Go to ABSE to see a demo of the resonance I spoke of a few
days ago. The filter resonance is excited by load current changes
in the equipment, in our case an amplifier. This kind of phenomena
will occur with any CLC filter network as commonly seen in
power supplies. The frequency of resonance is usually below the
audio range, as it is in this example. Cheers, John Stewart

I have an old Wurlitzer electronic organ, with a field coil loudspeaker
arranged as the filter choke. Unit was built in 1951.

About 1973, the thing developed a strange problem, if you played low "D" on
the pedal the note got louder and louder and finally there was a big
crackling noise (extremely loud) and the process would repeat.

Problem? Burned out input cap in a CLC filter. The last one, which fed the
voltage amplifier and the screens on the output tubes was OK, but the input
cap was gone. With cap gone, the resonance of the power supply moved way up,
and low "D" would excite it. It would build up until the 6L6Gs flashed over.
I think it flashed from plate to screen, but I am not sure. You could
definitely see it, however. Good thing I didn't burn out the field in the
speaker.

Curious, that was the only obvious clue to the problem. The hum rejectiion
of the output stage was fine, PP of course, with the screen supply still
filtered.

"John Stewart" wrote in message
...
Go to ABSE to see a demo of the resonance I spoke of a few
days ago. The filter resonance is excited by load current changes
in the equipment, in our case an amplifier. This kind of phenomena
will occur with any CLC filter network as commonly seen in
power supplies. The frequency of resonance is usually below the
audio range, as it is in this example. Cheers, John Stewart

What are the numbers on the amps that you have?
I managed to salvage 3 amps and 2 12" field coils and 2 15" field coils out
of a Wurlitzer that was being pitched.
too bad the idiots didnt let me get there first
they had to see how strong the paper was in a few of the speakers and
punched holes in them.
I managed to fix the speaker paper but I havent run the amps yet as I dont
know how they were run.

Each amp of mine has plugs for 2 speakers they are 7 pin much like the 6pin
on a Hammond B3/Leslie 122 cable.
anyways my amps are numbered as Model 6420

thanks in advance for any help

Doug

"BFoelsch" wrote in message
...
I have an old Wurlitzer electronic organ, with a field coil loudspeaker
arranged as the filter choke. Unit was built in 1951.

About 1973, the thing developed a strange problem, if you played low "D"
on the pedal the note got louder and louder and finally there was a big
crackling noise (extremely loud) and the process would repeat.

Problem? Burned out input cap in a CLC filter. The last one, which fed the
voltage amplifier and the screens on the output tubes was OK, but the
input cap was gone. With cap gone, the resonance of the power supply moved
way up, and low "D" would excite it. It would build up until the 6L6Gs
flashed over. I think it flashed from plate to screen, but I am not sure.
You could definitely see it, however. Good thing I didn't burn out the
field in the speaker.

Curious, that was the only obvious clue to the problem. The hum rejectiion
of the output stage was fine, PP of course, with the screen supply still
filtered.

"John Stewart" wrote in message
...
Go to ABSE to see a demo of the resonance I spoke of a few
days ago. The filter resonance is excited by load current changes
in the equipment, in our case an amplifier. This kind of phenomena
will occur with any CLC filter network as commonly seen in
power supplies. The frequency of resonance is usually below the
audio range, as it is in this example. Cheers, John Stewart

Hi John ,

Looks to me you tried realy hard to proof this .....
R-load = R-choke-dc and R-psu-trannie is 200 x that .....
What fool would build a PSU like that ?
To get a B+ on the load of just 200V you need a
(200 x2) x 200V = 80,000V trannie ....

Do I look at it the wrong way or am I just mad ......

Ronald .

"John Stewart" schreef in bericht
...
Go to ABSE to see a demo of the resonance I spoke of a few
days ago. The filter resonance is excited by load current changes
in the equipment, in our case an amplifier. This kind of phenomena
will occur with any CLC filter network as commonly seen in
power supplies. The frequency of resonance is usually below the
audio range, as it is in this example. Cheers, John Stewart

I couldn't view JS's post here on the CLC filter, I could only see his header
on the list.
news.individual doesn't always gather every post from around the globe,
and maybe a kookaburra occasionally forgets to place the digits in the right
box at the right time.

But John kindly sent me his post with schematic
and resonant response details.

Here is a near copy of what I sent back to JH,
with some additional editing for the group :-

--------------

John Stewart wrote:

Go to ABSE to see a demo of the resonance I spoke of a few
days ago. The filter resonance is excited by load current changes
in the equipment, in our case an amplifier. This kind of phenomena
will occur with any CLC filter network as commonly seen in
power supplies. The frequency of resonance is usually below the
audio range, as it is in this example. Cheers, John Stewart

The revised schematic of a CLC filter is much more like what one would see in
the real world,
and I commend you on the presentation with appropriate notes.

The response of the filter shows a 16 dB peak at 3.2Hz.

While this may seem to be a disaster for the audio of the amp,
I never have seen any problem, if the CLC components you show were indeed the
basis
for an SS amplifier's supply filter, were such a filter used
the ripple filtering would be quite fabulous, even if 5 amps DC were to be
drawn, which would give 1.1v 100 Hz ripple at C1, and about 1 mV of ripple
at C2.

Mains fluctuations here are typically +/- 20 mV and you can watch the
rectified voltages leaping about when the CRO is set to a sensitive position.
Its caused by everyone else in the area turning things on and off like stoves,
heaters,
aircon, hot water heaters, large SS amps, welders, etc.
But when the mains is transformed down to 55v, the mains "jitter"
is reduced a heck of a lot, and the amount of 3.2 Hz noise which will
cause a 16 dB peak at C2 is never going to spoil the audio.
Speakers can't reproduce much sound at 3.2 Hz, and the amount of cone
movements seen due to such supply
voltage undulations is sweet ferk awl.


The impedance of the 10,000 uF at 3.2Hz = 5 ohms, and this is a sufficiently
low impedance to prevent major v swings from a 200 ohm current source
you have detailed in your output load on your filter.
As I have explained at length in other posts, the R which terminates the
L-C2 filter needs to be 1.41 x ZL or ZC at 3.2 Hz to prevent the peaked LC
filter response.

In fact, the 200 ohms + voltage source is a silly real world value, and
doesn't represent a real amp at all, since if you have a typical collector load
from an emitter follower output stage connected to the C2, then the you have a
much higher impedance for the signal source than 200 ohms, but one which has
real grunt, because
you have power transistors in series with a lowZ load in series with the
supply cap.
Such a termination is a voltage generator in series with the dynamic collector
impedance,
plus the load, which is small by comparison.

So voltage movements at C2 are going to be inevitable if you
apply a signal in the amp at 3.2 Hz.
And and in fact at 0.1 Hz, the
10,000 uF has become a reactance of 160 ohms, and the
collector supply is virtually connected through your
1 ohm of total source impedance plus the 1 ohm dcR of the L.
So the response may be different again if the source is considered as a
rectified DV from a transformer,
and so clean DC operation is somewhat a myth, because it all depends on the
source being very low impedance indeed.
Maybe one needs several 200 amp hour truck batteries perhaps?
( don't laugh, I have seen an amp which did use large
auto batteries ).

Meanwhile at 16 Hz, the ZC2 = 1 ohm, and it is sufficiently low
to prevent much rail voltage change with transistors connected via the load of
4 ohms or higher, and the amount of total emitter follower voltage NFB
and global voltage NFB is about 80 dB at least, so any
IMD caused by the slight rail movements will be
also sweet fanny appleby.

Even if one sits there turning the amp on and off again repeatedly,
making the rail voltage leap up and down +/-10v or more,
nothing is heard at the output.
Well, all well designed SS amps will still work fine under such
idiotic working conditions.
The disturbances caused by the resonant behaviour *in this case*,
and in this case only if such a filter was used for a high power SS amp would be
minor, imho.

I still cannot see any reason not to use a CLC, or a simpler LC filter
as I have suggested.

Regards,

Patrick Turner.







------------------------------------------------------------------------
[Image] [Image]

Patrick, I don't understand what your point is with respect to the power
supply bounce problem that John Stewart has pointed out? There is no
question that this is a real effect that can be a serious problem with LC
type power supply filters. This was a significant problem with some AM
broadcast transmitters that used class B audio modulators. It became only
too apparent as higher, tighter, modulation levels were demanded from
transmitters, and it was discovered that the power supply bounce at
infrasonic frequencies was seriously limiting the modulation capability.
Granted this is not a problem with class A amplifiers where the current
draw from the power supply is constant, but it is an issue in class B
amplifiers. Many solid state amplifiers operate in class B and would be
subject to the problem if they used LC filters in the power supply. I
suspect the problem might not be too noticeable though in a home stereo
amplifier which operates with a lot of extra head room, unlike an AM
transmitter that is operating on the edge of its capability all the time.

In an earlier post you mentioned the use of a 10 Hz high pass filter, but
this does not eliminate the problem due to power supply bounce. As a
simple example to illustrate the problem, consider an amplifier with a
power supply resonance at 3.2 Hz as in John's example, and feed the
amplifier through a 10 Hz high pass filter with a 1 kHz sine that is keyed
on and off at a 3.2 Hz rate. This type of input signal, which I believe
is common in both voice and music, will pass through the high pass filter
unaffected, and will cause the current drawn from the power supply of a
class B amplifier to be modulated at a 3.2 Hz rate, thereby exciting the
resonance in the power supply and reducing the undistorted output that can
be obtained from the amplifier.


Regards,

John Byrns


In article , Patrick Turner
wrote:

I couldn't view JS's post here on the CLC filter, I could only see his header
on the list.
news.individual doesn't always gather every post from around the globe,
and maybe a kookaburra occasionally forgets to place the digits in the right
box at the right time.

But John kindly sent me his post with schematic
and resonant response details.

Here is a near copy of what I sent back to JH,
with some additional editing for the group :-

--------------

John Stewart wrote:

Go to ABSE to see a demo of the resonance I spoke of a few
days ago. The filter resonance is excited by load current changes
in the equipment, in our case an amplifier. This kind of phenomena
will occur with any CLC filter network as commonly seen in
power supplies. The frequency of resonance is usually below the
audio range, as it is in this example. Cheers, John Stewart

The revised schematic of a CLC filter is much more like what one would see in
the real world,
and I commend you on the presentation with appropriate notes.

The response of the filter shows a 16 dB peak at 3.2Hz.

While this may seem to be a disaster for the audio of the amp,
I never have seen any problem, if the CLC components you show were indeed the
basis
for an SS amplifier's supply filter, were such a filter used
the ripple filtering would be quite fabulous, even if 5 amps DC were to be
drawn, which would give 1.1v 100 Hz ripple at C1, and about 1 mV of ripple
at C2.

Mains fluctuations here are typically +/- 20 mV and you can watch the
rectified voltages leaping about when the CRO is set to a sensitive position.
Its caused by everyone else in the area turning things on and off like stoves,
heaters,
aircon, hot water heaters, large SS amps, welders, etc.
But when the mains is transformed down to 55v, the mains "jitter"
is reduced a heck of a lot, and the amount of 3.2 Hz noise which will
cause a 16 dB peak at C2 is never going to spoil the audio.
Speakers can't reproduce much sound at 3.2 Hz, and the amount of cone
movements seen due to such supply
voltage undulations is sweet ferk awl.


The impedance of the 10,000 uF at 3.2Hz = 5 ohms, and this is a sufficiently
low impedance to prevent major v swings from a 200 ohm current source
you have detailed in your output load on your filter.
As I have explained at length in other posts, the R which terminates the
L-C2 filter needs to be 1.41 x ZL or ZC at 3.2 Hz to prevent the peaked LC
filter response.

In fact, the 200 ohms + voltage source is a silly real world value, and
doesn't represent a real amp at all, since if you have a typical
collector load
from an emitter follower output stage connected to the C2, then the you have a
much higher impedance for the signal source than 200 ohms, but one which has
real grunt, because
you have power transistors in series with a lowZ load in series with the
supply cap.
Such a termination is a voltage generator in series with the dynamic collector
impedance,
plus the load, which is small by comparison.

So voltage movements at C2 are going to be inevitable if you
apply a signal in the amp at 3.2 Hz.
And and in fact at 0.1 Hz, the
10,000 uF has become a reactance of 160 ohms, and the
collector supply is virtually connected through your
1 ohm of total source impedance plus the 1 ohm dcR of the L.
So the response may be different again if the source is considered as a
rectified DV from a transformer,
and so clean DC operation is somewhat a myth, because it all depends on the
source being very low impedance indeed.
Maybe one needs several 200 amp hour truck batteries perhaps?
( don't laugh, I have seen an amp which did use large
auto batteries ).

Meanwhile at 16 Hz, the ZC2 = 1 ohm, and it is sufficiently low
to prevent much rail voltage change with transistors connected via the load of
4 ohms or higher, and the amount of total emitter follower voltage NFB
and global voltage NFB is about 80 dB at least, so any
IMD caused by the slight rail movements will be
also sweet fanny appleby.

Even if one sits there turning the amp on and off again repeatedly,
making the rail voltage leap up and down +/-10v or more,
nothing is heard at the output.
Well, all well designed SS amps will still work fine under such
idiotic working conditions.
The disturbances caused by the resonant behaviour *in this case*,
and in this case only if such a filter was used for a high power SS amp
would be
minor, imho.

I still cannot see any reason not to use a CLC, or a simpler LC filter
as I have suggested.

Regards,

Patrick Turner.


Surf my web pages at, http://users.rcn.com/jbyrns/

John Byrns wrote:

Patrick, I don't understand what your point is with respect to the power
supply bounce problem that John Stewart has pointed out?

John,
I don't understand what your point is with respect to the power
supply bounce "problem" that John Stewart has pointed out.

Please inform the group asap your proof that there is a problem with
the LC filter, or else get disbelieved, your credibility laying in tatters.

There is no
question that this is a real effect that can be a serious problem with LC
type power supply filters.

Sure, but I don't use values which cause the slightest problem.

Please read your old tube theory books, if you don't believe me,
for I don't intend arguing another long silly argument, and repeating myself
69 times because you don't get it. I already have pointed out in about 6 posts
what I think on the subject.

This was a significant problem with some AM
broadcast transmitters that used class B audio modulators.

So darn what? I didn't build the class B transmitters.
Blame the bean counter who did, and who tried to save on iron,
wire, capacitance....

It became only
too apparent as higher, tighter, modulation levels were demanded from
transmitters,

They have always been pushed hard.....

and it was discovered that the power supply bounce at
infrasonic frequencies was seriously limiting the modulation capability.
Granted this is not a problem with class A amplifiers where the current
draw from the power supply is constant,

But here you are wrong. The power supply doesn't change,
but SE which you include by saying "class A", will excite the resonance of the
LC filter if the F of the signal is taken down low enough, or the LC filter was
designed
by a ****wit who reckoned Fo = 50 Hz was OK.


but it is an issue in class B
amplifiers.

Its an issue is all amplifiers, period.

Many solid state amplifiers operate in class B and would be
subject to the problem if they used LC filters in the power supply. I
suspect the problem might not be too noticeable though in a home stereo
amplifier which operates with a lot of extra head room, unlike an AM
transmitter that is operating on the edge of its capability all the time.

So if there is zero problem with a domestic SS amplifier, even a sub woofer,
then what's your grouch?

AM transmitters need simply be designed with *hughmungous" chokes,
and "enormous" capacitors, and all will be well.

I have a power supply I picked up from a re-cycler just before he was to attack it
for the scarp value, and it puts out about 1,600v, and sure its got an LC
input filter, with huge L&C.





In an earlier post you mentioned the use of a 10 Hz high pass filter, but
this does not eliminate the problem due to power supply bounce.

I definately did no such thing.
What I did say was that you could add a C and R snubber network across the
L to make it resonant to 100 Hz, and I said it only works in CLC
filters where the Idc is constant, because L varies somewhat with Idc.
In a swinging choke design, the L may vary
from say 0.35H at 220 mA to 0.11H at 2 amps, and so the Fo
would change, so there would be less precise rejection of
the 100 Hz rectifier ripple.

I have used such C+R networks across L in CLC power supplies,
and enjoyed a reduction of 100 Hz hum of 15 dB, without increasing
higher harmonics much at all, and its like using a choke value 5 times the size.

But you wouldn't know about that eh.



As a
simple example to illustrate the problem, consider an amplifier with a
power supply resonance at 3.2 Hz as in John's example, and feed the
amplifier through a 10 Hz high pass filter with a 1 kHz sine that is keyed
on and off at a 3.2 Hz rate.

I will not consider an amplifier with such a high pass filter unless
you post a complete schematic and analysis of the circuit involved.
That should take you several hours to prepare.

Please, don't bull**** without a specific case to refer to,
it is intellectually wrong to try to bamboozle reading members in our group this
way.

Many others reading wouldn't have a clue about your concerns.


This type of input signal, which I believe
is common in both voice and music, will pass through the high pass filter
unaffected, and will cause the current drawn from the power supply of a
class B amplifier to be modulated at a 3.2 Hz rate, thereby exciting the
resonance in the power supply and reducing the undistorted output that can
be obtained from the amplifier.

Try connecting an amp you have to an LC filter with Fo = 3.2Hz,
and try sending your bursts of 1 kHz starting and stopping at whatever
F you want.

Record all your signal voltages and photograph all your CRO traces,
and let us know how you get on.

Define the problem, if there is one.

Then I might listen more to what you are saying.

Meanwhile, *well designed* CLC filters don't suck,
unless somebody proves to me they do,
which nobody has successfully done so far.

Patrick Turner



Regards,

John Byrns

In article , Patrick Turner
wrote:

I couldn't view JS's post here on the CLC filter, I could only see his header
on the list.
news.individual doesn't always gather every post from around the globe,
and maybe a kookaburra occasionally forgets to place the digits in the right
box at the right time.

But John kindly sent me his post with schematic
and resonant response details.

Here is a near copy of what I sent back to JH,
with some additional editing for the group :-

--------------

John Stewart wrote:

Go to ABSE to see a demo of the resonance I spoke of a few
days ago. The filter resonance is excited by load current changes
in the equipment, in our case an amplifier. This kind of phenomena
will occur with any CLC filter network as commonly seen in
power supplies. The frequency of resonance is usually below the
audio range, as it is in this example. Cheers, John Stewart

The revised schematic of a CLC filter is much more like what one would see in
the real world,
and I commend you on the presentation with appropriate notes.

The response of the filter shows a 16 dB peak at 3.2Hz.

While this may seem to be a disaster for the audio of the amp,
I never have seen any problem, if the CLC components you show were indeed the
basis
for an SS amplifier's supply filter, were such a filter used
the ripple filtering would be quite fabulous, even if 5 amps DC were to be
drawn, which would give 1.1v 100 Hz ripple at C1, and about 1 mV of ripple
at C2.

Mains fluctuations here are typically +/- 20 mV and you can watch the
rectified voltages leaping about when the CRO is set to a sensitive position.
Its caused by everyone else in the area turning things on and off like stoves,
heaters,
aircon, hot water heaters, large SS amps, welders, etc.
But when the mains is transformed down to 55v, the mains "jitter"
is reduced a heck of a lot, and the amount of 3.2 Hz noise which will
cause a 16 dB peak at C2 is never going to spoil the audio.
Speakers can't reproduce much sound at 3.2 Hz, and the amount of cone
movements seen due to such supply
voltage undulations is sweet ferk awl.


The impedance of the 10,000 uF at 3.2Hz = 5 ohms, and this is a sufficiently
low impedance to prevent major v swings from a 200 ohm current source
you have detailed in your output load on your filter.
As I have explained at length in other posts, the R which terminates the
L-C2 filter needs to be 1.41 x ZL or ZC at 3.2 Hz to prevent the peaked LC
filter response.

In fact, the 200 ohms + voltage source is a silly real world value, and
doesn't represent a real amp at all, since if you have a typical
collector load
from an emitter follower output stage connected to the C2, then the you have a
much higher impedance for the signal source than 200 ohms, but one which has
real grunt, because
you have power transistors in series with a lowZ load in series with the
supply cap.
Such a termination is a voltage generator in series with the dynamic collector
impedance,
plus the load, which is small by comparison.

So voltage movements at C2 are going to be inevitable if you
apply a signal in the amp at 3.2 Hz.
And and in fact at 0.1 Hz, the
10,000 uF has become a reactance of 160 ohms, and the
collector supply is virtually connected through your
1 ohm of total source impedance plus the 1 ohm dcR of the L.
So the response may be different again if the source is considered as a
rectified DV from a transformer,
and so clean DC operation is somewhat a myth, because it all depends on the
source being very low impedance indeed.
Maybe one needs several 200 amp hour truck batteries perhaps?
( don't laugh, I have seen an amp which did use large
auto batteries ).

Meanwhile at 16 Hz, the ZC2 = 1 ohm, and it is sufficiently low
to prevent much rail voltage change with transistors connected via the load of
4 ohms or higher, and the amount of total emitter follower voltage NFB
and global voltage NFB is about 80 dB at least, so any
IMD caused by the slight rail movements will be
also sweet fanny appleby.

Even if one sits there turning the amp on and off again repeatedly,
making the rail voltage leap up and down +/-10v or more,
nothing is heard at the output.
Well, all well designed SS amps will still work fine under such
idiotic working conditions.
The disturbances caused by the resonant behaviour *in this case*,
and in this case only if such a filter was used for a high power SS amp
would be
minor, imho.

I still cannot see any reason not to use a CLC, or a simpler LC filter
as I have suggested.

Regards,

Patrick Turner.

Surf my web pages at, http://users.rcn.com/jbyrns/

In article , Patrick Turner
wrote:

John Byrns wrote:

Patrick, I don't understand what your point is with respect to the power
supply bounce problem that John Stewart has pointed out?

I don't understand what your point is with respect to the power
supply bounce "problem" that John Stewart has pointed out.

My point, actually a question, is on what basis you say there is no power
supply bounce problem? This is a well known problem with class B
amplifiers using LC filters in the power supply.

Please inform the group asap your proof that there is a problem with
the LC filter, or else get disbelieved, your credibility laying in tatters.

There is no
question that this is a real effect that can be a serious problem with LC
type power supply filters.

Sure, but I don't use values which cause the slightest problem.

What values do you use to avoid the problem? How do you know you have
avoided the problem? If you are building class A amplifiers there will be
no problem in the first place. If you are building class B amplifiers
have you actually tested them with a suitable wave form to be sure there
isn't any power supply bounce and a loss of headroom? Keep in mind that
it isn't the frequency spectrum of the music or speech wave form that
triggers this problem, it is frequency spectrum of the envelope of the
music or speech signal. Putting a signal into a class B amplifier is like
putting it through an AM detector from the perspective of the power
supply. Think about it!

Most power supply filters of this type are under damped and hence resonate
at infrasonic frequencies, if you actually have a power supply design for
a class B amplifier which eliminates this problem, using an LC filter, why
don't you share the design and component values with us to prove your
point?

This was a significant problem with some AM
broadcast transmitters that used class B audio modulators.

So darn what? I didn't build the class B transmitters.
Blame the bean counter who did, and who tried to save on iron,
wire, capacitance....

Do you build class B amplifiers? I don't think it was entirely a matter
of bean counters. Again why don't you share with us the component values
that you say will solve this problem, which the bean counters didn't let
us use?

It became only
too apparent as higher, tighter, modulation levels were demanded from
transmitters,

They have always been pushed hard.....

How was that accomplished? The audio processing technology to do so only
became available about thirty years ago. Sure broadcasters have always
tried to push peaks up to or past 100%, but it is only in more recent
times that the average power and loudness has been greatly increased with
modern audio processing while still maintaining a modicum of fidelity.

and it was discovered that the power supply bounce at
infrasonic frequencies was seriously limiting the modulation capability.
Granted this is not a problem with class A amplifiers where the current
draw from the power supply is constant,

But here you are wrong. The power supply doesn't change,
but SE which you include by saying "class A", will excite the resonance of the
LC filter if the F of the signal is taken down low enough, or the LC filter
was designed by a ****wit who reckoned Fo = 50 Hz was OK.

Show me an LC power supply filter with an Fo of 50 Hz, Fo is typically in
the infrasonic region, and certainly below 50 Hz. A class A amplifier,
either PP or SE will only excite an infrasonic power supply resonance if
there is infrasonic energy in the program material , which generally there
isn't. A class B amplifier can excite power supply resonances even when
the program material contains no infrasonic energy, because of the varying
power demand it causes.

but it is an issue in class B
amplifiers.

Its an issue is all amplifiers, period.

Wrong, assuming the program material itself contains no infrasonic energy
it won't be a problem for a class A amplifier. You still haven't stopped
to think about how a class B amplifier operates, and the nature of the
current it draws from its power supply.

Many solid state amplifiers operate in class B and would be
subject to the problem if they used LC filters in the power supply. I
suspect the problem might not be too noticeable though in a home stereo
amplifier which operates with a lot of extra head room, unlike an AM
transmitter that is operating on the edge of its capability all the time.

So if there is zero problem with a domestic SS amplifier, even a sub woofer,
then what's your grouch?

If an amplifier has no problem I have no grouch. My grouch is that you
don't understand the problem and are making statements that are simply
wrong. You need to hit the books, this is not a newly discovered problem,
and it certainly isn't my theory, I'm not that smart.

AM transmitters need simply be designed with *hughmungous" chokes,
and "enormous" capacitors, and all will be well.

That would insure a greatly under damped filter, I assume that your point
is that if you push Fo low enough the problem will go away even if the
filter is under damped. I will give you the benefit of the doubt on that
for now, I don't know one way or the other if it would work or not, I will
have to think about it some. Larger value capacitors are practical today,
but where are you going to get those "hughmungous" chokes" by which I
assume you mean chokes with a very large inductance?

A better way for an AM broadcast transmitter to avoid the problem is to
use a twelve pulse rectifier in the power supply with only a simple
capacitor for filtering. The twelve pulse rectifier gets the hum down,
and a filter without an L can't resonate at infrasonic frequencies.

In an earlier post you mentioned the use of a 10 Hz high pass filter, but
this does not eliminate the problem due to power supply bounce.

I definately did no such thing.

You most certainly did, in message #
on October 24th you said; "The amp
in question will be used as a sub amp, but will still have input filtering
to give a pole at 10 Hz."

As a
simple example to illustrate the problem, consider an amplifier with a
power supply resonance at 3.2 Hz as in John's example, and feed the
amplifier through a 10 Hz high pass filter with a 1 kHz sine that is keyed
on and off at a 3.2 Hz rate.

I will not consider an amplifier with such a high pass filter unless
you post a complete schematic and analysis of the circuit involved.
That should take you several hours to prepare.

Please, don't bull**** without a specific case to refer to,
it is intellectually wrong to try to bamboozle reading members in our group
this way.

I suggest you think about how a class B amplifier operates, once you come
to an understanding of that the issue will become crystal clear.

Many others reading wouldn't have a clue about your concerns.

That's quite true but my point in posting is two fold, first to put the
truth on the record whether anyone understands it or not, and second to
put some tidbits out there for people to think about if they should become
interested later.

Meanwhile, *well designed* CLC filters don't suck,
unless somebody proves to me they do,
which nobody has successfully done so far.

That's not because LC filters don't have problems, or "suck" as you say,
it is simply because your skull is thick as a brick, as has been noted in
the past.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

Patrick, I don't understand what your point is with respect to the power
supply bounce problem that John Stewart has pointed out?

I don't understand what your point is with respect to the power
supply bounce "problem" that John Stewart has pointed out.

My point, actually a question, is on what basis you say there is no power
supply bounce problem? This is a well known problem with class B
amplifiers using LC filters in the power supply.

Please inform the group asap your proof that there is a problem with
the LC filter, or else get disbelieved, your credibility laying in tatters.

There is no
question that this is a real effect that can be a serious problem with LC
type power supply filters.

Sure, but I don't use values which cause the slightest problem.

What values do you use to avoid the problem?

Try reading all of the mountains of my postings about CLC and LC filters,
and if you don't get the general idea, then you never will.

How do you know you have
avoided the problem? If you are building class A amplifiers there will be
no problem in the first place.

Class A SE amps will excite resonances in PS.
Such amps are very good exciters of such resonances, better than a
class B PP amp.

If you are building class B amplifiers
have you actually tested them with a suitable wave form to be sure there
isn't any power supply bounce and a loss of headroom?

No loss of headroom at 25 Hz, perhaps some loss at 3.2 Hz, but then
we don't have any signals at 3.2 Hz....

Keep in mind that
it isn't the frequency spectrum of the music or speech wave form that
triggers this problem, it is frequency spectrum of the envelope of the
music or speech signal. Putting a signal into a class B amplifier is like
putting it through an AM detector from the perspective of the power
supply. Think about it!

Each rail sees a drain on the DC going one way when a burst of power at say 1 kHz is
applied,
and the rail sags maybe a max of 10% under the increased draw, and its this rail
sag which limits
the max power. Then when the burst stops, the current from the rectifier flows
through the choke to
fill it back up to the low current voltage, and perhaps there are some
ripples in way the cap fills back up, but all my observations revela that the
ripples are tiny,
and repeatedly shunting the cap with an R to increase the supply current
by 10 times from the idle condition did not cause any unwanted problems with LF
resonances.



Most power supply filters of this type are under damped and hence resonate
at infrasonic frequencies, if you actually have a power supply design for
a class B amplifier which eliminates this problem, using an LC filter, why
don't you share the design and component values with us to prove your
point?

I have shared all my info on numerous occasions, the latest being
a new post titled "LC filters, good or bad?"

Please read what I say, before telling the world I say nothing.

Nobody is yet game to debunk what I have said, but I welcome them to try.



This was a significant problem with some AM
broadcast transmitters that used class B audio modulators.

So darn what? I didn't build the class B transmitters.
Blame the bean counter who did, and who tried to save on iron,
wire, capacitance....

Do you build class B amplifiers?

I am building a substantially class B SS amp right now.
It will have a very small class A content, and Idc will vary
between about 250 mA at idle to a max of around 6 amps
under the worst load conditions of 2 ohms.

I don't think it was entirely a matter
of bean counters. Again why don't you share with us the component values
that you say will solve this problem, which the bean counters didn't let
us use?

I am re-furbishing an old Phase Linear amp. I didn't want to use the original +/-91
v rails.
In this case, I am using chokes to make LC input rail filters to get +/- 57v rails.
The chokes in their pots with roof pitch potting compound are 5 inches high
by about 4 inches square, and there isn't a company which would ever
use such technology, because nobody else does, and the reliance is on the
use of large cap values and cap input filters.
The alternative would have been to re-wind the power tranny, but I didn't want to do
that.

I have used cap inputs,
and on one amp I have 8" long x 3" dia 100,000 uF caps for each rail
on a 2 x 300 watt amp.

The chokes are a 2" stack of 1.25" tongue material with 345 turns of 1.25mm dia
wire,
with a gap of 0.21 mm each side of the E&I build up.


It became only
too apparent as higher, tighter, modulation levels were demanded from
transmitters,

They have always been pushed hard.....

How was that accomplished? The audio processing technology to do so only
became available about thirty years ago. Sure broadcasters have always
tried to push peaks up to or past 100%, but it is only in more recent
times that the average power and loudness has been greatly increased with
modern audio processing while still maintaining a modicum of fidelity.

Ever heard of compression? its been around for years.
RDH4 has several examples.



and it was discovered that the power supply bounce at
infrasonic frequencies was seriously limiting the modulation capability.
Granted this is not a problem with class A amplifiers where the current
draw from the power supply is constant,

But here you are wrong. The power supply doesn't change,
but SE which you include by saying "class A", will excite the resonance of the
LC filter if the F of the signal is taken down low enough, or the LC filter
was designed by a ****wit who reckoned Fo = 50 Hz was OK.

Show me an LC power supply filter with an Fo of 50 Hz, Fo is typically in
the infrasonic region, and certainly below 50 Hz.

There are a ****e load of ****wits in our world, but luckily not may are interested
in LC filter circuits, and so few examples of such circuits exist with Fo = 50 Hz.



A class A amplifier,
either PP or SE will only excite an infrasonic power supply resonance if
there is infrasonic energy in the program material , which generally there
isn't.

Indeed, so the resonance at below 10 Hz won't matter, now will it!

But you could send full range pink noise to an SE amp and infrasonic
LC PS resonaces would sure be excited, along with transformer
saturation effects etc...

A class B amplifier can excite power supply resonances even when
the program material contains no infrasonic energy, because of the varying
power demand it causes.

yes, but what's the problem?

Define it, quantify it, don't ****around endlessly talking about it from your damned
armchair!
Build something, measure it, and get back to us about your test results.


but it is an issue in class B
amplifiers.

Its an issue is all amplifiers, period.

Wrong, assuming the program material itself contains no infrasonic energy
it won't be a problem for a class A amplifier. You still haven't stopped
to think about how a class B amplifier operates, and the nature of the
current it draws from its power supply.

BS, I know exactly how class B amps work.



Many solid state amplifiers operate in class B and would be
subject to the problem if they used LC filters in the power supply. I
suspect the problem might not be too noticeable though in a home stereo
amplifier which operates with a lot of extra head room, unlike an AM
transmitter that is operating on the edge of its capability all the time.

So if there is zero problem with a domestic SS amplifier, even a sub woofer,
then what's your grouch?

If an amplifier has no problem I have no grouch. My grouch is that you
don't understand the problem and are making statements that are simply
wrong.

I do understand, and I said above, you have deliberately and obstinately
refused to read my posts fully, or go to your workshop
and fully test some circuits.


You need to hit the books, this is not a newly discovered problem,
and it certainly isn't my theory, I'm not that smart.

You lack of smartness is obvious, so get away from the PC,
your use of it only makes you look dumber by the minute,
and get to your workshop and *do something*,
rather than blather on trying to vainly criticise something without any facts
you have discovered for yourself.



AM transmitters need simply be designed with *hughmungous" chokes,
and "enormous" capacitors, and all will be well.

That would insure a greatly under damped filter, I assume that your point
is that if you push Fo low enough the problem will go away even if the
filter is under damped. I will give you the benefit of the doubt on that
for now, I don't know one way or the other if it would work or not, I will
have to think about it some. Larger value capacitors are practical today,
but where are you going to get those "hughmungous" chokes" by which I
assume you mean chokes with a very large inductance?

Wind the ferkin things!

Its not as if decent chokes are like rocking horse poo;
you have to go and wind them!!!!



A better way for an AM broadcast transmitter to avoid the problem is to
use a twelve pulse rectifier in the power supply with only a simple
capacitor for filtering. The twelve pulse rectifier gets the hum down,
and a filter without an L can't resonate at infrasonic frequencies.

We are not trying to build a supply for a transmitter here.
We don't want to enter the realm of 3 phase circuitry, or pulsed PS,
but we are discussing LC filters in simple homey gear.




In an earlier post you mentioned the use of a 10 Hz high pass filter, but
this does not eliminate the problem due to power supply bounce.

I definately did no such thing.

You most certainly did, in message #
on October 24th you said; "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 was refering to the pole of the amp response.
But I thought you were referring to the response of the PS filter.

Sorry for the misunderstanding of your unclear statement.



As a
simple example to illustrate the problem, consider an amplifier with a
power supply resonance at 3.2 Hz as in John's example, and feed the
amplifier through a 10 Hz high pass filter with a 1 kHz sine that is keyed
on and off at a 3.2 Hz rate.

I will not consider an amplifier with such a high pass filter unless
you post a complete schematic and analysis of the circuit involved.
That should take you several hours to prepare.

Please, don't bull**** without a specific case to refer to,
it is intellectually wrong to try to bamboozle reading members in our group
this way.

I suggest you think about how a class B amplifier operates, once you come
to an understanding of that the issue will become crystal clear.

I suggest you do some amp building to catch up with the rest of us.



Many others reading wouldn't have a clue about your concerns.

That's quite true but my point in posting is two fold, first to put the
truth on the record whether anyone understands it or not, and second to
put some tidbits out there for people to think about if they should become
interested later.

Meanwhile, *well designed* CLC filters don't suck,
unless somebody proves to me they do,
which nobody has successfully done so far.

That's not because LC filters don't have problems, or "suck" as you say,
it is simply because your skull is thick as a brick, as has been noted in
the past.

Be as personal as you like, but it won't make the problem of
your poor credibility on this whole issue go away.

You need to spend a day or 3 in your shed, and do some real research,
instead of bull****ting around to condemn PS ideas that have been around for
maybe 100 years, maybe longer, if we include telephone systems.
Many implementations of the ideas have been poor, and we can all cite cases where
the PS was a POS, and bean counters have a lot to answer for in allowing the
public to form the wrong general opinion about what is in fact a sensible idea,
if we leave out the bean counter notions of what is not a sensible
cost to be incurred.



Patrick Turner.

In article , Patrick Turner
wrote:

John Byrns wrote:

How do you know you have
avoided the problem? If you are building class A amplifiers there will be
no problem in the first place.

Class A SE amps will excite resonances in PS.
Such amps are very good exciters of such resonances, better than a
class B PP amp.

This is simply not correct and indicates that you still don't understand
the fundamental interaction of class B amplifiers and LC power supply
filters. A class A amplifier, either SE or PP, will not have any
infrasonic components in the power drawn from the power supply unless the
program material contains infrasonic components. On the other hand the
power drawn from the power supply by a class B amplifier can easily
contain infrasonic components because the average power drawn by a class B
amplifier varies at the "syllabic" rate following the envelope of the
speech or music wave form applied to the amplifier, even when the speech
or music wave form itself contains no infrasonic energy. This is the
point that seems to be eluding you.

If you are building class B amplifiers
have you actually tested them with a suitable wave form to be sure there
isn't any power supply bounce and a loss of headroom?

No loss of headroom at 25 Hz, perhaps some loss at 3.2 Hz, but then
we don't have any signals at 3.2 Hz....

The speech or music wave form doesn't contain any signals at 3.2 Hz, at
least if we have a suitable high pass filter at the input of the
amplifier, but the envelope of the signal wave form can easily contain
significant energy at frequencies like 3.2 Hz. Again you don't understand
the cause of the problem, it's not really that hard, just stop and think
about it a little.

There are effectively four questions to be answered here.

1. Can class B amplifiers produce infrasonic currents in the power supply
when the program signal feed into the class B amplifier contains no
infrasonic energy?

2. Can this infrasonic energy produce B+ "bounce" when LC filters are used
in the power supply?

3. Does the effect of power supply bounce on the dynamic headroom of a
domestic amplifier actually matter?

4. How can an LC power supply filter be designed to avoid the bounce problem?

The answers are.

#1. Yes, a class B amplifier can easily produce infrasonic variations in
the power drawn from the B+ supply, even with program material that
contains no infrasonic energy. This is obvious from an examination of the
operation of class B amplifiers.

#2. Yes, the infrasonic components of the power drawn by a class B
amplifier can excite resonances in LC power supply filters, typically
reducing the available output power by as much a 3 dB on a dynamic basis.

#3. Whether this loss of dynamic headroom is important in a domestic
amplifier is probably a subjective matter open to individual judgment.

#4. This is a potentially interesting question, especially when considered
in a commercial context, more discussion of this question is certainly
warranted.

Most power supply filters of this type are under damped and hence resonate
at infrasonic frequencies, if you actually have a power supply design for
a class B amplifier which eliminates this problem, using an LC filter, why
don't you share the design and component values with us to prove your
point?

I have shared all my info on numerous occasions, the latest being
a new post titled "LC filters, good or bad?"

Please read what I say, before telling the world I say nothing.

Our posts passed in the mail as it were, I had not read your "LC filters,
good or bad?" post when I posted my previous message. Your post makes a
good starting point towards answering question #4, but considerably more
analysis and design work is needed to determine whether any of your
suggested solutions can actually be made to work effectively.

Your suggestion of using pink noise as a test signal does not seem
entirely practical to me, and is clearly based on the fallacy that the
infrasonic energy is directly present in the program signal. A pink noise
test will certainly fail if the amplifier contains a high pass filter at
the input. Also it is not clear that a pink noise signal would contain
much energy at the resonant frequency of the power supply filter. A pink
noise signal contains considerable energy a very low frequencies
approaching DC which would swamp out the resonant effects in the power
supply. At the very least it would seem that you would need to cut the
pink noise off at a frequency just below the resonant frequency of the
power supply filter if it is to have a chance of working as a test
signal. I think a much better choice for a test signal is my suggestion
of a gated 1 kHz tone, where the amplitude of the tone would be adjusted
to drive the amplifier to full output, and then the tone would be gated on
and off with a square wave at the resonant frequency of the power supply
filter. This test signal, which itself contains no infrasonic energy,
would produce considerably more energy at the power supply resonant
frequency than would a pink noise signal driving the amplifier to full
output.

It became only
too apparent as higher, tighter, modulation levels were demanded from
transmitters,

They have always been pushed hard.....

How was that accomplished? The audio processing technology to do so only
became available about thirty years ago. Sure broadcasters have always
tried to push peaks up to or past 100%, but it is only in more recent
times that the average power and loudness has been greatly increased with
modern audio processing while still maintaining a modicum of fidelity.

Ever heard of compression? its been around for years.
RDH4 has several examples.

Yes I have heard of compressors, but the compressors and limiters used
years ago couldn't produce anywhere near the modulation density that for
example an AM Optimod can, at least with any pretense of fidelity.

In an earlier post you mentioned the use of a 10 Hz high pass filter,
but this does not eliminate the problem due to power supply bounce.

I definately did no such thing.

You most certainly did, in message #
on October 24th you said; "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 was refering to the pole of the amp response.
But I thought you were referring to the response of the PS filter.

Sorry for the misunderstanding of your unclear statement.

What was unclear about my statement? It was a pretty literal rewording of
your statement that I quoted above, or don't you understand your own
statement?


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

How do you know you have
avoided the problem? If you are building class A amplifiers there will be
no problem in the first place.

Class A SE amps will excite resonances in PS.
Such amps are very good exciters of such resonances, better than a
class B PP amp.

This is simply not correct and indicates that you still don't understand
the fundamental interaction of class B amplifiers and LC power supply
filters. A class A amplifier, either SE or PP, will not have any
infrasonic components in the power drawn from the power supply unless the
program material contains infrasonic components.

The infrasonic signals are found in full range pink noise
test signals, and need only be applied to the SE amp which can easily
be configured to have a cut off at 5 Hz at small % of their powers,
and and if any resonance between 1 Hz and 20 Kz can be seen by tracking the signal
response
at the supply cap after the L of an LC or CLC filter.
Its a standard test.

Or specifically, a large sine wave can be swept from say 1 Hz to 50 Hz
and any resonant behavoir of the PS is soon spotted.
I have done this plenty of times.

On the other hand the
power drawn from the power supply by a class B amplifier can easily
contain infrasonic components because the average power drawn by a class B
amplifier varies at the "syllabic" rate following the envelope of the
speech or music wave form applied to the amplifier, even when the speech
or music wave form itself contains no infrasonic energy. This is the
point that seems to be eluding you.

I read you loud and clear, and about the worst test signal tou coula apply
is to switch the PL after the LC from one which draws 0.2 amps, to 2 amps,
and watch the voltage level response at the C of the LC.
I have refered to doing this several times in the thread of my posts, but still
you accuse me of
not being aware, and of not understanding.

Nothing has eluded me.

So examine resonance, you either apply a signal at the resonant F,
or sweep past with a test signal, plotting the response,
or apply a transient wave, and examine the decay ripple.
And tou should do all this at a signal level far in excess of what may occur with
real world music conditions, ( or pink noise ).

When *YOU* have done your tests, get back to me and see if we agree,
but meanwhile stop being a lazy loudmouth.


If you are building class B amplifiers
have you actually tested them with a suitable wave form to be sure there
isn't any power supply bounce and a loss of headroom?

No loss of headroom at 25 Hz, perhaps some loss at 3.2 Hz, but then
we don't have any signals at 3.2 Hz....

The speech or music wave form doesn't contain any signals at 3.2 Hz, at
least if we have a suitable high pass filter at the input of the
amplifier, but the envelope of the signal wave form can easily contain
significant energy at frequencies like 3.2 Hz. Again you don't understand
the cause of the problem, it's not really that hard, just stop and think
about it a little.

No need to make a bigger fool of yourself by repeating, and implying that I have
not
thought of what is obvious to me.



There are effectively four questions to be answered here.

1. Can class B amplifiers produce infrasonic currents in the power supply
when the program signal feed into the class B amplifier contains no
infrasonic energy?

Yes of course they can.


2. Can this infrasonic energy produce B+ "bounce" when LC filters are used
in the power supply?

Perhaps, but just how much bounce depends on design, and the
level of signal, and we await John Byrn's test results to see what happens.
Don't hold your breath folks, he's allergic to doing anything in his workshop.



3. Does the effect of power supply bounce on the dynamic headroom of a
domestic amplifier actually matter?

Good question.
Depends on a few things. What are they John?


4. How can an LC power supply filter be designed to avoid the bounce problem?

Use lots of L and C.



The answers are.

#1. Yes, a class B amplifier can easily produce infrasonic variations in
the power drawn from the B+ supply, even with program material that
contains no infrasonic energy. This is obvious from an examination of the
operation of class B amplifiers.

Sure, but I think you overstate a problem



#2. Yes, the infrasonic components of the power drawn by a class B
amplifier can excite resonances in LC power supply filters, typically
reducing the available output power by as much a 3 dB on a dynamic basis.

So you say that the maximum voltage output of a a class B amp with music is
reduced 3 dB,
therefore only 1/2 full power is available since clipping will raise thd
drastically above what
is theoretically expected.

Where are your test results to proove such a typical situation occurs????



#3. Whether this loss of dynamic headroom is important in a domestic
amplifier is probably a subjective matter open to individual judgment.

We are not sure it happens John.

Don't be lazy, get busy to find out in your workshop.


#4. This is a potentially interesting question, especially when considered
in a commercial context, more discussion of this question is certainly
warranted.

Commercial interest will always try to whitle down the
cost of the the power supply, and output transformer size, etc, and keep the
profits high.

Quad II is a typical example of a "bean counter" implemtation of what is a very
fine original idea.

But without most "commercial interest's" disinterest in what costs $2 more than
the lowest possible
denominator, I suggest some fine amps can be built with CLC or LC input filters.



Most power supply filters of this type are under damped and hence resonate
at infrasonic frequencies, if you actually have a power supply design for
a class B amplifier which eliminates this problem, using an LC filter, why
don't you share the design and component values with us to prove your
point?

I have shared all my info on numerous occasions, the latest being
a new post titled "LC filters, good or bad?"

Please read what I say, before telling the world I say nothing.

Our posts passed in the mail as it were, I had not read your "LC filters,
good or bad?" post when I posted my previous message. Your post makes a
good starting point towards answering question #4, but considerably more
analysis and design work is needed to determine whether any of your
suggested solutions can actually be made to work effectively.

You are just the man to do all the R&D, so the world may see
whether the same test results can be gained by someone else
independantly duplicating the experiments.

Whether something in science is true or not depends on
the outcomes of independant experiments concurring.

If you ain't gonna do the tests, then what can you say you really know?



Your suggestion of using pink noise as a test signal does not seem
entirely practical to me, and is clearly based on the fallacy that the
infrasonic energy is directly present in the program signal.

Its VERY practical. I have a pink noise tester which is built for testing
speakers, but it does well with amps.
It uses a small signal current passing through a transistor backwards,
and the resulting noise is amplified, then filtered
through a LPF to convert it from white noise to pink noise,
then there is an eq circuit to make sure the levels of signal at any F from around
2 Hz to 25 kHz
when tested using a tunable filter with a Q of over 20.
So I know this signal will provide plenty of signal at LF to test
amps better than simply using a rock'n'roll or heavy metal signal.


A pink noise
test will certainly fail if the amplifier contains a high pass filter at
the input.

Rubbish.
The applied signal is the full audio band.

If the HF component above say 100 Hz is filtered out,
then an even better picture of the LF signal behaviour is gained.

Also it is not clear that a pink noise signal would contain
much energy at the resonant frequency of the power supply filter.

The random noise at any F is reatively equal anywhere along the band.

A pink
noise signal contains considerable energy a very low frequencies
approaching DC which would swamp out the resonant effects in the power
supply.

No.

The pink noise is very much like music, except that the tones in pink noise
are not musically related. Heavy metal signals are almost
as randomly unmusical as pink noise.

At the very least it would seem that you would need to cut the
pink noise off at a frequency just below the resonant frequency of the
power supply filter if it is to have a chance of working as a test
signal.

This always improves the test results one heck of a lot.

In a tube amp, adding a pole at say 14 Hz prevents most of the saturation seen
with a pink noise signal, even at 1/2 the output voltage levels, ( 1/4 power )

The average maximum pink noise signal output levels will be considerably lower
than a sine wave
test because like music, the pink noise signal level has a huge dynamic range, and
the range is randomly
varying, along with random variations in F and phase.

The level at which any amplifier can be used with music is determined by what
level
clipping begins to occur with the high peaks of the programme or test signal,
so any amp with a 300 watt capacity usually has no more than
30 watts asked of it, which means the average signal level is about 1/3 of the
full sine wave instantaneous level at voltage clipping.

A large amount of compression is applied to pop "music" to allow
fuller use of the available power, to get the sound levels up to ruin the young
folks hearing,
( but oh how exciting it all is for the little darlings ).
But still the average power level won't be more than 1/2 the maximum,
say at 0.7 of the max output signal.




I think a much better choice for a test signal is my suggestion
of a gated 1 kHz tone, where the amplitude of the tone would be adjusted
to drive the amplifier to full output, and then the tone would be gated on
and off with a square wave at the resonant frequency of the power supply
filter.

Yes, such a test signal is a rigourous one.

Let me know your test results when you have perfomed them.

This test signal, which itself contains no infrasonic energy,
would produce considerably more energy at the power supply resonant
frequency than would a pink noise signal driving the amplifier to full
output.

We await the results of your tests.



It became only
too apparent as higher, tighter, modulation levels were demanded from
transmitters,

They have always been pushed hard.....

How was that accomplished? The audio processing technology to do so only
became available about thirty years ago. Sure broadcasters have always
tried to push peaks up to or past 100%, but it is only in more recent
times that the average power and loudness has been greatly increased with
modern audio processing while still maintaining a modicum of fidelity.

Ever heard of compression? its been around for years.
RDH4 has several examples.

Yes I have heard of compressors, but the compressors and limiters used
years ago couldn't produce anywhere near the modulation density that for
example an AM Optimod can, at least with any pretense of fidelity.

The problem with compression is that you are applying
gross 3rd harmonic distortion by flattening the range of the signal swing.

But perhaps a PC program does it all better; I have never played with any of that
****,
since I am solely interested in *adequate* if not superlative fidelity in my hi-fi
amps.




In an earlier post you mentioned the use of a 10 Hz high pass filter,
but this does not eliminate the problem due to power supply bounce.

I definately did no such thing.

You most certainly did, in message #
on October 24th you said; "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 was refering to the pole of the amp response.
But I thought you were referring to the response of the PS filter.

Sorry for the misunderstanding of your unclear statement.

What was unclear about my statement? It was a pretty literal rewording of
your statement that I quoted above, or don't you understand your own
statement?

Don't let yourself become more muddled than you already are.

Patrick Turner.



Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

In article , Patrick Turner
wrote:

John Byrns wrote:

A pink noise
test will certainly fail if the amplifier contains a high pass filter at
the input.

Rubbish.
The applied signal is the full audio band.

Many amplifiers incorporate high pass filters at the input, so a "full
audio band" signal won't do the job.

If the HF component above say 100 Hz is filtered out,
then an even better picture of the LF signal behaviour is gained.

A high pass filter eliminates energy below the cutoff freqeuncy of the
filter, not above the cutoff freqeuncy.

Also it is not clear that a pink noise signal would contain
much energy at the resonant frequency of the power supply filter.

The random noise at any F is reatively equal anywhere along the band.

I think you are confusing pink noise with white noise here.

The problem with compression is that you are applying
gross 3rd harmonic distortion by flattening the range of the signal swing.

"gross 3rd harmonic distortion" is not a problem in a well designed
compressor. The control signal in a compressor is relatively slow acting
so that if the VGA is blameless distortion should be minimal. I think you
are confusing compression with clipping which does introduce fairly
serious distortion, although I would say that inter modulation distortion
is a bigger problem than 3rd harmonic distortion. The clippers in modern
audio processors also have so called "distortion cancellation" circuitry
which is supposed to mitigate the problem.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

A pink noise
test will certainly fail if the amplifier contains a high pass filter at
the input.

Rubbish.
The applied signal is the full audio band.

Many amplifiers incorporate high pass filters at the input, so a "full
audio band" signal won't do the job.

Indeed such filters exist, and indeed the effects of the LF content of pink
noise is avoided
by this means, but when testing audio amps, its advisable to
set them up to have as full a bandwidth as possible without input filtering
and when NFB is connected.
When the amp shows reasonable performance, then is the time to add approporiate
bandwidth limiting.
This certainly applies with hi-fi amps which are never used at full power,
so their response of 5 Hz to 100 kHz is commendable,
but PA amps may need to be tailored to between 20 Hz and 20 kHz.

Many tube amps with NFB and with slow CR coupling time constants will have a
bw from 3 Hz to 50 kHz, and thus effect of the amp upon the PS is easily
measured.



If the HF component above say 100 Hz is filtered out,
then an even better picture of the LF signal behaviour is gained.

A high pass filter eliminates energy below the cutoff freqeuncy of the
filter, not above the cutoff freqeuncy.

See above,



Also it is not clear that a pink noise signal would contain
much energy at the resonant frequency of the power supply filter.

The random noise at any F is reatively equal anywhere along the band.

I think you are confusing pink noise with white noise here.

It appears I am, but let me explain further....

White noise is defined on page 619 of RDH4.

I don't know if there is any reference to pink noise, its not even in the index.

But white noise is mentioned on page 619 in connection to testing of speakers.

Pink noise is white noise which has been filtered by a low pass filter with a 3
dB per octave slope,
so the same power will exist betwen say 100 Hz and 120 Hz as between 1,000 Hz
and 1,020 Hz.

It is explained at
http://www.cim.mcgill.ca/~clark/nord.../nm_noise.html.

I use a tunable filter with a constant Q to extract the signal from the DUT
using pink noise, and at anywhere along the band, the measured amplitude
using an RMS meter is constant using my filter and with the pink noise source,
which is equalised white noise.

With speakers that I know are flat, I get a flat response measurement.

There is more at
http://hyperphysics.phy-astr.gsu.edu...dio/equal.html



The problem with compression is that you are applying
gross 3rd harmonic distortion by flattening the range of the signal swing.

"gross 3rd harmonic distortion" is not a problem in a well designed
compressor. The control signal in a compressor is relatively slow acting
so that if the VGA is blameless distortion should be minimal.

If you squash a sine wave down, it then has an enormous amount of
odd order thd.
If the sine wave is linear up to 1/2 its amplitude, then has a rounded top,
then you get mainly 3H without al the other odd order harmonics.


I think you
are confusing compression with clipping which does introduce fairly
serious distortion, although I would say that inter modulation distortion
is a bigger problem than 3rd harmonic distortion. The clippers in modern
audio processors also have so called "distortion cancellation" circuitry
which is supposed to mitigate the problem.

No doubt there are all sort of processes used that I know little about,
and not sure I want to know, I am flat out maintaining
fidelity.

It is obviously possible to bias a variable mu tube tube off as the volume
level increases, thus reducing the gain of the amplifier, thus
preventing the overload of the tube.
But it all taints the fidelity a bit.
So a large LF bass signal will cause the gain to reduce, and the singer's
voice may seem to go quiet, and come back loud as the bass signal fades,
and so one F modulates another F.
This isn't fidelity, its nothing like a real world concert, where what you hear
is what you get.

Patrick Turner.




Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/