On Mon, 13 Dec 2004 23:27:18 +0200, "Iain M Churches"
wrote:

I would be interested to know the phase response
of a high end SS amp (OK, I know, they are ALL high end:-)
at 20Hz and say 20kHz, or more points if the info is
available.

If someone can supply me with some figs (from say the
legendary Krell:-) I cold use these a benchmark for the
measurements of my own amp.

It obviously depends on the bandwidth of the amp, which doesn't have
much to do with 'high end' credentials. A Naim amp for instance is
specifically designed to have limited bandwidth, and will have phase
angles of around 10-20 degrees at 20Hz and 20-30 degrees at 20kHz. At
the other extreme, a Spectral will have phase angles of less than five
degrees at these frequencies. My own Audiolab 8000P measures about 5-7
degrees at 20Hz and 8-10 degrees at 20kHz (approximated due to being
measured on a 'scope, since I don't have a vector voltmeter). The
Krell is about the same. Some SS amps are DC coupled, which will give
them zero phase angle at 20Hz.

The only thing 'legendary' about the classic Krells is that they
operate in Class A at their 8 ohm rated output, and that they double
that rated output for each halving of impedance down to 1 ohm.
Otherwise, they are not especially wideband or especially low
distortion. They are however, like most good SS amps and a very few
valve amps, sonically transparent.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

"Stewart Pinkerton" wrote in message
...
On Mon, 13 Dec 2004 23:27:18 +0200, "Iain M Churches"
wrote:

It obviously depends on the bandwidth of the amp, which doesn't have
much to do with 'high end' credentials.

Understood:-)
But I assumed the better the amp the better the performance.


A Naim amp for instance is
specifically designed to have limited bandwidth, and will have phase
angles of around 10-20 degrees at 20Hz and 20-30 degrees at 20kHz. At
the other extreme, a Spectral will have phase angles of less than five
degrees at these frequencies. My own Audiolab 8000P measures about 5-7
degrees at 20Hz and 8-10 degrees at 20kHz (approximated due to being
measured on a 'scope, since I don't have a vector voltmeter). The
Krell is about the same. Some SS amps are DC coupled, which will give
them zero phase angle at 20Hz.

Hmm! Thanks for the info. These are pretty impressive figs.
I have access to a phase shift generator/meter combination (digital of
course:-) made by Feedback Electronics in the UK.
That's why I was interested to obtain some kind of a yardstick.

An initial investigation of my 50W amplifier which has the first and second
stages DC coupled, shows -3 degrees at 10Hz, and -8 degrees at 50kHz.
So about the same as the Audiolab 8000P, and very close to the Spectral?
Not too shabby:-))

This amp has custom-designed Sowter transformers. A similar amp with
Hammond iron may not perform so well.

Iain

On Tue, 14 Dec 2004 11:25:59 +0200, "Iain M Churches"
wrote:


"Stewart Pinkerton" wrote in message
.. .
On Mon, 13 Dec 2004 23:27:18 +0200, "Iain M Churches"
wrote:

It obviously depends on the bandwidth of the amp, which doesn't have
much to do with 'high end' credentials.

Understood:-)
But I assumed the better the amp the better the performance.


A Naim amp for instance is
specifically designed to have limited bandwidth, and will have phase
angles of around 10-20 degrees at 20Hz and 20-30 degrees at 20kHz. At
the other extreme, a Spectral will have phase angles of less than five
degrees at these frequencies. My own Audiolab 8000P measures about 5-7
degrees at 20Hz and 8-10 degrees at 20kHz (approximated due to being
measured on a 'scope, since I don't have a vector voltmeter). The
Krell is about the same. Some SS amps are DC coupled, which will give
them zero phase angle at 20Hz.

Hmm! Thanks for the info. These are pretty impressive figs.
I have access to a phase shift generator/meter combination (digital of
course:-) made by Feedback Electronics in the UK.
That's why I was interested to obtain some kind of a yardstick.

An initial investigation of my 50W amplifier which has the first and second
stages DC coupled, shows -3 degrees at 10Hz, and -8 degrees at 50kHz.
So about the same as the Audiolab 8000P, and very close to the Spectral?
Not too shabby:-))

This amp has custom-designed Sowter transformers. A similar amp with
Hammond iron may not perform so well.

Iain


I wouldn't get too bothered by the phase response of an amplifier,
other than as an indicator of bandwidth (in a competently designed
amplifier where rolloffs are defined by single poles, -3dB = 90
degrees). Do remember that a speaker follows, with phase shifts that
are all over the place by comparison.

d

Pearce Consulting
http://www.pearce.uk.com

In article , Iain M Churches
wrote:


Hmm! Thanks for the info. These are pretty impressive figs. I have
access to a phase shift generator/meter combination (digital of
course:-) made by Feedback Electronics in the UK. That's why I was
interested to obtain some kind of a yardstick.

An initial investigation of my 50W amplifier which has the first and
second stages DC coupled, shows -3 degrees at 10Hz, and -8 degrees at
50kHz. So about the same as the Audiolab 8000P, and very close to the
Spectral? Not too shabby:-))

Are the above values with an 8 Ohm load from an 8 Ohm tap?

If you are concerned about this, I'd suggest you also check with reactive
and/or mismatched loads. FWIW Naim amps have tended in the past to have a
series resistor (about 0.2 Ohms IIRC) in their output, and then tell the
user to employ about 10 microH worth of special cable in series. (Which is
conveniently omitted from magazine measurements but is in place when using
the amp as recommended.)

Whereas many non-Naim SS amps either have direct output or have a small
series inductor. These can affect the phase response at 20kHz and mean that
values obtained into an 8 Ohm resistor are not always a good guide.
Similarly, if your valve amp has any reactance or an impedance much above
0.1 Ohms, this may interact with reactive loads and give a distinct change
in result.

How much this matters is a different matter, though. The figures you quoted
look - in themselves - perfectly respectable. Hence unless your amp has a
particularly high output impedance and/or is reactive at HF/LF I'd doubt
this matters compared to the way most speakers and rooms will affect the
phase of what reaches the listener. :-)

This amp has custom-designed Sowter transformers. A similar amp with
Hammond iron may not perform so well.

Since you are doing this, I'd be interested to know the complex output
impedance you get across the audio band as this seems to be the kind of
measurement that rarely gets done and published for *any* amp these days...

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

"Jim Lesurf" wrote in message
...
In article , Iain M Churches
wrote:
An initial investigation of my 50W amplifier which has the first and
second stages DC coupled, shows -3 degrees at 10Hz, and -8 degrees at
50kHz. So about the same as the Audiolab 8000P, and very close to the
Spectral? Not too shabby:-))

Are the above values with an 8 Ohm load from an 8 Ohm tap?

Yes, they are.

If you are concerned about this, I'd suggest you also check with reactive
and/or mismatched loads.

Not so much concerned as interested:-) The figs in themselves do not mean
much if I have no yardstick - hence ther question.



The figures you quoted
look - in themselves - perfectly respectable. Hence unless your amp has a
particularly high output impedance

The damping factor is 20, which for a valve amp very respectable:-)

and/or is reactive at HF/LF I'd doubt
this matters compared to the way most speakers and rooms will affect the
phase of what reaches the listener. :-)

Understood.

This amp has custom-designed Sowter transformers. A similar amp with
Hammond iron may not perform so well.

Since you are doing this, I'd be interested to know the complex output
impedance you get across the audio band as this seems to be the kind of
measurement that rarely gets done and published for *any* amp these
days...

Can you suggest the best way to ascertain these figs?


Iain

In article , Iain M Churches
wrote:

"Jim Lesurf" wrote in message
...


The figures you quoted look - in themselves - perfectly respectable.
Hence unless your amp has a particularly high output impedance

The damping factor is 20, which for a valve amp very respectable:-)

Given the qualification in the statement, yes. :-)

However bear in mind that a damping factor of 20 for an 8 Ohm load implies
an output impedance of 0.4 Ohms. If this kind of value exists across the
audio band then you can expect interactions with a typical loudspeaker's
impedance to produce changes in the power-frequency response that can be
very audible, and changes in the phase-frequency response that may be far
higher than the values with an 8 Ohm load. Hence I'd tend to regard such a
damping factor as being a bit low, but despite that I'd expect the results
to sound fine in most cases.

Also bear in mind - as implied by Arny's comments - that these values may
well be dependent upon the output power level, particularly with amps that
have low levels of feedback. (Which is implied by having a high output
impedance unless the feeback is from the primary.)

None of the above may matter much in use. Depends on circumstances. But it
can mean that values measured using an 8 Ohm load don't really tell you
what will happen when you connect to a speaker.


Since you are doing this, I'd be interested to know the complex output
impedance you get across the audio band as this seems to be the kind
of measurement that rarely gets done and published for *any* amp these
days...

Can you suggest the best way to ascertain these figs?

Basically the same as when measuring 'damping factor' (a term I don't
personally like as I think it is misleading).

Either:

1) Set the signal level and vary the load, and note the o/p level for two
different loads. Then use this to work out the o/p impedance. If the amp is
stable, one 'load' can be o/c as this gives you the o/p emf directly.

or

2) Ensure the amp is trying to o/p zero, but drive a signal into the
*output* via a suitable resistor. Use the other channel of the power amp to
do this if it is convenient. Then measure the drive level and the level
that appear on the o/p and work out the o/p impedance that way.

Repeat for various frequencies.

For a low feedback amp you may need to do this at high power as well as low
to see if the values change. However this may be difficult via (2) as some
amps may not like this process.

IIRC In a previous posting you say you have a phase meter, so by measuring
the signal phases as well as amplitudes you can work out the o/p impedance
as a complex value and hence determine the entire complex impedance as a
function of frequency. This can matter due to interactions with the
complex load presented by a loudspeaker. For an example of this have a look
at the model of the '303' amp and '57' speaker which is linked to my 'Audio
Misc' page (URL in my sig, below). This shows why - when the amp output
impedance isn't essentially zero - these effects can dwarf any departures
for a flat response when measureing with an 8 Ohm load. :-)

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

"Jim Lesurf" wrote in message
...
In article , Iain M Churches
wrote:

"Jim Lesurf" wrote in message
...


Thanks for your reply, Jim. If you don't mind I will cut it into
sections so that I can read and digest the information which each
contains before replying .


The damping factor is 20, which for a valve amp very respectable:-)

Given the qualification in the statement, yes. :-)

However bear in mind that a damping factor of 20 for an 8 Ohm load implies
an output impedance of 0.4 Ohms. If this kind of value exists across the
audio band then you can expect interactions with a typical loudspeaker's
impedance to produce changes in the power-frequency response that can be
very audible, and changes in the phase-frequency response that may be far
higher than the values with an 8 Ohm load. Hence I'd tend to regard such a
damping factor as being a bit low, but despite that I'd expect the results
to sound fine in most cases.

Over the years, I have been involved in the measuring of quite a lot
of push pull UL valve amps on a comparative basis. I have also
been fortunate enough to have worked alongside or met
some good valve amp builders or designers
(ex. the legendary Jim Kerr of Kerr McCosh)
Sadly, I was young at the time, and did not ask many of the questions
to which I would now like the answers:-(

I even saw Henry Leak, when my Dad took me along to
return his TL12 amp for a valve change. We went to the factory in
Acton London W3. I remember seeing rows of ladies in their forties,
nimbly assembling circuit boards, which still are pristine by any
standards.

Only a few of the best sounding amps have large amounts of NFB.
15dB seems typical. Some using pentodes at the front end have a lot
more.

Most of them have a DF between 10-15, so I thought my fig of 20
was very respectable. Both Tremaine and Olson stated that in tests,
a listening panel were unable to detect changes in increase in DF
above about 12. I know that SS amps can easily achieve a much higher
figure.

DF is an interesting parameter, and I feel it has a lot to do in explaining
why amplifiers with a similar FR or PBW sound so different.
Arthur Radford built a series of speakers for studio use, with valve
amplifiers built in on the bottom of the stand. The speaker impedance
was 24 Ohms, with a corresponding output transformer.
This way, he could achieve a greater DF. They did
sound very good indeed:-)

Normally we express DF in terms of the ratio of the internal output
impedance to the load impedance. (and so arrive at the figure of
0.4 Ohms which you quoted above) DF = Zload/Zout

But according to Tremaine, the equation is more complex,
and should also take into consideration the DC resistance of
the voice coil, which is a limiting factor.

So the equation becomes DF = Zload/(Zout+Rvc).
Assuming a Rvc=6 Ohms, a DF of 16 calculated by the first equation
becomes 1.23 using the second equation, and even with zero output
impedance, a DF of infinity in the first equation gives us only 1.33
when calculated in the second.

ref: Tremaine, pp1120


Kunniottaen!
Iain










Also bear in mind - as implied by Arny's comments - that these values may
well be dependent upon the output power level, particularly with amps that
have low levels of feedback. (Which is implied by having a high output
impedance unless the feedback is from the primary.)

None of the above may matter much in use. Depends on circumstances. But it
can mean that values measured using an 8 Ohm load don't really tell you
what will happen when you connect to a speaker.

In article , Iain M Churches
wrote:

"Jim Lesurf" wrote in message
...
In article , Iain M Churches
wrote:

"Jim Lesurf" wrote in message
...


Thanks for your reply, Jim. If you don't mind I will cut it into
sections so that I can read and digest the information which each
contains before replying .

OK. :-)

[snip some discussion of damping factor]



Over the years, I have been involved in the measuring of quite a lot of
push pull UL valve amps on a comparative basis. I have also been
fortunate enough to have worked alongside or met some good valve amp
builders or designers (ex. the legendary Jim Kerr of Kerr McCosh) Sadly,
I was young at the time, and did not ask many of the questions to which
I would now like the answers:-(

I have had similar experiences/regrets. I also now regret I did not keep a
much more careful 'diary' of what I did 20-30 years ago as I now sometimes
struggle to recall results, discussions, etc. Loads of info familiar to me
at the time, but which I have now largely forgotten. Peril of old age, and
youth being wasted on the young. :-)

[snip]

DF is an interesting parameter, and I feel it has a lot to do in
explaining why amplifiers with a similar FR or PBW sound so different.

I would agree. Although I feel the term 'damping factor' is quite
misleading as it allows people to assume that the main effect is in damping
the mechanical resonance of the loudpeaker, when I think this is generally
not the reason for its main audible effects.

[snip]

But according to Tremaine, the equation is more complex, and should also
take into consideration the DC resistance of the voice coil, which is a
limiting factor.

Yes. I think that is an important point. It shows why - once the
amplifier's damping factor exceeds about 20 any further increases may not
really have any effect - as your estimate below implies.

So the equation becomes DF = Zload/(Zout+Rvc). Assuming a Rvc=6 Ohms, a
DF of 16 calculated by the first equation becomes 1.23 using the second
equation, and even with zero output impedance, a DF of infinity in the
first equation gives us only 1.33 when calculated in the second.

Although I can't comment on the specific values for a given speaker, I'd
agree with the implication. That once we have an amplifier output impedance
which is somewhat less than the series resistance of the speaker unit we
should find that the effect on 'damping' becomes much the same as we
continue to reduce the amp's output impedance. i.e. if your only concern is
damping the loudspeaker resonance then a very low output impedance isn't
required.

However I'd say that the real effect resides elsewhere... :-)

Consider a loudspeaker whose load impedance varies from, say, 5 Ohms at
some frequencies up to, say, 32 Ohms at others. Then consider using this
with an amp that has a damping factor of, say, 20 (ref 8 Ohms).

The speaker is nominally designed on the basis of being fed with a voltage
source (i.e. of negligable source impedance). This has been the standard
practice in the domestic audio business in my experience.

The damping factor translates into a nominal amp output impedance of 0.4
Ohms.

At the frequencies where the speaker has an impedance of 5 Ohms, the 0.4
Ohms between the nominal EMF voltage of the amp output combined with the
speaker load impedance acts as a potential divider, and the voltage
appearing on the speaker terminals will be 5/(5+0.4) = 0.926 of the level
which would appear there if the same amp had a zero output impedance. This
corresponds to the power changing by -0.67 dB.

Whereas, at frequencies where the speaker has an impedance of 32 Ohms, the
corresponding drop will be 32/32.4 = 0.987 or -0.1 dB.

The result is variations at the 0.5dB level in the response.

If the speaker impedance varies more than this, and if the speaker and amp
have reactive components in their impedances, the variations may be larger.
This also has implications for the phase response which would need to be
considered if this is felt to be an issue.

Hence we can find that the simple interaction between the amplifier's
output impedance and the frequency dependence of the speaker impedance can
make slight but audible changes to the overall response. When listeners are
unaware of this mechanism it can then given them a 'cue' which may affect
their judgements. They may also assume the audible differences are due to
something else.

How much this matters, and whether the result is felt to be 'better' or
'worse' will depend on the circumstances. However it is a 'hidden' variable
which people tend to ignore or misunderstand and complicates the issue so
far as I am concerned.

FWIW One of the amps developed and sold by the company I used to work for
had an output impedance that not only went low, but became slightly
negative in the mid-bass region. This replaced an earlier SS design amp
that had an output impedance that approached an Ohm. :-) (I did not do
the initial design for either of these.) You may not be surprised to be
told that the two amps were felt to sound quite different in many
circumstances. (Although I'm sure there were other contributions to this.
:-) )

My personal preference tends to be to have a low output impedance. If
someone wants the changes that a higher value gives it is easier to add a
few tenths of an Ohm in series than it is to get a lower output impedance
from a commercial design. I tend also to prefer to remove hidden variables
which are essentially uncontrolled.

That said, although my pref is to go for amp o/p impedances below 0.1 Ohm,
I see no reason to dismiss amps that have relatively high impedances. They
may work very nicely, and give results someone prefers in relevant
circumstances. The typical effects are quite small, anyway. However for
these reasons I tend to avoid the term 'damping factor' as for reasons you
have indicated yourself. It tends to draw attention away from the more
likely reasons for any audible effect in many cases.

An extra issue with low feedback is that the output impedance may be level
dependent as well as frequency dependent, and this further complicates the
issue...

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

In message , Iain M Churches
writes

"Jim Lesurf" wrote in message
...
In article , Iain M Churches
wrote:

"Jim Lesurf" wrote in message
...


Thanks for your reply, Jim. If you don't mind I will cut it into
sections so that I can read and digest the information which each
contains before replying .


The damping factor is 20, which for a valve amp very respectable:-)

Given the qualification in the statement, yes. :-)

However bear in mind that a damping factor of 20 for an 8 Ohm load implies
an output impedance of 0.4 Ohms. If this kind of value exists across the
audio band then you can expect interactions with a typical loudspeaker's
impedance to produce changes in the power-frequency response that can be
very audible, and changes in the phase-frequency response that may be far
higher than the values with an 8 Ohm load. Hence I'd tend to regard such a
damping factor as being a bit low, but despite that I'd expect the results
to sound fine in most cases.

Over the years, I have been involved in the measuring of quite a lot
of push pull UL valve amps on a comparative basis. I have also
been fortunate enough to have worked alongside or met
some good valve amp builders or designers
(ex. the legendary Jim Kerr of Kerr McCosh)
Sadly, I was young at the time, and did not ask many of the questions
to which I would now like the answers:-(

I even saw Henry Leak, when my Dad took me along to
return his TL12 amp for a valve change. We went to the factory in
Acton London W3. I remember seeing rows of ladies in their forties,
nimbly assembling circuit boards, which still are pristine by any
standards.

Only a few of the best sounding amps have large amounts of NFB.
15dB seems typical. Some using pentodes at the front end have a lot
more.

Most of them have a DF between 10-15, so I thought my fig of 20
was very respectable. Both Tremaine and Olson stated that in tests,
a listening panel were unable to detect changes in increase in DF
above about 12. I know that SS amps can easily achieve a much higher
figure.

DF is an interesting parameter, and I feel it has a lot to do in explaining
why amplifiers with a similar FR or PBW sound so different.
Arthur Radford built a series of speakers for studio use, with valve
amplifiers built in on the bottom of the stand. The speaker impedance
was 24 Ohms, with a corresponding output transformer.
This way, he could achieve a greater DF. They did
sound very good indeed:-)

Normally we express DF in terms of the ratio of the internal output
impedance to the load impedance. (and so arrive at the figure of
0.4 Ohms which you quoted above) DF = Zload/Zout

But according to Tremaine, the equation is more complex,
and should also take into consideration the DC resistance of
the voice coil, which is a limiting factor.

So the equation becomes DF = Zload/(Zout+Rvc).
Assuming a Rvc=6 Ohms, a DF of 16 calculated by the first equation
becomes 1.23 using the second equation, and even with zero output
impedance, a DF of infinity in the first equation gives us only 1.33
when calculated in the second.

ref: Tremaine, pp1120


Kunniottaen!
Iain





If the amp is feeding a speaker directly, with no crossover, then a very
low output impedance is not required, because the dc resistance of the
voice coil is dominant. The same cannot be said for multi-unit speakers
with complex crossovers, where the interaction of the non-linearity (in
both frequency and level) of the valve amplifier output Z, and the
crossover can cause problems.

--
Chris Morriss

On Tue, 14 Dec 2004 11:25:59 +0200, "Iain M Churches"
wrote:


"Stewart Pinkerton" wrote in message
.. .
On Mon, 13 Dec 2004 23:27:18 +0200, "Iain M Churches"
wrote:

It obviously depends on the bandwidth of the amp, which doesn't have
much to do with 'high end' credentials.

Understood:-)
But I assumed the better the amp the better the performance.

That depends how you define 'batter'. The Spectral for instance has
megahertz bandwidth, so immaculate phase response at 20kHz, but is
notoriously unstable. The Naim OTOH has been highly regarded by
audiophiles for about thirty years, despite being sensitive to
capacitive lopads and demanding highly inductive speaker cable to
maintain its warranty. VCome to think of it, the phase response is a
lot worse if you include that cable, as you should since Naim consider
iot to be an essential part of the circuit driving the speaker.
Basically, there's a *lot* more to 'high fidelity' than sheer
bandwidth. Many decades ago, the brilliant Gilbert Briggs of
Wharfedale coined a very useful phrase - the wider you open the
window, the more the muck flies in!

An initial investigation of my 50W amplifier which has the first and second
stages DC coupled, shows -3 degrees at 10Hz, and -8 degrees at 50kHz.
So about the same as the Audiolab 8000P, and very close to the Spectral?
Not too shabby:-))

That's very impressive for a valve amp.

This amp has custom-designed Sowter transformers. A similar amp with
Hammond iron may not perform so well.

Impressive results explained! :-)
--

Stewart Pinkerton | Music is Art - Audio is Engineering