[Iain Churches]

In addition to the fact that no capacitor might be better
than any capacitor, what are the benefits of DC coupling
between the first stage and the phase splitter?

Iain

[Patrick Turner]

Iain Churches wrote:

In addition to the fact that no capacitor might be better
than any capacitor, what are the benefits of DC coupling
between the first stage and the phase splitter?

Iain

In a Williamson amp there is direct coupling between V1 and V2,
and this eliminates a time constant which would otherwise add to those
already in the amp
and lead to certain LF instability.

Practical issues like stablity were in the minds of designers in 1950.

Issues ( or ideas ) such as "no cap might be better than any cap" only
came
to prominence when ppl began to use bjts and a guy called Linn invented
the
fully direct coupled bjt amp using both pnp and npn devices.

Later on would be cognescenties writing in audiophile magazines
began talking about caps as though they were downright evil things to be
avoided
at all costs.
Hardly any of these dudes had ever built an amp.

But without caps, society would grind to a halt until a substitute
could be found.

In my own rather decent system i have maybe 6 RC couplings in the preamp
and power amp signal path, and I believe gross circuit modification
to remove them would lead to no subjectively detectable change, although
the change to the circuit to be rid of the caps may make a detectable
change.

I intend leaving the caps where they are.

Patrick Turner.

[Bob H.]

On Feb 20, 6:03 am, "Iain Churches" wrote:
In addition to the fact that no capacitor might be better
than any capacitor, what are the benefits of DC coupling
between the first stage and the phase splitter?

Iain


The only advantage I see is that a cap is eliminated. To some, this
is the reason it's done.

The problem is achieving a relative negative bias between grid and
cathode of the following stage.

If going into a concertina splitter (split load), a large cathode cap
is already used in the following stage, so you should be in the
ballpark for proper biasing range to begin with. Adjusting the
cathode and plate resistors together, or adding a bias resistor under
the cathode resistor with a cap bypass are methods to achieve relative
negative bias for the phase splitter. Another place to adjust is the
plate resistor and operating current of the preceeding stage.
The tricky part is keeping optimum loading of both stages while
adjusting for biasing of the following stage.

I'd make sure the tubes you're direct coupling are lower gain ones,
which afford some leeway with biasing to compensate for tube to tube
differences in plate resistance, age, etc. A 6sn7 can operate well
with a bias change in volts if set properly, while a 12ax7 usually has
millivolts of range before they might be positive biased.
A cursory look through some vintage schematics show 12ax7's being
direct coupled to the splitter, so I guess it was done.


Have fun
Bob H.

[Iain Churches]

"Patrick Turner" wrote in message
...


Iain Churches wrote:

In addition to the fact that no capacitor might be better
than any capacitor, what are the benefits of DC coupling
between the first stage and the phase splitter?

Iain

In a Williamson amp there is direct coupling between V1 and V2,
and this eliminates a time constant which would otherwise add to those
already in the amp
and lead to certain LF instability.

That's exactly the advantage I am looking for. The amp I am
building has a very respectable square wave at 1kHz

http://www.kolumbus.fi/iain.churches...kHz_8_Ohms.jpg

and a fairly respectable square wave at 5kHz.

http://www.kolumbus.fi/iain.churches...kHz_8_Ohms.jpg

but some phase shift at 100Hz, which I would like
to improve:

http://www.kolumbus.fi/iain.churches...0Hz_8_Ohms.jpg


Later on would be cognescenties writing in audiophile magazines
began talking about caps as though they were downright evil things to be
avoided at all costs.

Yes. I have seen such articles. I was thinking here about any benefits in
lower phase shift and improvement in stability.

In my own rather decent system i have maybe 6 RC couplings in the preamp
and power amp signal path, and I believe gross circuit modification
to remove them would lead to no subjectively detectable change, although
the change to the circuit to be rid of the caps may make a detectable
change.

In this case, I am thinking about one cap only, between the
first stage and the phase inverter.

Iain

[Iain Churches]

"Bob H." wrote in message
oups.com...
On Feb 20, 6:03 am, "Iain Churches" wrote:
In addition to the fact that no capacitor might be better
than any capacitor, what are the benefits of DC coupling
between the first stage and the phase splitter?

The only advantage I see is that a cap is eliminated. To some, this
is the reason it's done.

Hi Bob.

As mentioned in my reply to Patrick, I was wondering if it
might improve phase response.

The problem is achieving a relative negative bias between grid and
cathode of the following stage.

I can probably manage that:-))

If going into a concertina splitter (split load), a large cathode cap
is already used in the following stage, so you should be in the
ballpark for proper biasing range to begin with. Adjusting the
cathode and plate resistors together, or adding a bias resistor under
the cathode resistor with a cap bypass are methods to achieve relative
negative bias for the phase splitter. Another place to adjust is the
plate resistor and operating current of the preceeding stage.
The tricky part is keeping optimum loading of both stages while
adjusting for biasing of the following stage.

Yes I see. My phase splitter is common cathode.

I'd make sure the tubes you're direct coupling are lower gain ones,
which afford some leeway with biasing to compensate for tube to tube
differences in plate resistance, age, etc. A 6sn7 can operate well
with a bias change in volts if set properly, while a 12ax7 usually has
millivolts of range before they might be positive biased.
A cursory look through some vintage schematics show 12ax7's being
direct coupled to the splitter, so I guess it was done.

In this case the input stage is a 6SN7 with a 6SL7 as a phase splitter.


Have fun
Bob H.

You too:-)

Iain

[Patrick Turner]

Iain Churches wrote:

"Patrick Turner" wrote in message
...


Iain Churches wrote:

In addition to the fact that no capacitor might be better
than any capacitor, what are the benefits of DC coupling
between the first stage and the phase splitter?

Iain

In a Williamson amp there is direct coupling between V1 and V2,
and this eliminates a time constant which would otherwise add to those
already in the amp
and lead to certain LF instability.

That's exactly the advantage I am looking for. The amp I am
building has a very respectable square wave at 1kHz

http://www.kolumbus.fi/iain.churches...kHz_8_Ohms.jpg

and a fairly respectable square wave at 5kHz.

http://www.kolumbus.fi/iain.churches...kHz_8_Ohms.jpg

but some phase shift at 100Hz, which I would like
to improve:

http://www.kolumbus.fi/iain.churches...0Hz_8_Ohms.jpg

Wow, never did a copy of what you are seeing on a 100mm CRO screen
appear SO BIG on my monitor...


The CR coupling has no effects on the HF square wave traces.

And the direct coupling of v1-v2 won't make much difference to the phase
shift at 100Hz.

After NFB is applied, all will be well.

And if you take a look at the LF shelving networks I use then you will
see I deliberately
*increase* the phase shift a bit between 20Hz and 100Hz compared to
using
plain CR coupling with one low time constant.
but the FB takes care of it all and the shelving network makes the
stability excellent
and recovery and the bass sounds really well.

Later on would be cognescenties writing in audiophile magazines
began talking about caps as though they were downright evil things to be
avoided at all costs.

Yes. I have seen such articles. I was thinking here about any benefits in
lower phase shift and improvement in stability.

In my own rather decent system i have maybe 6 RC couplings in the preamp
and power amp signal path, and I believe gross circuit modification
to remove them would lead to no subjectively detectable change, although
the change to the circuit to be rid of the caps may make a detectable
change.

In this case, I am thinking about one cap only, between the
first stage and the phase inverter.

Some Williamsons I have seen have RC coupling V1 -V2 instead of the
direct couple.
These often oscillate badly at LF without any load because open gain
becomes high
and the there is no margin of stability.
Recovery after overload causes yo-yo of the output as the amp tries
desperately to reset
its dc voltage levels.
Motorboating in such arrangements can also be a problem,
especially with a preamp whose PS is dependant on the power amp,
and there is a tone control with bass boost.
Turn up the bass and it begins to oscillate.....

There was a lotta garbage produced in the 1950s...

Patrick Turner


Iain

[John Byrns]

In article ,
Patrick Turner wrote:

Wow, never did a copy of what you are seeing on a 100mm CRO screen
appear SO BIG on my monitor...

Agreed, linear dimensions are about 2X too large!


Regards,

John Byrns

--
Surf my web pages at, http://fmamradios.com/

[Iain Churches]

"John Byrns" wrote in message
...
In article ,
Patrick Turner wrote:

Wow, never did a copy of what you are seeing on a 100mm CRO screen
appear SO BIG on my monitor...

Agreed, linear dimensions are about 2X too large!


Fits my 17 and 19 inch monitors perfectly.

Cheers
Iain

[Bob H.]

It just adds stabilitiy to reduce the chance of oscillation when using
negative feedback around the output transformer.

[Iain Churches]

"John Byrns" wrote in message
...
In article ,
Patrick Turner wrote:

Wow, never did a copy of what you are seeing on a 100mm CRO screen
appear SO BIG on my monitor...

Agreed, linear dimensions are about 2X too large!


Regards,

John Byrns

Sorry about the size of the CRO pics. I checked them carefully
on my own PC and they fitted comnfortably into the frame. It
may be that my ISP somehow resizes pics, as I never get problems
with having to scroll.

New smaller images (reduced 50%) are at:

http://www.kolumbus.fi/iain.churches/Pics/1kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/5kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg

Please let me know if these are OK. It will be useful to
know for the future.

Thanks
Iain

[Iain Churches]

"Bob H." wrote in message
ups.com...

It just adds stabilitiy to reduce the chance of oscillation when using
negative feedback around the output transformer.

So added stability seems like a very valid reason:-))

I asked the question as I wondered if DC coupling
might improve the LF phase response shown at:

http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg

Regards
Iain

[Iain Churches]

"Patrick Turner" wrote in message
...

Wow, never did a copy of what you are seeing on a 100mm CRO screen
appear SO BIG on my monitor...

Soirry about that:-) I checked them on my own monitor
and the piucs fitted the screen perfectly without scrolling.
At Joihn'äs suggestioin I have reduced by 50%.

New pics at:

http://www.kolumbus.fi/iain.churches/Pics/1kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/5kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg


The CR coupling has no effects on the HF square wave traces.

And the direct coupling of v1-v2 won't make much difference to the phase
shift at 100Hz.

OK. That was my may concern. I was looking for a way to get
some improvement here. I don't have the figs with me, but IIRC
the phase shift was about +10 degs at 20Hz.

After NFB is applied, all will be well.

Did you think I could get such good results open loop? :-)
I already have 12dB NFB applied, with a stability margin of
15dB. This gives me (very conveniently) an input sensitivity
of 0dBV.


There was a lotta garbage produced in the 1950s...

There still is. Look East, young man, Look East:-)

Iain

[robert casey]

Iain Churches wrote:
In addition to the fact that no capacitor might be better
than any capacitor, what are the benefits of DC coupling
between the first stage and the phase splitter?


One issue with DC coupling that caps help you avoid is the shifts and
drifts of currents and voltages due to component aging and heat. A
small and usually insignificant drift near the input would be magnified
(by the gain of the amp) at the output stages. Which could cause bad
biasing of the output tubes which could run them too hot. Human ears
don't hear DC biases (unless you have a cold that clogs the internal
workings of the ear and you can't equalize air pressure on the eardrum)
so you don't need to reproduce DC in your audio amp.

[Iain Churches]

"robert casey" wrote in message
ink.net...
Iain Churches wrote:
In addition to the fact that no capacitor might be better
than any capacitor, what are the benefits of DC coupling
between the first stage and the phase splitter?


One issue with DC coupling that caps help you avoid is the shifts and
drifts of currents and voltages due to component aging and heat. A small
and usually insignificant drift near the input would be magnified (by the
gain of the amp) at the output stages. Which could cause bad biasing of
the output tubes which could run them too hot. Human ears don't hear DC
biases (unless you have a cold that clogs the internal workings of the ear
and you can't equalize air pressure on the eardrum) so you don't need to
reproduce DC in your audio amp.

Hi Robert. Thanks for your input. My question was asked as the result
of looking at some square waves of my amplifier's performance. Sqr
Wave testing is something fairly new to me, and I need to be able to
interpret what I see before I can know if I have a problem needing a
solution.

Iain

[John Byrns]

In article ,
"Iain Churches" wrote:

"John Byrns" wrote in message
...
In article ,
Patrick Turner wrote:

Wow, never did a copy of what you are seeing on a 100mm CRO screen
appear SO BIG on my monitor...

Agreed, linear dimensions are about 2X too large!


Regards,

John Byrns

Sorry about the size of the CRO pics. I checked them carefully
on my own PC and they fitted comnfortably into the frame. It
may be that my ISP somehow resizes pics, as I never get problems
with having to scroll.

New smaller images (reduced 50%) are at:

http://www.kolumbus.fi/iain.churches/Pics/1kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/5kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg

Please let me know if these are OK. It will be useful to
know for the future.

Those are just about right for me. The problem for me with the
schematic is that it is too wide, yet not real high.


Regards,

John Byrns

--
Surf my web pages at, http://fmamradios.com/

[Phil Allison]

"Iain Churches"


I asked the question as I wondered if DC coupling
might improve the LF phase response shown at:

http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg



** It don't need fixing.

Good audio power amps * HAVE * 100Hz square wave responses like that -
because good audio power amps need to have sub sonic roll off included.



........ Phil

[Iain Churches]

"Phil Allison" wrote in message
...

"Iain Churches"


I asked the question as I wondered if DC coupling
might improve the LF phase response shown at:

http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg



** It don't need fixing.

Good audio power amps * HAVE * 100Hz square wave responses like that -
because good audio power amps need to have sub sonic roll off included.


Hi Phil,

My interpretation of the 100Hz square wave, by comparison with
pics in text books, is that it does not exhibit sub sonic roll-off (which
would be shown by a trace with sides of equal height and a concave
top bar) but phase shift. IIRC it's about 10degrees, so maybe you are
correct - "It don't need fixing"

Iain

[Patrick Turner]

"Bob H." wrote:

It just adds stabilitiy to reduce the chance of oscillation when using
negative feedback around the output transformer.

DC coupling can add to the margin of stabliity.

A Williamson has cap coupling between phase inverter output and the
balanced drive amp
and from this to the output tubes, and then the Lp of the OPT shunts the
anode circuit.
So you have C-R, C-R, R-L, or 3 time constants and the -3dB pole for
each
is at around 3Hz or less.

This means that at some F below 3Hz, there is 60 degrees of phase lead
and a total open loop
phase shift of 180 degrees and if open loop gain is above 1.0 at this
point, it will oscillate,
and many Williamsons do just that.

The reason why old DTN said you need at least 100H of Lp in the OPT was
because of phase shift
and stability.
Many ppl since 1947 have built OPT with nothing like 100H of Lp at 5Vrms
a-a
at 50Hz, or LF, as DTN suggested must be possible for stability.
Some sa,ples of Williamsons oscillate at a low level where Lp 100H and
as the
level of oscillation signal builds, so does the amount of Lp,
so the amplitude of the oscillations are limited from getting any
larger.
This often occurs without an output load, and indicates borderline
stability margins.

The answer to the problem of such instabilities is to eliminate the
TWO consecutive stages of pure C-R couplings, and use a shelving filter
which places the ultimate LF pole well below 1Hz or wherever it may be;
a 0.47uF plus 220k
R give a pole at 1.5Hz.
If you have a 1M plus 0.033 uF parallel network between the 0.47 and
220k,
there is an additional pole created for where Z 0.033uF = 220k, which is
at 22Hz,
and then there is a shelf created at -14.8dB, and a new pole at 0.28Hz.
The phase lead i such a shelved filter/coupling arrangement is always
never exceeding about 45 degrees until 0.28Hz when the gain of the
circuit has been reduced 14.8dB.

Many LF shelving network examples exist in schematics of tube amps at my
website.

I suggest anyone uncertain as to the effectiveness of such techniques in
NFB amps
should PLOT THE RESPONSE of the open loop character of any amp they are
struggling to build,
with/without the shelving networks for LF ( and HF )

beacuse of phase shift, it is impossible to expect stability when gain
is above 1.0 and phase shift exceeds
180d. The shelving networks reduce both gain and the phaseshift, and at
2Hz and say 50kHz
where an amp may otherwise oscillate, the shelving networks will
increase the stability
by maybe 10dB margin, or enough to allow the amp to be unconditionally
stable.
Meanwhile, at all F between 20Hz and 20kHz, the phase shift with NFB
is reduced a lot from the amount without any FB, to maybe less than 10
degrees, and negligible.

All amplifiers are in effect bandpass filters.

The open loop gain response is an arched line with 0 degrees phase shift
somewhere
in the top of the arch, and each side of the arch indicates phase shift,
and also decline in open loop
gain.

The rules of stability are well explained in RDH4, although very poorly
understood,
and those fellas Nyquist and Bode didn't make it much easier for lay
people without much idea of mathematical discipline to
understand NFB. Luckily, one needn't have heard the name of either guy
to be able to build a beautiful
amp with 20dB of global NFB, one just has to accept the realities of the
phase shift,
and know how to set up compensation networks to prevent both excessive
gain AND phase shift
to produce an oscillator rather than amplifier when you don't want an
oscillator.

I do not know a living soul who can plot the open loop gain and phase
response without GNFB of any amp,
then tell me confidently what compensation networks I MUST use, and what
all the values of R & C must be.

Even RDH4 says that with the vaguraries of iron cored inductors and the
parasitic
leakage L and shunt C, stablisiation is best achioeved on a cut and try
method.

The method only works if you KNOW what to look for, and KNOW how to test
an amp.

Every diy person who insists on never knowing either will sooner
or later be forced to take his creation to someone who does.
The trouble with many diyers, they get to age 50, and make ONE amp
system, and never make another,
and we never hear again from them because their mission is over.

Quad-II got around the problem of having a low amount of Lp
and large amount of LL and poor PS filter and coupling time constants
by having TWO lots of NFB, one in the output stage, and the other
with a SMALL amount of GNFB, and only 1 C&R coupling between input tubes
and output tubes.
If anyone tries increasing the global FB in a Quad-II amp, by say only
another 8dB,
the darn amp will show instability problems.

Leak Also had stability problems with both triode and UL versions
of the TL12 Pt1 amps going unstable at LF due to too many C&R couplings.
Leak didn't see the merit of having the input pentode directly connected
to the following
LTP, and used CR coupling, and to know how to fix the attrocious Leak
behaviour,
see my website for the revised circuit with adequate compensation
networks.

The later Mullard 520 circuit with its V1 EF86 direct coupled to the
following 12AX7 LTP
adressed the problem of LF instability.

Leak must have thought it too unreliable to have direct coupling,
because of the dcV drift
of the V1 anode voltage, and the tendency for V2 to become overloaded
with grid current.

The problem often with direct coupling between V1 and V2 triodes is the
bias condition for the V2 phase inverter.
if you want +100V for the grid at V2 inverter, and you have a 22k Rk,
then you'd have about 4mA
for Ik, way too much to expect from a 1/2 6SL7, or 12AX7, but OK for
12AU7 and 6CG7 or 6SN7.
Then you want 22k for the anode R, and there must be the right anode to
cathode DC voltage,
based on basic optimum operations of the tube with a 44k load if it was
a gain tube.
Load line analysis is essential to find out where one wants the B+, and
how negative one wants
the grid to be with respect to the cathode so that the voltage drive
applied from V1
to V2 grid never produces grid I by exceeding the bias voltage.
Usually the B+ supply for the phase inverter will need to be higher than
that for the V1 gain tube
but ppl try to keep them the same, and the damn phase inverter still
works a bit
but its plagued with grid current if the RLa and RLk are too low a
value, so much
working out and trials must be done to get the concertina phase inverter
to work
entirely blame free, ie, be able to produce 20Vrms at least from both a
and k without grid I
smudging up the music signal.

Such concerns are paramount in amps such as Dynaco ST70 where the phase
inverter must produce
the full drive voltage to the output tube grids, so blame free output
should be up
to 50Vrms at a and k if you have KT88 in a Mk-III.


Every tube used in any gain line up should be set up so its loading and
bias operation
is optimal, regardless of whether it is in a phase inverter circuit or
not,
and so the stages could each produce much more clean voltage than
actually needed.

Patrick Turner.

[Patrick Turner]

Iain Churches wrote:

"John Byrns" wrote in message
...
In article ,
Patrick Turner wrote:

Wow, never did a copy of what you are seeing on a 100mm CRO screen
appear SO BIG on my monitor...

Agreed, linear dimensions are about 2X too large!


Regards,

John Byrns

Sorry about the size of the CRO pics. I checked them carefully
on my own PC and they fitted comnfortably into the frame. It
may be that my ISP somehow resizes pics, as I never get problems
with having to scroll.

New smaller images (reduced 50%) are at:

http://www.kolumbus.fi/iain.churches/Pics/1kHz.jpg

This one fills 3/4 of the width and full height of the 17" monitor of
mine.
The green trace is 3mm wide.

Size could be 50% less, and we'd all still see what you are seeing on
your CRO.

What's the trace look like when you have nothing but 0.22uF at across
the output terminals,
and a low level 5kHz square wave with the full amount of GNFB applied?

Patrick Turner.


http://www.kolumbus.fi/iain.churches/Pics/5kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg

Please let me know if these are OK. It will be useful to
know for the future.

Thanks
Iain

[Patrick Turner]

Iain Churches wrote:

"Patrick Turner" wrote in message
...

Wow, never did a copy of what you are seeing on a 100mm CRO screen
appear SO BIG on my monitor...

Soirry about that:-) I checked them on my own monitor
and the piucs fitted the screen perfectly without scrolling.
At Joihn'äs suggestioin I have reduced by 50%.

New pics at:

http://www.kolumbus.fi/iain.churches/Pics/1kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/5kHz.jpg
http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg


The CR coupling has no effects on the HF square wave traces.

And the direct coupling of v1-v2 won't make much difference to the phase
shift at 100Hz.

OK. That was my may concern. I was looking for a way to get
some improvement here. I don't have the figs with me, but IIRC
the phase shift was about +10 degs at 20Hz.

After NFB is applied, all will be well.

Did you think I could get such good results open loop? :-)
I already have 12dB NFB applied, with a stability margin of
15dB. This gives me (very conveniently) an input sensitivity
of 0dBV.


There was a lotta garbage produced in the 1950s...

There still is. Look East, young man, Look East:-)

I look East, and I am aghast.

The asians seem to have a very poor understanding of gain/phase shift/FB
stability issues.

But then the brightest young men of the East couldn't give a **** about
silly old amps that grandpa likes; they are busily engaged in producing
rockets that can shoot down US satellites,
becoming lawyers, doctors, and getting rich fast and dodging the BS of
the Chinese Communist Party.

Tube amp building in the East attracts the not so very bright guys......


Patrick Turner.



Iain

[Phil Allison]

"Iain Churches Total IDIOT "


I asked the question as I wondered if DC coupling
might improve the LF phase response shown at:

http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg



** It don't need fixing.

Good audio power amps * HAVE * 100Hz square wave responses like
hat -
because good audio power amps need to have sub sonic roll off included.



My interpretation of the 100Hz square wave, by comparison with
pics in text books, is that it does not exhibit sub sonic roll-off (which
would be shown by a trace with sides of equal height and a concave
top bar) but phase shift.


** The square wave is affected by phase shift ( ie is tilted) which is *
CAUSED * by a sub sonic roll-off.

Any stupid book that says otherwise needs to be burned !!!

A "concave " square wave shows a response dip or notch exists at the
frequency.

Bet you misread it, as always.



IIRC it's about 10degrees, so maybe you are
correct - "It don't need fixing"


** The phase shift induced tilt you are seeing is that which exists *
between* the fundamental and the harmonics - mainly the 3rd and 5th.

For a square wave to look square - the harmonic components must be in an
exact phase relationship.



........ Phil

[Patrick Turner]

Phil Allison wrote:

"Iain Churches Total IDIOT "


I asked the question as I wondered if DC coupling
might improve the LF phase response shown at:

http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg



** It don't need fixing.

Good audio power amps * HAVE * 100Hz square wave responses like
hat -
because good audio power amps need to have sub sonic roll off included.



My interpretation of the 100Hz square wave, by comparison with
pics in text books, is that it does not exhibit sub sonic roll-off (which
would be shown by a trace with sides of equal height and a concave
top bar) but phase shift.

** The square wave is affected by phase shift ( ie is tilted) which is *
CAUSED * by a sub sonic roll-off.

Any stupid book that says otherwise needs to be burned !!!

A "concave " square wave shows a response dip or notch exists at the
frequency.

Bet you misread it, as always.

IIRC it's about 10degrees, so maybe you are
correct - "It don't need fixing"

** The phase shift induced tilt you are seeing is that which exists *
between* the fundamental and the harmonics - mainly the 3rd and 5th.

For a square wave to look square - the harmonic components must be in an
exact phase relationship.

....... Phil

The curve downwards of the "horizontals" of the square wave is the
action
of the time constant, C being discharged after the transient uprights.
But you're right with the harmonic explanations, and the meaning of a
concave line.
Could be convex, indicating a peak someplace.

Patrick Turner.

[Iain Churches]

"Patrick Turner" wrote in message
...

New smaller images (reduced 50%) are at:

http://www.kolumbus.fi/iain.churches/Pics/1kHz.jpg

This one fills 3/4 of the width and full height of the 17" monitor of
mine.
The green trace is 3mm wide.

Size could be 50% less, and we'd all still see what you are seeing on
your CRO.

OK. I have reduced them again by 50% making 25% of the original
size. If this is OK I shall bear this in mind for the future. So thanks
for
your comments.

What's the trace look like when you have nothing but 0.22uF at across
the output terminals,
and a low level 5kHz square wave with the full amount of GNFB applied?

:-(((( I was hoping you would not ask at this juncture.
The amp is stable into open circuit, but goes totally bananas (HF
oscillation)
with a 0.22µF connected.

Regards to all
Iain

[Iain Churches]

"Phil Allison" wrote in message
...

"Iain Churches Total IDIOT "


I asked the question as I wondered if DC coupling
might improve the LF phase response shown at:

http://www.kolumbus.fi/iain.churches/Pics/100Hz.jpg



** It don't need fixing.

Good audio power amps * HAVE * 100Hz square wave responses like
at -
because good audio power amps need to have sub sonic roll off included.



My interpretation of the 100Hz square wave, by comparison with
pics in text books, is that it does not exhibit sub sonic roll-off (which
would be shown by a trace with sides of equal height and a concave
top bar) but phase shift.


** The square wave is affected by phase shift ( ie is tilted) which is *
CAUSED * by a sub sonic roll-off.

Any stupid book that says otherwise needs to be burned !!!

A "concave " square wave shows a response dip or notch exists at the
frequency.

Bet you misread it, as always.

Hi Phil. This has been a civilised thread up to now. I hope we
can keep it that way. You are an amp technician so things that are
obvious to you are not so obvious to someone like me.

At the same time, as a classical recording engineer, there are
things that I do on a daily basis which would probably be well
beyond your field of expertise.


** The phase shift induced tilt you are seeing is that which exists *
between* the fundamental and the harmonics - mainly the 3rd and 5th.

For a square wave to look square - the harmonic components must be in an
exact phase relationship.

OK. Thanks for that.
Iain



....... Phil

[Phil Allison]

"Iain Churches Total IDIOT "


** It don't need fixing.

Good audio power amps * HAVE * 100Hz square wave responses like
t -
because good audio power amps need to have sub sonic roll off included.


My interpretation of the 100Hz square wave, by comparison with
pics in text books, is that it does not exhibit sub sonic roll-off
(which
would be shown by a trace with sides of equal height and a concave
top bar) but phase shift.


** The square wave is affected by phase shift ( ie is tilted) which is
* CAUSED * by a sub sonic roll-off.

Any stupid book that says otherwise needs to be burned !!!

A "concave " square wave shows a response dip or notch exists at the
frequency.

Bet you misread it, as always.

Hi Phil.


** Never " Hi Phil " me - **** head.


This has been a civilised thread up to now.


** ********.

Never the case with a slimy, pommy crim & sub human **** called Churches
involved.



** The phase shift induced tilt you are seeing is that which exists *
between* the fundamental and the harmonics - mainly the 3rd and 5th.

For a square wave to look square - the harmonic components must be in an
exact phase relationship.

OK. Thanks for that.


** So the dumb as dog ****, ASD ****ed **** misread the book.

Just will never admit it.




......... Phil

[Phil Allison]

"Patrick Turner"


** The square wave is affected by phase shift ( ie is tilted) which is
*
CAUSED * by a sub sonic roll-off.

Any stupid book that says otherwise needs to be burned !!!

A "concave " square wave shows a response dip or notch exists at the
frequency.

Bet you misread it, as always.


IIRC it's about 10degrees, so maybe you are
correct - "It don't need fixing"

** The phase shift induced tilt you are seeing is that which exists *
between* the fundamental and the harmonics - mainly the 3rd and 5th.

For a square wave to look square - the harmonic components must be in an
exact phase relationship.



The curve downwards of the "horizontals" of the square wave is the
action of the time constant, C being discharged after the transient
uprights.


** The straight lines of the wave would become curved in that case, not
merely tilted - as they are at much lower frequencies.

It is just down to relative phase shift in the harmonics, despite it being
only a few degrees.


But you're right with the harmonic explanations, and the meaning of a
concave line.


** Been staring at what various amps do to audio frequency square waves for
over 40 years - it does get kinda ground into the brain.

Someone ought do a web site on the CORRECT analysis of square waves shapes
as seen on a CRO with various phase and frequency anomalies applied.




........ Phil

[Patrick Turner]

Iain Churches wrote:

"Patrick Turner" wrote in message
...

New smaller images (reduced 50%) are at:

http://www.kolumbus.fi/iain.churches/Pics/1kHz.jpg

This one fills 3/4 of the width and full height of the 17" monitor of
mine.
The green trace is 3mm wide.

Size could be 50% less, and we'd all still see what you are seeing on
your CRO.

OK. I have reduced them again by 50% making 25% of the original
size. If this is OK I shall bear this in mind for the future. So thanks
for
your comments.

What's the trace look like when you have nothing but 0.22uF at across
the output terminals,
and a low level 5kHz square wave with the full amount of GNFB applied?

:-(((( I was hoping you would not ask at this juncture.
The amp is stable into open circuit, but goes totally bananas (HF
oscillation)
with a 0.22µF connected.

Under exactly what schematic conditions were the pics of the waveforms
recorded?
Pure open loop with no shelving networks and maximum LF extensions?

With/without NFB?

High level/low level?

Each time a wave form is shown, a full specification of the test should
be given for the waveform to have any scientific or useful relevance to
ppl
capable of making good amplifiers without outside assistances, ie, they
know enough.

So why does the amp oscillate with a 0.22uF?

What happens with 0.047, 0.1, 0.47, 1.0 , 2.0 and 4.7 uF ?

What happens with 1mH?

What is the open loop sine wave response at low levels especially at
between 32khz and 320kHz?
And what about the power bandwidth open loop where the sine wave
response is measured
at where thd just under 1%?

What effect is there with HF stability during tests with FB with when
running a 10Hz signal at just above the onset of
OPT saturation?
Are there any bursts of RF during LF staturation?

What does the CRO trace do when you change from using a 1khz signal at
clipping suddenly
to -20dB? Does the trace wobble up and down before settling?

Can you provoke the amp into slow oscillations by using bursts of 1khz
signals that go
from no signal to about 10dB input voltage over load and off again?

What happens when you change the burst F to 100Hz, 10kHz?

With the rated R load at the output, what is the error signal sine wave
response
like at the output of V1 between 1Hz and 1MHz?

At what F does the OPT exhibit series or parallel resonances?

Does the total plate current to the output stage at -20db signal, R
load, remain constant
between 1Hz and 1MHz?

What is the output resistance with NFB and without?

What is the Rout at 40Hz and 20kHz, with/without GNFB?

At the 1kHz clipping level, what is the increase in plate current?
What about at 40Hz and 20kHz?

What does the 5 kHz square wave signal look like at each anode of the
output stage
with NFB? without NFB?

With GNFB, is there only very slight ringing with a 5kHz square wave at
the output, ie, less than 3dB overshoot,
maybe 2 cycles of ringing F,
but lots more ringing at each output stage anode?

What is the output R load required to be for pure class A operation up
to clipping?

What is the graph profile of the power output vs RL at 2% thd, (
clipping )
for 1kHz, for all R loads of 1,2,3,4,5,6,8,12,16,24,32,64 ohms?

What does this tell you about load matching of the amp?

You have a 2uF, and connect 1.5ohms in series with it, and have 15 ohms
across this R+C
in parallel. Is this an approximate minimum model load for a well known
ESL?

You try such a load on your amp. What is the sine wave response at low
level to 100kHz?
What about at -6dB? What is the maximum power response curve with such a
load
between 1Hz and 1MHz?

Assuming you are able to tame the gross oscillations with a pure 0.22uF
load,
and you think you have optimum critical damping, so no cap causes
oscillations, what is the maximum
power bandwidth with thd not exceeding 2%, and with the rated load R?
with 3 times R? with R/2?

With NFB connected, With various values of pure C connected as loads,
what is the low level sine wave response for
all C chosen? does any value of C load cause the sine wave response to
rise more than
10dB above the 1 kHz rated R load output level?

What is the amount of boost to the HF below 20kHz due to having C loads
connected when GNFB is connected at low levels?

What value of C is required to cause the 20kHz sine wave at low levels
to increase by 3 dB?

What is the sine wave low level response with and without GNFB when very
carefully measuring between 0.5Hz and 100Hz
so that your sig gene sag at LF is fully compensated for, and you have
no input capacitor to tailor
a LF pole external to the amp signal path?

What is the phase shift plot for the amp at low levels, rated R load?
With open loop, without shelving networks?
With open loop, with shelving networks?
With NFB for both situations?
Note, oscillation may prevent the phase response measuring with GNFB and
no shelving networks.

How much do you rely on the cap shunting the reistor feeding back the
GNFB?
What is the effect of too high a C value here?
To small a value?
How do you determine the value of this important C?

Do you need Zobel networks acoss each 1/2 primary of the OPT?
Across any other windings or part therof?
Why?



I could maybe think of another 10 questions, but that will do for now.

Only YOU can answer them, and I suggest you prepare some exercize book
graph paper drawn up using the
8mm between horizontal lines to show 3dB steps in response levels, and
then
divide the page vertically into columns for F beginning at 5Hz to 10Hz,
then 100Hz, 1kHz, 10kHz, 100kHz, and 1MHz.

Try drawing 12 even spaces between say 10Hz and 100Hz.
This way you can nominate each vertical line at
10, 12, 15, 18, 22, 27, 33, 39, 47, 56, 68, 82 and 100 for the decade of
F, and you have a nice logarithmic
page notated, even though line spacing are equal. Interpolation
will allow postioning of even number frequencies.

Notice that the scal is the same as standard resistor values.

It is derived by simply starting with 1.0, then multiplying
by 1.2115 as you move up in value. 1.2115 to the power of 12 = 10.0
approx.

With a log page, the distance between 1.0 and 2.0 = that between 2.0 and
4.0
and between 4.0 and 8.0. Distances between octaves are equal.

Thus slopes of attenuation for RC networks have the same slope for the
same order
attenuation anywhere along the graph.

You could spend all day plotting with a PC unless you have a program
which does it automatically.
Then when you have a response saved, you click "print".
Or perhaps you post the responses someplace for the group to inspect and
comment.

Do you know where I can download a FREE program which will plot a
response of anything?
I have a couple of old spare PCs, two with W98se, so I could set one up
where its dusty and dirty, in my work shed, and proceed more
modernisticly.
Do I need a sheet of plastic over the PC?
I'll answer for you, YES.

Plotting by hand with a PC may be way too slow. My way is quick for a
"hand job",
so you give yourself more time for more questions, more answers,
and building a better amp..

If you had a pink noise source of even noise spread, and a suitable
selective filter,
this may also be used for response measurements, and more.
Pink noise is like music, very like heavy metal music at least,
except that musical tones have some harmonicly related waves, pink noise
has randomly varying
F, phase and amplitude.

HOWEVER, if you could find a signal sweep that puts out a slowly
changing F sine wave between
1Hz and 1MHz, ( or split into a number of CALIBRATED bands ), you might
be able to plot
amplitude in dB against F very much faster than my method using pencil
and paper in a book
if you had an old response plot grapher machine. They used to use a roll
of paper with a pencil/pen driven
with a dc motor. Maybe there are a few around left over from 1960.
I don't fancy making one up myself, and the PC option looks much easier,
and it would be more accurate..

Patrick Turner.



Regards to all
Iain

[Patrick Turner]

Phil Allison wrote:

"Patrick Turner"


** The square wave is affected by phase shift ( ie is tilted) which is
*
CAUSED * by a sub sonic roll-off.

Any stupid book that says otherwise needs to be burned !!!

A "concave " square wave shows a response dip or notch exists at the
frequency.

Bet you misread it, as always.


IIRC it's about 10degrees, so maybe you are
correct - "It don't need fixing"

** The phase shift induced tilt you are seeing is that which exists *
between* the fundamental and the harmonics - mainly the 3rd and 5th.

For a square wave to look square - the harmonic components must be in an
exact phase relationship.



The curve downwards of the "horizontals" of the square wave is the
action of the time constant, C being discharged after the transient
uprights.

** The straight lines of the wave would become curved in that case, not
merely tilted - as they are at much lower frequencies.

well, in fact the sloped lines of the wave at 100Hz are the top fairly
straight bit of the
curved slopes of time constant discharge.
If the waves were 10Hz square waves, you'd see the defineite curve
better;
1hz square waves would show what looked like pulses....

It is just down to relative phase shift in the harmonics, despite it being
only a few degrees.

But you're right with the harmonic explanations, and the meaning of a
concave line.

** Been staring at what various amps do to audio frequency square waves for
over 40 years - it does get kinda ground into the brain.

Someone ought do a web site on the CORRECT analysis of square waves shapes
as seen on a CRO with various phase and frequency anomalies applied.

....... Phil

There is already some info on the Web about
'amplifier square wave phase shift effects' according to the Google
search.

Try

http://users.tpg.com.au/ldbutler/Waveforms.htm


(( A related topic, intermodulation distortion, see

http://users.tpg.com.au/ldbutler/Intermodulation.htm ))

I went 6 pages deep in the google search, and there didn't seem to
obviously be much else
on phase shift of square wave harmonics.

I dunno how CORRECT you want the analysis for interpreting square wave
shapes.

But most tube amps have a generic predictable response at LF at least,
ie, a plain roll off
even in open loop unless there is a shelving network.

The LF response of the ordinary amp may have an ultimate damped 3rd
order response due to
say 3 cascaded but separated, or effectively buffered stages in the amp.

But let's be simple, and consider the amp equal to a simple C and R
filter at LF.

The harmonics within the square wave are phase shifted by dissimilar
amounts, with the
higher F not much shifted, but lowest F quite a lot, and how much
can be simply calculated.
A 100Hz sq.wave has 100Hz as its lowest sine wave within, and if the
pole was at 100Hz,
there would be -3db attenuation and 45d of phase lead.

But for the 5H within the sq.wave, phase shift would be a lot less, a
few degrees only.

Since many harmonics are present in a definite proportion to give a nice
looking square wave,
we tend to forget many H are present, but a high Q filter will show they
are there alright.


FB around an active filter with more than 12dB slope will produce over
shooting on square waves,
as will a plain passive C & L 12 dB/octave filter with moderate damping
R.

In a tube amp with so many C, L and R values all through it, we do in
effect have an active bandpass
filter and with usually a slope of attenuation that exceeds 10dB/octave
somewhere at least,
and hence some over shoot on square waves is impossible to avoid,
certianly so
when we add a pure C across the output which can add another order of
attenuation
at HF. So the best we can do is to damp the amp with R critiocally at HF
to prevent runaway
ultrasonic or RF oscillations. Its very easy to achieve this when you
know the simple tricks,
and a very lousy amp can be tamed. The worse the OPT, the
less FB you can use and still get good bandwidth. Usually its mainly the
OPT that determines final stablity and bandwidth,
and the amount of NFB and it isn't due to whether or not you have one
less C&R coupling out of 3 possible sets.

Most OPT are resonant at one or more F above the audio spectrum. Some
real crook not well interleaved OPT become
resonant even at 10kHz, but you don't see many this bad;
even the well interleaved OPT that Williamson designed 60 years back had
resonances at above 50kHz.
Resonances are unavoidable in OPTs. Just what the F is, and whether its
series or parallel is
often practically impossible to determine in the design stage, so we
just stick
to empirical methods to get a decent OPT
with MINIMIZED LEAKAGE INDUCTANCES AND CAPACITANCES.
I have found that with E&I trannies, and using 5P x 6S winding
interleaving, and
keeping the P&S windings well apart, not tightly wound close,
15Hz to 270kHz BW is EASILY OBTAINABLE AT LESS THAN 5% WINDING LOSSES.
Final response of an amp with such a an OPT is often better than many SS
amps.

Sure, such a decent tranny is twice the weight that a brandname maker
would prepare, but who cares about that crowd!
they make amps for a profit, and don't much worry about much else,
except that it
conform to lowest common denominator quality. Thus we have Quad-II,
could have been much better, but certainly
good enough for most PPL.

I hear Quad have released a celebratory issue of the Quad-II amps again;
I wonder if there are any
improvements to the trannies included? I assume the Chinese makers are
capable fellows....
( I do question their willingness though, when less effort = more
profits ).

Meanwhile, OPTs from Plitron, Amplimo?, Sowter, and other specialist
makers
do tend to make it easier to use NFB, use less shelving networks and
compensatory measures,
so NFB's job is easier, more effective, and its use
less blameable by the Bureau of Anti Negative Feedback Fetishistas.

Patrick Turner.

[Phil Allison]

"Patrick Turner"

Someone ought do a web site on the CORRECT analysis of square waves
shapes
as seen on a CRO with various phase and frequency anomalies applied.


There is already some info on the Web about
'amplifier square wave phase shift effects' according to the Google
search.

http://users.tpg.com.au/ldbutler/Waveforms.htm


** LOL !!!

From some Yank ham radio ****** - so full of stupid errors.

Looks like he used the same erroneous old source as Churches did.

" The evil that men do ......... "



http://users.tpg.com.au/ldbutler/Intermodulation.htm ))


** All RF theory based - little relevance to audio.


I dunno how CORRECT you want the analysis for interpreting square wave
shapes.


** Simply not completely wrong or hopelessly ambiguous will do.

I did a Goolge search long ago and found nothing of value.

Now that I have a small digital camera ......




....... Phil

[John Byrns]

In article ,
"Phil Allison" wrote:

"Patrick Turner"

The curve downwards of the "horizontals" of the square wave is the
action of the time constant, C being discharged after the transient
uprights.

** The straight lines of the wave would become curved in that case, not
merely tilted - as they are at much lower frequencies.

It is just down to relative phase shift in the harmonics, despite it being
only a few degrees.

Hi Phil,

The straight lines are curved, take a closer look the curvature is
easily seen in the photo. Because the effective time constant is
relatively long compared to the test frequency the curvature is just not
extreme.


Regards,

John Byrns

--
Surf my web pages at, http://fmamradios.com/

[John Byrns]

In article ,
"Phil Allison" wrote:

"Patrick Turner"

http://users.tpg.com.au/ldbutler/Intermodulation.htm ))


** All RF theory based - little relevance to audio.

Hi Phil,

I've got news for you, "RF" and "audio" electronics follow the same
theory, it's only in practice, where simplifying assumptions are made,
that things diverge.


Regards,

John Byrns

--
Surf my web pages at, http://fmamradios.com/

[Phil Allison]

"John Byrns"
"Patrick Turner"

The curve downwards of the "horizontals" of the square wave is the
action of the time constant, C being discharged after the transient
uprights.

** The straight lines of the wave would become curved in that case, not
merely tilted - as they are at much lower frequencies.

It is just down to relative phase shift in the harmonics, despite it
being
only a few degrees.

Hi Phil,

The straight lines are curved, take a closer look the curvature is
easily seen in the photo.


** The lower line is DEAD straight while the upper ones have a *tiny*
curvature.

The point about the visible tilt being due to "phase shift " is that all the
frequency components of that square wave are within 1% of their correct
amplitude values as for a perfect square wave.

A phase correction network alone with fix the tilt.



........ Phil

[Phil Allison]

"John Byrns"
"Phil Allison"
"Patrick Turner"

http://users.tpg.com.au/ldbutler/Intermodulation.htm ))


** All RF theory based - little relevance to audio.


I've got news for you,


** No you have not - you ****ing smartarse prick.


"RF" and "audio" electronics follow the same theory,


** See how my comment is right under a URL ?????

What do you suspect that means - ****head ???


That *PAGE* is written all about RF with no references to audio.

It was written by a bloody Yank Ham !!



......... Phil

[John Byrns]

In article ,
"Phil Allison" wrote:

"John Byrns"
"Phil Allison"
"Patrick Turner"

http://users.tpg.com.au/ldbutler/Intermodulation.htm ))


** All RF theory based - little relevance to audio.


I've got news for you,


** No you have not - you ****ing smartarse prick.


"RF" and "audio" electronics follow the same theory,


** See how my comment is right under a URL ?????

What do you suspect that means - ****head ???


That *PAGE* is written all about RF with no references to audio.

It was written by a bloody Yank Ham !!

Hi Phil,

None of that changes the fact that "RF" and "audio" follow the same
theory, you can curse all you want but it won't change the facts.


Regards,

John Byrns

--
Surf my web pages at, http://fmamradios.com/

[Phil Allison]

"John Byrns ****head & Asshole "



http://users.tpg.com.au/ldbutler/Intermodulation.htm ))


** All RF theory based - little relevance to audio.


I've got news for you,


** No you have not - you ****ing smartarse prick.


"RF" and "audio" electronics follow the same theory,


** See how my comment is right under a URL ?????

What do you suspect that means - ****head ???

That *PAGE* is written all about RF with no references to audio.

It was written by a bloody Yank Ham !!



None of that changes the fact that "RF" and "audio" follow the same
theory,


** Utterly IRRELEVANT !!!!!!

Not got one tiny thing to do with * my * comment

- you pathetic ASD ****ed ASSHOLE !!




....... Phil

[Patrick Turner]

Phil Allison wrote:

"Patrick Turner"

Someone ought do a web site on the CORRECT analysis of square waves
shapes
as seen on a CRO with various phase and frequency anomalies applied.


There is already some info on the Web about
'amplifier square wave phase shift effects' according to the Google
search.

http://users.tpg.com.au/ldbutler/Waveforms.htm

** LOL !!!

From some Yank ham radio ****** - so full of stupid errors.

Looks like he used the same erroneous old source as Churches did.

OK, but given time, Iain may spend enough long hours until the dawns
of many early mornings in his shed,
and there experiment with sq.waves and various order filters to see the
transient promoted effects, and ring frequencies produced.

I think anyone who does actually set up the experiments does learn much
real stuff about such phenomena.
But if the word WHY does not scream loud in their ears, maybe they never
learn.
Answers are not always easy to fathom.
Is *WHY!* a young man's cry, or an an old man's cry?.
Imagine a world without WHY!

I know **** all about quantum physics. There are Quarks and Bosuns in
there someplace,
photons that are maybe waves, maybe particles, but really I am ignorant,
and as experts
find more and more particles, the atom seems to be like Russian Doll
with a new face on each Doll
as you unpack yet another one, and the experts argue about what is
there, and might be there or not.
I hope they don't blow the world up looking.

How far does one have to go to understand how to make a good amp?

I think far enough by seat of pants calculation, and applied experience
of
networks around amps with FB so they just won't oscillate.
Usually, there is only ONE set of valid R&C compensation network values
valid for a given amp
to make it optimally stable yet have widest bandwith, most constant Rout
for the audio spectrum,
lowest thd etc etc.
After setting up dozens of amps with networks that work, one gets a
feeling about what R&C values to try,
and then you vary them to suit the amp.
I find a useful tool to be a radio tuning cap with two gangs of C giving
40pF to 720pF,
and a series pot of 50k.
Such a tool is ideal for Zobel values to shelve gain and alter phase
shift favourably.
One watches the 5kHz sq.wave on the CRO with a 0.22uF cap load, and
varies the R & C across the V1 anode load for
least overshoot,without allowing the sine wave response into the rated R
load to diminish too much.
There are 4 things to consider simultaneously. But unless one does, the
outcome will be
less than ideal. So you have to take several settings of R&C and change
to sine wave,
check the -3db with R load instead of the 0.22uf, then check with other
values of C...
Its all a bit mystifying to the uninitiated, and it all takes time to
achieve,
and methinks NFB cops a real beat up bashing by those who just don't get
it, ie the knack
of applying NFB without spoiling the amp, or the music.




" The evil that men do ......... "

http://users.tpg.com.au/ldbutler/Intermodulation.htm ))

** All RF theory based - little relevance to audio.

Yes, but IMD theory at AF has simularities with RF.

The difference is that with RF, we have a modulated carrier when we want
one,
and want the large intermodulation products, say the upper and lower
sidebands
of the AM wave so ppl can receive it.
But in AF, when we have two F present in a signal we DON'T want the IMD
products.
But if a large 80Hz signal and small 5kHz signal are both present, the
non linearities of the
devices cause the frequencies of 5,080Hz and 4,420Hz to appear...
The higher F is modulated like a an RF AM carrier wave by the lower F.
When a multitude of music F are present with great amplitude variations
and phase variations,
we don't want the extra crap thankyou, sum and difference F between all
signals present.
Its this spuriae which makes music suffer the most.
And grid current in direct coupled stages not well set up is a major
cause of very bad sounding IMD.

I didn't look to closely at the above sources, and it still didn't
explain
the phase shift of individual harmonics in a square wave after analysing
the shape of the sq.wave
deviation from the perfect wave.

Would it be useful to do such measurements?

I dunno how CORRECT you want the analysis for interpreting square wave
shapes.

** Simply not completely wrong or hopelessly ambiguous will do.

I did a Goolge search long ago and found nothing of value.

Now that I have a small digital camera ......

You should be able to catch all the action as it happens eh!

Sound waves produced by speakers being fed a square wave and picked up
by a mic
would give more dramatic amounts of phase movements of the various
harmonics.
The mauled sq.waves in rooms from speakers change shape as you move the
mic,
and the phase is all over the joint.
The CRO will show the wave, but all of the interpretation of the wave
content is difficult by just looking at it,
beyond the basics we already know.

Patrick Turner.


...... Phil

[Phil Allison]

"Patrick Turner "

From some Yank ham radio ****** - so full of stupid errors.

Looks like he used the same erroneous old source as Churches did.

OK, but given time, Iain may spend enough long hours until the dawns
of many early mornings in his shed,
and there experiment with sq.waves and various order filters to see the
transient promoted effects, and ring frequencies produced.


** All someone really needs is a 1/3 octave graphic equaliser in the link
from generator to CRO.

See it all happening before your eyes.




........ Phil

[Patrick Turner]

Phil Allison wrote:

"John Byrns ****head & Asshole "

http://users.tpg.com.au/ldbutler/Intermodulation.htm ))


** All RF theory based - little relevance to audio.


I've got news for you,


** No you have not - you ****ing smartarse prick.


"RF" and "audio" electronics follow the same theory,


** See how my comment is right under a URL ?????

What do you suspect that means - ****head ???

That *PAGE* is written all about RF with no references to audio.

It was written by a bloody Yank Ham !!



None of that changes the fact that "RF" and "audio" follow the same
theory,

** Utterly IRRELEVANT !!!!!!

Not got one tiny thing to do with * my * comment

- you pathetic ASD ****ed ASSHOLE !!

...... Phil



Here is the reference URL and Phil's comment immediately after it :-

http://users.tpg.com.au/ldbutler/Intermodulation.htm ))


** All RF theory based - little relevance to audio.

Some would agree with Phil, and some might say
well the production of RF sidebands, intermodulation products, are the
same
phenomena at RF or at AF, so the explanations given about the RF
behaviour
at the URL are also valid for AF, hence relevant, even if not explained
above for AF.

But there is nothing at the URL above which is relevant to phase shifted
harmonics of square waves.


Can we move on?

Patrick Turner

[John Byrns]

In article ,
"Phil Allison" wrote:

"John Byrns"
"Patrick Turner"

The curve downwards of the "horizontals" of the square wave is the
action of the time constant, C being discharged after the transient
uprights.

** The straight lines of the wave would become curved in that case, not
merely tilted - as they are at much lower frequencies.

It is just down to relative phase shift in the harmonics, despite it
being
only a few degrees.

Hi Phil,

The straight lines are curved, take a closer look the curvature is
easily seen in the photo.

** The lower line is DEAD straight while the upper ones have a *tiny*
curvature.

The point about the visible tilt being due to "phase shift " is that all the
frequency components of that square wave are within 1% of their correct
amplitude values as for a perfect square wave.

Hi Phil,

I agree that the amplitude errors of a the square wave components is
less than 1%, but that does not imply that the tilt we see is not the
result of a time constant as Patrick originally asserted, There is no
contradiction in a simple time constant causing exactly the type of
waveform we see in Ian's photo.


Regards,

John Byrns

--
Surf my web pages at, http://fmamradios.com/

[John Byrns]

In article ,
Patrick Turner wrote:

Here is the reference URL and Phil's comment immediately after it :-

http://users.tpg.com.au/ldbutler/Intermodulation.htm ))


** All RF theory based - little relevance to audio.

Some would agree with Phil, and some might say
well the production of RF sidebands, intermodulation products, are the
same
phenomena at RF or at AF, so the explanations given about the RF
behaviour
at the URL are also valid for AF, hence relevant, even if not explained
above for AF.

But there is nothing at the URL above which is relevant to phase shifted
harmonics of square waves.

Can we move on?

Hi Patrick,

Sure we can move on, getting back to square waves, it is not so much the
harmonics that are phase shifted, it is mainly the fundamental that is
phase shifted, the higher the harmonic the less the phase shift.


Regards,

John Byrns

--
Surf my web pages at, http://fmamradios.com/