice amps, but they wont be the best by any
means. If you read up on power supply rejection, transformer
regulation and ss reg design you'd produce some much better kit.


And what do you produce and sell?

Complex junk riddled with SS crap all through the circuit?

I have compared the sound of my gear at audiophile meetings
with examples of high end with all their complex and very unreliable
SS regs, and recieved very favourable reports.


thats to be expected, again do some reading on subjective tests.


You are one of the crowd who bases his judgements of "bestness"
on measured results.


nope


If you wanna come back with some figures to support the idea that
well designed CLC are ****e in amplifiers, then do so,


not what i claimed. And no thanks, ive no interest in educating
someone not interested in learning.


NT
  #35   Report Post  
John Stewart
 
Posts: n/a
Default

Patrick Turner wrote:


With the CLC filter there will be a resonance, possibly in the AF range.
I used Patrick's proposed values in this simulation. See the results at ABSE.

Nowhere did I ever suggest all the values for the CLC filter shown in the
test schematic at your post at ABSE.


The post was not meant to criticize your choice of components for the filter
or to embarrass you. I simply used the values which happened to be in the
discussion. I could just as easily chosen any number of other sets of values.


But the choice of values just happened to be extraordinarily similar to what I had
chosen.


Not only that, they are the same & I said so somewhere in my post.

Then you appeared to use this choice to support the general idea that CLC filters were
poor engineering.


Not at all. As usual you are jumping to conclusions. I am detecting insecurity here.

The values you posted to gain the results for resonance were :-

PS generator resistance = 1kOhm,
C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.


You got that correct, anyway.

I assume your model used the questionable Duncan Amps freely


downloadable power supply designer program which I found gave me some
incomprehensible results.


Not at all. I have not tried the software you refer too, nor have I bothered to
download it from the net.


Duncan amps presents the circuit in an almost identical fashion.

But in a power supply, who would ever have
such a filter?


You, I & many others.

Why on earth would there be 1 kohm of series R between a voltage source
and C1?


You are having a problem of understanding the difference between a real circuit
& a short cut for simulation only. The simulation is to demonstrate resonance.
Don't worry to much, it will happen in a real circuit anyway.

Why isn't the power supply represented by a low impedance with diodes and
providing a DC flow, which affects the way the choke works, and its inductance value.


Yes, but it is low Z only while the diodes are conducting at the rectifier end of the
filter. The load end can oscillate all over depending on load changes & it does.
Swinging chokes were once popular with the Class B set. They are no longer
shown in the Hammond catalogue. The choke I used for my tests below changes
about 20% from no load to full load.

To newbies, the post of yous could be very confusing.


If you go to ABSE you will see a result measured today on a real circuit.
The filter is a PI of a pair of 20 microfarad caps & a Hammond 10 H
choke. The choke resistance is 82 ohms. It is in a regulated PS but just as
easily could have been seen in an amp of the 50's. The 40 volt jumps
are caused by a 70 ma change, first on then off. The trace demonstrates
well what I tried to make others aware of before you got bent out of shape.

It's time you listened a bit to the advice of others.

If I have time later this week I will post some stuff on what happens to
the filter in an SE amp.

Cheers to all, John Stewart



  #36   Report Post  
Patrick Turner
 
Posts: n/a
Default



"N. Thornton" wrote:

Patrick Turner wrote in message ...
"N. Thornton" wrote:
John Stewart wrote:
lazyadm1n wrote:


PS generator resistance = 1kOhm,

whats a PS generator?

C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.
I assume your model used the questionable Duncan Amps freely
downloadable power supply designer program which I found gave me some
incomprehensible results.

The schematic you posted is misleading to those who
may think that 10mF = 10 uF!!!!!!

it is. Lots of uF caps are marked mF. Always puzzles newbies. If your
software treats mF as = 1000uF, its a bit odd.

In the case of an LC filter, any resonance of the LC circuit
is damped when the value of the inductance changes considerably
with a varying charge rate into the C.

really? how do you reach that conclusion?


You obviously have not carefully observed the "swinging choke"
inductance value with varying power levels going into the amp circuit.
I described it in detail in my posts here.


A silly comment. The changing L of a swinging choke does not kill
resonance, it merely moves it about frequency wise.

In my case with the LC filters for this SS amp,
the tested L for a 250 ohm load with a +55v rail was 0.35 H, or 350 mH.
When 32 ohms was added to the 250 ohms for total 28 ohms,
Idc became 1.9 amps dc, and L became 0.12 H, or 120 mH.

But the test circuit showed no sign of wild resonance voltage swings.
if there were, the choke goes from being higher L to a low value and which
saturates
for part of the rectifier wave forms, and becomes a very much lower
value of L when it does, so resonance seems damped
since L does not stay at a constant value.

I'm wondering how that would damp resonance.


The Fo changes with a changing L value, so little resonance occurs.


non sequitor

For resonance, you need a fixed value for L and C, and the Q
of the LC circuit must be high....


nope.

I suggest you build a few PS, and observe, observe....


posing

There may be some amps which have a resonance in the CLC filter,
or LC filter ot CLCLC filter which may be at just under 5 Hz.

In none of them have I found there is any audible artifact.

This is fundamental theory. No output stage has perfect PS rejection,
therefore psu gunk will cause atrifacts on the output. Inevitably.


But the unresolved gunk you speak of is a lot less than noise and distortions from
other sources.
Even with the Quad II attrociously inadequate PS, the IM distortion caused by
the PS is about the same as the THD levels of the amps.
With an improved CLC plate supply, the IM from PS is reduced about 30 dB at least.

And Quad II at 2 watt normal listening levels make only about 0.03%
thd when in stock standard form.


But even in SE amps, I sure don't get such problems in my amps.

there is no way for you not to.


What bull****. The levels of PS caused spuriae
in at the output of all my SE amps is inaudible, and measures far less
than the natural thd/imd of the signal processing, at all levels of operation.


you say inaudible, i say inevitable.

I have always recommended the use of very large value C2 in a CLC
filter of a tube amp, 470 uF being typical.

which will certainly help.

This class A amp has an OPT, and CR coupling, like a tube amp,
and the circuit has a total of 20 dB of NFB,
and there is no resonance problems at LF whatever,
or stability problems because
the circuit open loop gain is well below unity where the LF phase shift
is 180 degrees.

what happens at 90 degrees determines stability.


Er, you need 180 degrees of phase shift for instability.


wrong. Anything from 90 degs to 270 can do it with the right gain.
Think about this: lets say phase shift is 179 degs, and gain 2. You
draw those 2 cycles superimposed on paper, and try tell me it wont
oscillate.


Ah, so now you have suddenly introduced a wider range of degrees and gain conditions to suit
your argument.

I won't dissagree.



But lets not argue about a few degrees.


without
resorting to solid state regulators, which always do result in a power
loss, because of the voltage drop x load current across the series
pass element, or the B+ voltage x shunt element in the case of a shunt regulator.

yes, thats the price of better kit. (I cant imagine any grown up
designer using a shunt reg though.)


I can imagine a grown up using a shunt reg.


in an amp psu? I cant. Senseless idea.


Well that's your preference.

We sure don't mind if you think we are senseless.

A few zeners, maybe a regulator tube or two hurt nobody....




You either cant work out what you dont know, or are determined to
maintain an Uberexpert image to sell your goods. I'm sure you know
enough to make some fairly nice amps, but they wont be the best by any
means. If you read up on power supply rejection, transformer
regulation and ss reg design you'd produce some much better kit.


And what do you produce and sell?

Complex junk riddled with SS crap all through the circuit?

I have compared the sound of my gear at audiophile meetings
with examples of high end with all their complex and very unreliable
SS regs, and recieved very favourable reports.


thats to be expected, again do some reading on subjective tests.

You are one of the crowd who bases his judgements of "bestness"
on measured results.


nope

If you wanna come back with some figures to support the idea that
well designed CLC are ****e in amplifiers, then do so,


not what i claimed. And no thanks, ive no interest in educating
someone not interested in learning.


And you have bugger all interest in building anything,
testing it, and listening.

You have provided no proof that CLC filters are ****e.

I know all about regulators and how well they measure,
while they stay alive, so I will get by without your educational input..

If you have 20% mains variations where you live, it is unfortunate.

But if 20% variation means down to 207v, and up to 253v,
then all my amps will cope OK without regulators, and if the variations
took place at a rate of between below 1 Hz, I bet you couldn't
hear it.
My SS amps have no damn regulation either, and I can switch
off the 300 watter, and after 10 seconds, switch it back on,
after allowing the rails to fall 50%, and hear exactly nothing at all,
because the amp works fine over a wide range of voltages, and the
NFB copes with any outside source of noise or rail changes.
You can sit there turning my power amps on and off every 1/2 a second, and hear nothing,
until finally the heaters cool down a bit.......

If the mains variation exceeds +/-10%, then the heater voltage will stray
too far from being correct, and you'll maybe have to worry about tube life.

The job of regulation of the heater circuits might be something you would really
relish, since you appear unable and unwilling to use CLC filters,
and you are regulation fetishist.


Patrick Turner.



NT


  #37   Report Post  
Patrick Turner
 
Posts: n/a
Default



John Stewart wrote:

Patrick Turner wrote:


With the CLC filter there will be a resonance, possibly in the AF range.
I used Patrick's proposed values in this simulation. See the results at ABSE.

Nowhere did I ever suggest all the values for the CLC filter shown in the
test schematic at your post at ABSE.

The post was not meant to criticize your choice of components for the filter
or to embarrass you. I simply used the values which happened to be in the
discussion. I could just as easily chosen any number of other sets of values.


But the choice of values just happened to be extraordinarily similar to what I had
chosen.


Not only that, they are the same & I said so somewhere in my post.

Then you appeared to use this choice to support the general idea that CLC filters were
poor engineering.


Not at all. As usual you are jumping to conclusions. I am detecting insecurity here.


But I don't want everyone to incorrectly make conclusions about what I proposed,
by taking your simulations as simulations of what may occur in what I proposed.

Please try to make the distinctions clearer in all future postings, lest your vagueness
tendencies
cast unjustified negative critique upon an undeserving poster.


The values you posted to gain the results for resonance were :-

PS generator resistance = 1kOhm,
C1 = 10mF, which is milli Farads, or 10,000uF,
L = 250 milli Henrys.
C2 = 10mF, which is milli Farads, or 10,000 uF,
RL = 200 ohms.


You got that correct, anyway.

I assume your model used the questionable Duncan Amps freely


downloadable power supply designer program which I found gave me some
incomprehensible results.

Not at all. I have not tried the software you refer too, nor have I bothered to
download it from the net.


Duncan amps presents the circuit in an almost identical fashion.

But in a power supply, who would ever have
such a filter?


You, I & many others.


No, I don't use that circuit.

The source impedance shown is utterly too high, and totally different
to what I would use!!!!!!!

Please DO NOT say I would use something that I wouldn't.

You are discrediting me again!




Why on earth would there be 1 kohm of series R between a voltage source
and C1?


You are having a problem of understanding the difference between a real circuit
& a short cut for simulation only. The simulation is to demonstrate resonance.
Don't worry to much, it will happen in a real circuit anyway.


Your post about resonance is fine, but the circuit shown isn't what i would use.





Why isn't the power supply represented by a low impedance with diodes and
providing a DC flow, which affects the way the choke works, and its inductance value.


Yes, but it is low Z only while the diodes are conducting at the rectifier end of the
filter. The load end can oscillate all over depending on load changes & it does.
Swinging chokes were once popular with the Class B set. They are no longer
shown in the Hammond catalogue. The choke I used for my tests below changes
about 20% from no load to full load.


Demand for swinger chokes is so low, I can't think why Hammond would bother.

They try to suit the market for low priced volume orders.



To newbies, the post of yous could be very confusing.


If you go to ABSE you will see a result measured today on a real circuit.
The filter is a PI of a pair of 20 microfarad caps & a Hammond 10 H
choke. The choke resistance is 82 ohms. It is in a regulated PS but just as
easily could have been seen in an amp of the 50's. The 40 volt jumps
are caused by a 70 ma change, first on then off. The trace demonstrates
well what I tried to make others aware of before you got bent out of shape.

It's time you listened a bit to the advice of others.


I got bent out of shape because you were not able to
present the correct simulation of the LC input circuit exactly as I said it was, a tranny
feeding
silicon diodes which continually conduct through a swinging choke to a large cap.

This was very sloppy tired work of yours, and shows you don't read or listen yourself.

Now we have a new presentation of a CLF filter using
a pair of 20 uF caps, and a 10H choke with 82 ohms dcR.

If we consider the source impedance of the tranny and diodes to be lowZ,
then the L and C2 form a series resonant circuit to 0V with Fo = 11.2 Hz

The ZC = ZL at 11.2 Hz, and each = 703 ohms.

If there was an RL of 1,000 ohms connected to C2, say
400v at 400 mA, then the LC filter would be terminated
by the value of critical R to make the LC filter give a maximally flat
response and I see no reason to believe that there would be
any resonance problems.

I would never use such a filter as you have described here.

I would perhaps use C1 = 47 uF, if the rectifier was a tube type.
If the diodes were silicon, I'd use 470 uF for C1 if I wanted,
and try to use say 15 ohms R in series to reduce the peak charge currents
in the diodes.

I would then use 470 uF for C2, not the attrociously inadequate 20 uF,
entirely unfit for anchoring down the plate voltage of a PP power amp
let alone an SE power amp.
C2 needs to be low impedance, lest changing voltages at the supply cap cause IMD
of the HF by the LF signals.
470 uF gives more than enough low impedance, at 3.4 ohms at 100 Hz.

If a choke value was 2H, then Fo would be 5.2 Hz, sufficiently low,
and the choke can be smaller and lighter than the Hammond, and the circuit will work better.


If I have time later this week I will post some stuff on what happens to
the filter in an SE amp.


I have the CRO screen picture you sent privately, but in regretable fashion,
there were not only inadequate explanatory notes provided for the test curve result
conditions, there were no notes sent at all!!!!

Please provide the page of tightly packed information anyone with a disciplined mind would
expect
if you want yourself to be fully understood and listened to as you would wish.

At the moment, I have no up to date instant access to ABSE.

I really don't need to see the claimed resonance curves of LC input schemes
unless their relevance in real world amplifiers is also fully explained and included,
and of course there is all the info in RDH4, so before ppl dismiss CLC
as stupid and ill-concieved, then they have a job ahead of them
to proove RDH4 was also a pile of BS, which I suggest it isn't.

Patrick Turner.



Cheers to all, John Stewart


  #38   Report Post  
Patrick Turner
 
Posts: n/a
Default


I have a few more thoughts on LC input filters
for a class B SS amp, where L is a swinging choke
with L = 0.35H at 220 mA, and drops to 0.11H at 2 amps,
and C = 10,000 uF,
The power tranny winding for the full wave SS rectifier
has 1 ohm dcR, and there is 60vrms from the winding,
so there would be +53 volts at the C with Idc = 220 mA.
Mains F is assumed to be 50 Hz.


A couple of you guys have mounted a healthy and pithy challenge
to the wisdom of what I have proposed, with the theme of their challenge
based around the idea that the undamped resonance of the LC
filter will have a serious effect on the audio of the amp,
and especially on its power ceiling ability, and you'd think this
could easily be tested using a *full range* pink noise signal,
which *will contain* enough quite LF signals to excite whatever resonances in a system
if they exist.
Pink noise is a good test signal because it most closely resembles
hard rock and roll, heavy metal etc, but with deeper going bass signals.
( Such signals soon tell you the weakness in any tube amp, such as in an OPT,
because the OPT will damn well saturate when the average levels of the test signal are way below
the sine wave
max power for say 25 Hz, when there won't be any sign of saturation
when compared to the 1 kHz max level signal;
full level signals below say Fsat of 14 Hz will make awful
knocking noise from the OPT, and cause short circuit current
bursts through the saturated coils and OP tubes...)

If the instananeous L value was 0.25H and C is 10,000 uF,
the Fo = 3.2 Hz, and should somebody want to reduce this F
they could do worse than use 40,000 uF, which could be made up easily
with 4 x 10,000 uF caps of medium ripple current ability, because
with a choke there are no heavy peak charge currents
that you get with a cap input filter.
If the C is quadrupled, the Fo is reduced by 1/2 to 1.6Hz.

But one could also apply critical R damping to the LC
so that the response of the LC filter does not have the huge peak
at 3.2 Hz that John Stewart has so kindly simulated and presented with his
recent post at ABSE.
At 3.2Hz, each of the L and C reactances = 5 ohms at Fo = 3.2 Hz.
If the R across the C were to be 7 ohms, the
- 3 dB point of the filter would be at 3.2 Hz, and the response unpeaked
and thus *non resonant*.

But since the LC filter is to be used for a DC power supply, placing
7 ohms across the C would create a huge waste of DC flow, and
so that idea isn't practical.

But then if you had 7 ohms from the 10,000 uF to say
100,000 uF as a low frequency snubber network,
then indeed the the LC filter would be critically damped without
wasting DC power.

One could also apply an active dynamic shunt impedance
which acts at least like a 7 ohm resistive impedance, but which consists
of a power transistor which only conducts when an AC signal at the C
exceeds a certain threshold at around 3.2 Hz, which suggests
the signal to the base of the transistor be from a
parallel resonant circuit fed by an R from the 10,000 uf.

In other words we could use what may be a simple shunt regulator,
which passes no more direct current at idle than say 220 mA,
but which otherwise acts like at least 7 ohm resistor, but only
where its needed.
I believe such a scheme could be not unlike a "capacitance multiplyer",
and be a lot cheaper and easier to fit into the amp than
40,000 uF, but in fact it would be like a Q divider, rather than like a
Q multiplyer, used in RF circuitry to increase the Q of a resonant circuit
to increase selectivity.
Any active device across the C would have to be protected
against back emfs in the choke.

The next idea that comes to mind is to place
7 ohms across the choke, which would damp the resonance peak in the
LC, but which would change the rate of attenuation
of the filter and make the filter into an RC filter with 6 dB/octave slope
above the pole F of around 3.2 Hz,
so that for an LC input filter, the
attenutation of the 100 Hz rectifier hum would be less than with a pure
LC filter.
With the LC values chosen, at 100 Hz, ZL = 157 ohms,
and ZC = 0.16 ohms, so the approximate 24 vrms of 100Hz signal
at the input to the choke is a factor of 0.16 / 157 = 0.001, so the hum
is reduced to 24 mV.

With 7 ohms across the L, the Z ( R+L ) at 100 Hz
ia approx 7 ohms, so the
attenuation of the 24vrms 100 Hz would be down to
( 0.16 / 7 ) x 24v = 0.54 volts.

But no DC power losses would occur in the R across the choke, because
nearly all the DC power loss is in the dcR of the choke, and little
DC flows in the 7 ohms across the L.

Perhaps slightly more than 7 ohms would be sufficient to damp the LC resonance,
maybe 10 ohms is enough, so I ask someone with a simulation
program to set up the LC filter connected to a full wave SS
rectifier from a 60 vrms secondary PT winding with a dcR = 1 ohm.

In a tube amp, a similar arrangement could be simulated.

You could have the following values
for an LC input filter for a class AB amp, where L is a swinging choke
with L = 35H at 22 mA, and drops to 11H at 200mA,
and C = 100 uF.
The power tranny winding for the full wave SS rectifier
has 10 ohm dcR, and there is 600vrms from the winding,
so there would be +530 volts at the C with Idc = 22 mA.
Mains F is assumed to be 50 Hz.
Critical damping R would have to be 700 ohms,
and so a snubber network with 700 ohms and say 1,000 uF across the
100 uF would remove resonance effects.

But immediately we would see that the 100 uF is perhaps
not sufficient to anchor the CT of the OPT,
where 1/4 of the RLa-a + the Ra of the tubes is connected to 100 uF,
so that if this load and source R was say 2,000 ohms,
then at 10 Hz the 100 uF has become 160 ohms,
and at 3.2 Hz it has become 480 ohms, and too high a value
to function as an anchor.
There is a strong case to increase the amount of C by a large factor,
which isn't hard in a world full of cheap small 470 uF x 450v electros.
4 in a series/parallel array will give 470 uF, to cope with the expected working voltage
of 530v or the soaring voltage of +848v, in case there was no
bleeder current of 22 mA.
The value of the choke, and its range of values must
be maintained irrespective of what sized C we choose, since the DC flow
commands the L value so that L must never be less than RL / 940.
( Notice that if Idc is zero, RL = infinity,
and so L would have to be an infinite value to make the B+ = 0.89 x Vrms
of the AC transformer winding. )
But if the bleeder R current was say 44 mA, then RL = 12k,
so L critical needs to be 12.7 H,
and at 200 mA, it can be allowed to fall to 2.8H.
It is good practice to use more L than the critical calculated values,
and hence 17H falling to 4.3H would be about right,
and at a current where L was in fact 12.7H
and the C = 470 uF, the Fo would be 2 Hz,
and if anyone wants to say such a resonant F will upset the music, they
best get a move on to proove it, remembering
that there isn't much 2 Hz info in music, and the
amp gain at 2 Hz is low, and that the OPT primary
impedance at 2 Hz is also rather low, etc.
With 200 mA of dc flow, and with
4.3H in series with 470 uF, the
240vrms of 100Hz from the rectifier is reduced by a factor of ZC / ZL =
Z 470uF / Z 4.3H = 3.4 / 2,700 = 0.0012,
so the hum would be 0.3vrms at the CT at 200 mA of dc flow.
But Fo would have risen to 3.54 Hz,
but then the RL connected across the C = 530 / 0.2 = 2,650 ohms,
which would help damp the resonance to some extent.

There is a another set of circumstances to analyse on your simulators.

Patrick Turner.








  #39   Report Post  
Patrick Turner
 
Posts: n/a
Default


As a follow up to the thread I started above,
I have some test measurements which may be of interest to
people such as John Byrns, who said that the resonant behaviour
of the LC filter used with a class B amplifier would result in being forced to limit the signal
output
to 3dB less than theoretical maximum, or words to that effect.

I just ran a test on the LC input filter described here :-

L is a swinging choke
with L = 0.35H at 228 mA, and drops to 0.11H at 2 amps,
and C = 10,000 uF,
The power tranny winding for the full wave SS rectifier
has 1 ohm dcR, and there is 65vrms from the winding,
so there would be +57 volts at the C with Idc = 228 mA.
Mains F is 50 Hz.


The resistance to get 228mA was 250 ohms, and
I then tried repeatedly adding 32 ohms, to make the supply voltage fall
to 53v, and DC current increase to 1.87 amps.
The speed at which I added/removed the 32 ohms was varied to generate
the highest possible signal voltage across the C, and all was monitored with a CRO,
and the voltage variation with an F around 4 Hz was 4.6 vrms,
or about +/-7 peak volts, and I seriously doubt that any music signal condition
could generate a greater disturbance to the supply.
But where the F of adding subtracting the 32 ohms to cause the transient
was reduced to less or more than around 4 Hz, the voltage swing of the supply
was about 1/2 the 7 volts peak.
So when the supply has a sudden 1.6 amp increase in load current, the rail
dips to just under the final rail volts before stabilising at 53v, and when the current is reduced
1.6 amps,
the rail voltage swings up to above the 57v rail before settling at 57v.
There was no more than about 1/2 a "ring cycle".
But I still don't see a problem here under normal music operation where
the maximum tried for rail swings are less than 13% of the rail voltage, and in any case
full power with a sine wave would produce a 9% drop in rail voltage.

It will be interesting to see what a pink noise or music signal will do.

But I have to wire up the amps to test further, and that will take all next week.

Patrick Turner.





A couple of you guys have mounted a healthy and pithy challenge
to the wisdom of what I have proposed, with the theme of their challenge
based around the idea that the undamped resonance of the LC
filter will have a serious effect on the audio of the amp,
and especially on its power ceiling ability, and you'd think this
could easily be tested using a *full range* pink noise signal,
which *will contain* enough quite LF signals to excite whatever resonances in a system
if they exist.
Pink noise is a good test signal because it most closely resembles
hard rock and roll, heavy metal etc, but with deeper going bass signals.
( Such signals soon tell you the weakness in any tube amp, such as in an OPT,
because the OPT will damn well saturate when the average levels of the test signal are way below
the sine wave
max power for say 25 Hz, when there won't be any sign of saturation
when compared to the 1 kHz max level signal;
full level signals below say Fsat of 14 Hz will make awful
knocking noise from the OPT, and cause short circuit current
bursts through the saturated coils and OP tubes...)

If the instananeous L value was 0.25H and C is 10,000 uF,
the Fo = 3.2 Hz, and should somebody want to reduce this F
they could do worse than use 40,000 uF, which could be made up easily
with 4 x 10,000 uF caps of medium ripple current ability, because
with a choke there are no heavy peak charge currents
that you get with a cap input filter.
If the C is quadrupled, the Fo is reduced by 1/2 to 1.6Hz.

But one could also apply critical R damping to the LC
so that the response of the LC filter does not have the huge peak
at 3.2 Hz that John Stewart has so kindly simulated and presented with his
recent post at ABSE.
At 3.2Hz, each of the L and C reactances = 5 ohms at Fo = 3.2 Hz.
If the R across the C were to be 7 ohms, the
- 3 dB point of the filter would be at 3.2 Hz, and the response unpeaked
and thus *non resonant*.

But since the LC filter is to be used for a DC power supply, placing
7 ohms across the C would create a huge waste of DC flow, and
so that idea isn't practical.

But then if you had 7 ohms from the 10,000 uF to say
100,000 uF as a low frequency snubber network,
then indeed the the LC filter would be critically damped without
wasting DC power.

One could also apply an active dynamic shunt impedance
which acts at least like a 7 ohm resistive impedance, but which consists
of a power transistor which only conducts when an AC signal at the C
exceeds a certain threshold at around 3.2 Hz, which suggests
the signal to the base of the transistor be from a
parallel resonant circuit fed by an R from the 10,000 uf.

In other words we could use what may be a simple shunt regulator,
which passes no more direct current at idle than say 220 mA,
but which otherwise acts like at least 7 ohm resistor, but only
where its needed.
I believe such a scheme could be not unlike a "capacitance multiplyer",
and be a lot cheaper and easier to fit into the amp than
40,000 uF, but in fact it would be like a Q divider, rather than like a
Q multiplyer, used in RF circuitry to increase the Q of a resonant circuit
to increase selectivity.
Any active device across the C would have to be protected
against back emfs in the choke.

The next idea that comes to mind is to place
7 ohms across the choke, which would damp the resonance peak in the
LC, but which would change the rate of attenuation
of the filter and make the filter into an RC filter with 6 dB/octave slope
above the pole F of around 3.2 Hz,
so that for an LC input filter, the
attenutation of the 100 Hz rectifier hum would be less than with a pure
LC filter.
With the LC values chosen, at 100 Hz, ZL = 157 ohms,
and ZC = 0.16 ohms, so the approximate 24 vrms of 100Hz signal
at the input to the choke is a factor of 0.16 / 157 = 0.001, so the hum
is reduced to 24 mV.

With 7 ohms across the L, the Z ( R+L ) at 100 Hz
ia approx 7 ohms, so the
attenuation of the 24vrms 100 Hz would be down to
( 0.16 / 7 ) x 24v = 0.54 volts.

But no DC power losses would occur in the R across the choke, because
nearly all the DC power loss is in the dcR of the choke, and little
DC flows in the 7 ohms across the L.

Perhaps slightly more than 7 ohms would be sufficient to damp the LC resonance,
maybe 10 ohms is enough, so I ask someone with a simulation
program to set up the LC filter connected to a full wave SS
rectifier from a 60 vrms secondary PT winding with a dcR = 1 ohm.

In a tube amp, a similar arrangement could be simulated.

You could have the following values
for an LC input filter for a class AB amp, where L is a swinging choke
with L = 35H at 22 mA, and drops to 11H at 200mA,
and C = 100 uF.
The power tranny winding for the full wave SS rectifier
has 10 ohm dcR, and there is 600vrms from the winding,
so there would be +530 volts at the C with Idc = 22 mA.
Mains F is assumed to be 50 Hz.
Critical damping R would have to be 700 ohms,
and so a snubber network with 700 ohms and say 1,000 uF across the
100 uF would remove resonance effects.

But immediately we would see that the 100 uF is perhaps
not sufficient to anchor the CT of the OPT,
where 1/4 of the RLa-a + the Ra of the tubes is connected to 100 uF,
so that if this load and source R was say 2,000 ohms,
then at 10 Hz the 100 uF has become 160 ohms,
and at 3.2 Hz it has become 480 ohms, and too high a value
to function as an anchor.
There is a strong case to increase the amount of C by a large factor,
which isn't hard in a world full of cheap small 470 uF x 450v electros.
4 in a series/parallel array will give 470 uF, to cope with the expected working voltage
of 530v or the soaring voltage of +848v, in case there was no
bleeder current of 22 mA.
The value of the choke, and its range of values must
be maintained irrespective of what sized C we choose, since the DC flow
commands the L value so that L must never be less than RL / 940.
( Notice that if Idc is zero, RL = infinity,
and so L would have to be an infinite value to make the B+ = 0.89 x Vrms
of the AC transformer winding. )
But if the bleeder R current was say 44 mA, then RL = 12k,
so L critical needs to be 12.7 H,
and at 200 mA, it can be allowed to fall to 2.8H.
It is good practice to use more L than the critical calculated values,
and hence 17H falling to 4.3H would be about right,
and at a current where L was in fact 12.7H
and the C = 470 uF, the Fo would be 2 Hz,
and if anyone wants to say such a resonant F will upset the music, they
best get a move on to proove it, remembering
that there isn't much 2 Hz info in music, and the
amp gain at 2 Hz is low, and that the OPT primary
impedance at 2 Hz is also rather low, etc.
With 200 mA of dc flow, and with
4.3H in series with 470 uF, the
240vrms of 100Hz from the rectifier is reduced by a factor of ZC / ZL =
Z 470uF / Z 4.3H = 3.4 / 2,700 = 0.0012,
so the hum would be 0.3vrms at the CT at 200 mA of dc flow.
But Fo would have risen to 3.54 Hz,
but then the RL connected across the C = 530 / 0.2 = 2,650 ohms,
which would help damp the resonance to some extent.

There is a another set of circumstances to analyse on your simulators.

Patrick Turner.


  #40   Report Post  
John Byrns
 
Posts: n/a
Default

In article , Patrick Turner
wrote:

As a follow up to the thread I started above,
I have some test measurements which may be of interest to
people such as John Byrns, who said that the resonant behaviour
of the LC filter used with a class B amplifier would result in being
forced to limit the signal output
to 3dB less than theoretical maximum, or words to that effect.


I never said words to that effect! What I actually said, complete with
typo, was "reducing the available output power by as much a 3 dB on a
dynamic basis." I didn't say that an LC power supply filter always
reduces the output by 3 dB, I said "by as much as 3 dB", it can be less
depending on the design of the power supply and filter.

I can't actually take credit for the 3 dB number, I took it from the
literature on the subject. This is not a new effect, you need read
something besides the RDH4, you might want to stop by the library to
review the existing literature on the subject.

I just ran a test on the LC input filter described here :-

L is a swinging choke
with L = 0.35H at 228 mA, and drops to 0.11H at 2 amps,
and C = 10,000 uF,
The power tranny winding for the full wave SS rectifier
has 1 ohm dcR, and there is 65vrms from the winding,
so there would be +57 volts at the C with Idc = 228 mA.
Mains F is 50 Hz.


The resistance to get 228mA was 250 ohms, and
I then tried repeatedly adding 32 ohms, to make the supply voltage fall
to 53v, and DC current increase to 1.87 amps.
The speed at which I added/removed the 32 ohms was varied to generate
the highest possible signal voltage across the C, and all was monitored
with a CRO, and the voltage variation with an F around 4 Hz was 4.6 vrms,
or about +/-7 peak volts, and I seriously doubt that any music signal
condition could generate a greater disturbance to the supply.
But where the F of adding subtracting the 32 ohms to cause the transient
was reduced to less or more than around 4 Hz, the voltage swing of the supply
was about 1/2 the 7 volts peak.
So when the supply has a sudden 1.6 amp increase in load current, the rail
dips to just under the final rail volts before stabilising at 53v, and when
the current is reduced 1.6 amps,
the rail voltage swings up to above the 57v rail before settling at 57v.
There was no more than about 1/2 a "ring cycle".


1/2 a "ring cycle" is more than enough to cause trouble. Your results
sound like your design has about a 1 dB loss of output due to power supply
bounce, although your test results as presented are incomplete, so 1 dB is
only an estimate on my part. 1 dB fits within the "as much a 3 dB"
description I posted, obviously your design is not nearly as bad as it
might be, although I don't think 1 dB is good enough to meet the FCC
specs. for AM broadcast transmitters in the USA when the modulator and RF
final amplifier share the same power supply as is typical.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/
Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT +1. The time now is 07:03 PM.

Powered by: vBulletin
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 AudioBanter.com.
The comments are property of their posters.
 

About Us

"It's about Audio and hi-fi"