John Curl's Blowtorch preamplifier part II

[RNMarsh]

Originally Posted by RNMarsh

I am unable to find the paper being presented at this months AES. Pls give exact info... type of paper, author name and paper title. The number 8722 did not bring it up. Thx RNM

Found it -- now if it is avail to download yet --- > Still on island off an island. Traveling tomorrow. Talk at ya later.

[scott wurcer]

Originally Posted by RNMarsh

[Something the FFT throws away (ignors), usually.]

Thx - RN

Usually the user throws it away by improperly posing the problem. We are back here I see. The FFT "throws away" nothing.

[gpapag]

For people who have access to the AES Convention Paper 8722:
Any similarities with the content of this patent?

METHOD AND APPARATUS TO MEASURE DIFFERENTIAL PHASE AND FREQUENCY MODULATION DISTORTIONS FOR AUDIO EQUIPMENT

George

[magnoman]

Originally Posted by jan.didden

DC offset at the output Phase shift @ 15kHz
-2V -25 degr
-1V -24 degr
0V -23 degr
+1V -22 degr
+2V -21 degr

jan

Jan, I'm probably completely missing the point, but isn't the phase shift very high for 15KHz and thus likely configured with some very high closed loop gain.
Even for a simple single pole at 22 phase I think it would take just over 0.5% gain change per degree phase shift, consistent with cmrr of ~40db at 15KHz.

Thanks
-Antonio

[RNMarsh]

Originally Posted by scott wurcer

Usually the user throws it away by improperly posing the problem. We are back here I see. The FFT "throws away" nothing.

I dont mean in theory... but often in practice. Thus the word - ignore was included. did you ignore that? Anyway - beside the point of my main thrust... re cmr and PIM.

[the FFT algor usually drops the dc term as it isnt needed to determine harmonics etc. But yes, it is there in the theory books]

[RNMarsh]

Originally Posted by gpapag

For people who have access to the AES Convention Paper 8722:
Any similarities with the content of this patent?

METHOD AND APPARATUS TO MEASURE DIFFERENTIAL PHASE AND FREQUENCY MODULATION DISTORTIONS FOR AUDIO EQUIPMENT

George

It looks to be the testing set up for the AES paper is here. No data on test results but that is the test concept, I'm sure. Thx-RNM

[PMA]

Richard, audio signal is assymetric, I agree. But has zero dc component..

[Bob Cordell]

Originally Posted by jan.didden

Actually in a way that phase modulation is quite easy to measure for all of us:
Measure the phase shift between input and output with zero Vout.
Then insert a DC offset of a few volts, and again measure it; you'll find it most probably has changed.
One typical table from his paper:

DC offset at the output Phase shift @ 15kHz
-2V -25 degr
-1V -24 degr
0V -23 degr
+1V -22 degr
+2V -21 degr

So there's you varying phase shift with output level.
Fascinating, isn't it?

jan

Hi Jan,

Although I haven't seen Ron's paper, I'm really looking forward to seeing it.

I think the phase deviations you show here are probably fairly easy to explain. As the operating points change, the forward gain of the amplifier changes and the closed-loop bandwidth changes. A far-out change in the closed-loop bandwidth, often acting roughly like a single pole at the gain crossover frequency, will cause a change in phase shift at frequencies well below that crossover frequency. That is what causes PIM.

These operating point forward gain changes can occur due to numerous things happening, but at high frequencies one of them could be the voltage-dependency of junction capacitances.

One thing of interest in the numbers above is that they seem to suggest that the closed-loop 3 dB frequency is fairly close-in, unless there is some other source of phase lag in the setup. Bear in mind that the phase lag of a single pole one octave below that pole is about 22 degrees. If the numbers above were attributable to a single pole, one would be inclined to speculate that the closed-loop pole was only about an octave above the test frequency of 15 kHz, implying a closed loop bandwidth on the order of only about 30 kHz. This would be very, very low.

It would be interesting to know what the closed-loop amplitude response of the amplifier under test is.

It would also be interesting to know how much the 20-kHz THD of that amplifier was under those same conditions.

The idea of looking at phase lag as a function of a DC offset is largely the same as Matti Otala's original proposal for measuring PIM. That is, mix a 60Hz signal with a 6 kHz sigbal and look at the phase modulation on the 6kHz "carrier" after the test signal is passed through the amplifier under test. This is sort of like the SMPTE IM test, except that we look for phase modulation on the carrier instead of amplitude modulation (AIM). The coherent IM analyzer I built to measure this is described in my JAES PIM paper published in 1983 (a copy of which is on my web site at CordellAudio.com - Home). As long as we can argue that the 60Hz stimulous is low enough in frequency to look like a DC offset for purposes of measurement, the two approaches are essentially the same.

In that paper, a TL071 operating at an inverting gain of 10 and with an output of 6V rms or the p-p equivalent of that value was found to have THD-20 of 0.013%, SMPTE IM (AIM) of 0.0017% and PIM of 1.6 nanoseconds rms when measured in accordance with Otala's scheme.

It is virtually impossible to have any significant amount of PIM without also having an easily measureable amount of THD-20, since the same nonlinearity that causes PIM will also cause HF THD.

Cheers,
Bob

[Esperado]

I wonder how we can be concerned in 2012 with phase's problems with amps able to have a flat bandwidth up to 1Mhz, witch is one of the requirements, on my point of view, for amps pretending to be very "high-end" for audio.
The others being level linearity and low noise
What else ?

[edit] In accordance to the last Bob Cordell's post, and its old and accurate papers on TIM, open loop bandwidths, and closed loop amps.

[Wavebourn]

Originally Posted by scott wurcer

Usually the user throws it away by improperly posing the problem. We are back here I see. The FFT "throws away" nothing.

It's the observer who does not see sidebands of different heights.