amplifier output impedance measurement

[jan.didden]

Just thought about another aspect of this. You know those input/output differencing tests by Peter Baxandall and Dave Hafler, where they subtract amp input from a (fraction of) amp output in opposite phase. What is left is the distortion, and you can thus listen to that in isolation.

To get a good subtraction you have to send the input signal through a low pass filter to mimic the low pass character of the amp, and you can with some care get a null of 60dB or more.

This would not work if there was any delay in the amp.

Jan

 

[atmasphere]

Funny thing is- I am familiar with that stuff. I think what is throwing me is I am seeing this at frequencies too low to be involved with the cutoff of a filter. IOW, I shouldn't see it at 100 Hz if the upper -3db point of the circuit is over 100KHz, at least, that is what I was taught in school....

 

[jan.didden]

Yeah you got a point there.
Can you post a screen shot or something.

Jan

 

[atmasphere]

Working on it! I think I must have been using a Tektronics 475 when I observed this- definitely not the 15MHz bench scope I use for basic troubleshooting. The 475 has been showing its age and has been on the blink (literally) recently...

 

[SY]

Funny thing is- I am familiar with that stuff. I think what is throwing me is I am seeing this at frequencies too low to be involved with the cutoff of a filter. IOW, I shouldn't see it at 100 Hz if the upper -3db point of the circuit is over 100KHz, at least, that is what I was taught in school....

What's the LF cutoff?

 

[atmasphere]

2Hz.

 

[SY]

Thanks- so it's a lot closer.

 

[rayma]

Just thought about another aspect of this. You know those input/output differencing tests by Peter Baxandall and Dave Hafler, where they subtract amp input from a (fraction of) amp output in opposite phase. What is left is the distortion, and you can thus listen to that in isolation.

To get a good subtraction you have to send the input signal through a low pass filter to mimic the low pass character of the amp, and you can with some care get a null of 60dB or more.

This would not work if there was any delay in the amp.

Jan

Some tests of this nature actually include a specific length of shielded cable to add a time delay to the input, for correct timing in the cancellation.

 

[DF96]

100kHz HF rolloff will give you an apparent 'delay' of 1.6us.

 

[rayma]

100kHz HF rolloff will give you an apparent 'delay' of 1.6us.

Yes, for a single frequency signal (a sine wave) we could view a phase shift as a sort of time delay (assuming no amplitude change).
However, for a signal that contains more than one frequency this would not work, since the waveform's time domain shape would not be preserved.
To keep the waveform the same shape, the phase shift must be proportional to frequency. This mathematically is a pure delay.
A low pass filter has a nonlinear phase shift, described by an arctangent function, as well as the amplitude changes.

 

[jan.didden]

Group delay comes to mind...

 

[Elvee]

Just thought about another aspect of this. You know those input/output differencing tests by Peter Baxandall and Dave Hafler, where they subtract amp input from a (fraction of) amp output in opposite phase. What is left is the distortion, and you can thus listen to that in isolation.

To get a good subtraction you have to send the input signal through a low pass filter to mimic the low pass character of the amp, and you can with some care get a null of 60dB or more.

This would not work if there was any delay in the amp.

In fact, it would work and it actually works, yet amplifiers have some tens (or more) of ns of pure (TL-like) delay.
You just need to have the phase shift right for the fundamental, the rest is unimportant and this can be implemented with a number (down to 1) of 1st order cells

Group delay comes to mind...

JD dreamt it, LV did it:

The parameter linking the two is called group delay, but there are some caveats: it has to do with cutoff frequencies, etc.
In addition, it isn't flat with frequency, unlike an ideal transmission line.

 

[jan.didden]

JD dreamt it, LV did it:

LV did it, Didden didn't do it

Jan

 

[atmasphere]

Just thought about another aspect of this. You know those input/output differencing tests by Peter Baxandall and Dave Hafler, where they subtract amp input from a (fraction of) amp output in opposite phase. What is left is the distortion, and you can thus listen to that in isolation.

To get a good subtraction you have to send the input signal through a low pass filter to mimic the low pass character of the amp, and you can with some care get a null of 60dB or more.

This would not work if there was any delay in the amp.

Jan

Sure it would- but you would not get an accurate measurement of the distortion...

 

[Johan Potgieter]

As an aside:

Didden didn't do it