Is there a simple clear explanation and diagram showing how to measure distortion using the compensation method? Its a term I had never seen in the last 50 years of designing and measuring audio. Perhaps I understand it as a different term?
There is a technique used to validate high speed data links called bit error rate testing. A variation that would be valid (and not easy) would be to use pseudorandom noise band limited to the audio band, ideally shaped to the typical spectrum of music and compare the input to the output of an audio link. You could derive some analytic techniques to see what aspect is getting changed. Using random noise should test all possible transistions and levels after some reasonable time.
There is a technique used to validate high speed data links called bit error rate testing. A variation that would be valid (and not easy) would be to use pseudorandom noise band limited to the audio band, ideally shaped to the typical spectrum of music and compare the input to the output of an audio link. You could derive some analytic techniques to see what aspect is getting changed. Using random noise should test all possible transistions and levels after some reasonable time.
This method of measuring distortion was defended by Sapozhkov in 1956 with his doctoral dissertation.
However, for its correct application, the delay line and the comparison circuit (adder) must be ideal.
Today, anyone who understands the essence can use this method in the simulator.
regards
Petr
However, for its correct application, the delay line and the comparison circuit (adder) must be ideal.
Today, anyone who understands the essence can use this method in the simulator.
regards
Petr
Attachments
As I thought, different terminology. The basic principle was used in RCA distortion analyzers from the 1930's:
And in more recent designs like the Cordell distortion magnifier use the same principle.
Any amplifier can be modeled as a low pass filter (no amps I can concieve of have unlimited bandwidth). Most audio amps are bandass filters. While I think simulators quickly become mental masturbators, you can easily explore a "perfect" amp in your simulations to look for the dreaded "speed distortion". Just model a bandpass filter that matches a typical amp. in fact that would be a more appropriate reference path than a simple phase shift + delay. You should also be able to easily figure out what the requirements would be for a bandlimited source to have no "speed" distortion.
Acoustic music is usually captured with microphones that low pass around 15 KHz. The wider band mikes usually are noisier and not preferred. The real wideband mikes are far noisier (1" vs 1/4" disphragm area). Its very hard to justify testing with spectral content above 20 KHz.
However if you look at the whole chain from mike to speaker there are a mix of high pass and low pass filters connected in series with no concern for how they interact. I believe there is more to learn down that path.
A stimulus used for testing band limited systems for level and phase distortion is the "sine squared pulse". https://iopscience.iop.org/article/10.1088/1742-6596/1195/1/012018/pdf It may well work for this. Happy Hunting.
Any amplifier can be modeled as a low pass filter (no amps I can concieve of have unlimited bandwidth). Most audio amps are bandass filters. While I think simulators quickly become mental masturbators, you can easily explore a "perfect" amp in your simulations to look for the dreaded "speed distortion". Just model a bandpass filter that matches a typical amp. in fact that would be a more appropriate reference path than a simple phase shift + delay. You should also be able to easily figure out what the requirements would be for a bandlimited source to have no "speed" distortion.
Acoustic music is usually captured with microphones that low pass around 15 KHz. The wider band mikes usually are noisier and not preferred. The real wideband mikes are far noisier (1" vs 1/4" disphragm area). Its very hard to justify testing with spectral content above 20 KHz.
However if you look at the whole chain from mike to speaker there are a mix of high pass and low pass filters connected in series with no concern for how they interact. I believe there is more to learn down that path.
A stimulus used for testing band limited systems for level and phase distortion is the "sine squared pulse". https://iopscience.iop.org/article/10.1088/1742-6596/1195/1/012018/pdf It may well work for this. Happy Hunting.
Only those who have a bear stepped on their ear can enjoy music from an amplifier with a bandwidth of 15 kHz. Happy listening!
Demian, this is exactly where Petr goes wrong for two reasons: 1) by comparing the output of an amp with an input signal with a BW that is by far exceeding the amp’s BW and 2) by neglecting the amp’s frequency dependent GD.
When subtracting signals, the only correct way would be to feed any appropriate signal to a distortion free bandpass filter with the exact BW and GD properties as from the DUT and compare its output to the DUT’s output.
Hans
How many times did I ask the “theorists” to take a burst from one period of a signal of any frequency in the audio range, pass it through a first-order low-pass filter of 100 kHz and show a section with a “cosmic slew rate”. I usually use 10kHz for both sine burst and triangular waveform.
So far, no "theorist" (or guru) has provided the result of such measurements.
It is done very simply, a triangle is selected in a suspicious area using a tangent to the signal (hypotenuse) and two legs (horizontal and vertical) so that dV / dt can be calculated.
Let me remind you that the slew rate SR is related to the total power bandwidth of the amplifier by the formula:
SR = 2πfmVm
Last edited:
Don't think so. The Schoeps MK4 capsule, perhaps the most popular cardioid for acoustic recordings, easily reaches 40kHz (with the proper head amp, CMC 6xt).
Yes, that‘s the practical limit, right at the startingpoint of any audio chain.
Hans, it's people like you who keep repeating the unsubstantiated mantra about infinite slew rate in bursts. Well, confirm it finally and the question will disappear by itself.
I have already cited data from psychoacoustics that in order to transmit a musical signal without quality loss, at least 18 harmonics must be transmitted. Moreover, each of them should be in its place and strictly its amplitude.
Even if we take the top note equal to 5 kHz, then the amplifier must be able to work without linear distortion in the band up to 90 kHz.
And if we take into account that the phase of the signal starts to spin an order of magnitude earlier, then the amplifier bandwidth should be at least 900 kHz.
Here you have the rationale on the fingers, as they say.
Only in this case can we expect that the amplifier will be able to convey the emotions that are embedded in the music.
It is hopeless.
Petr winds himself around like a eel.
The hard fact is that there are no 90 kHz.
Not even close.
40 kHz maybe with SACD and the very best microphones in a limited range of recordings.
Obviously he will never admit, he is wrong.
Lost time.
Lots have been written about the obvious shortcomings of the Oohashi test, that’s also violating the rule that no independen source could ever reproduce the test with statistical evidence.
The next step is probably pointing to the ever quoted Kunchur paper, who was thinking to have proven that our hearing system can detect 5usec time differences where in fact he proved that our hearing threshold is ca. 0.7 dB.
There is no single solid evidence whatsoever that human beings can hear, direct or indirect, anything above 20Khz despite all those whith different opinions.
Hans
The next step is probably pointing to the ever quoted Kunchur paper, who was thinking to have proven that our hearing system can detect 5usec time differences where in fact he proved that our hearing threshold is ca. 0.7 dB.
There is no single solid evidence whatsoever that human beings can hear, direct or indirect, anything above 20Khz despite all those whith different opinions.
Hans
The extended bandwidth Shoeps mikes seem to be directed to sound designers looking for effects, not for music recording. The specs all list 20 KHz for normal applications and you need a special preamp to get the extended bandwidth. And even if the sound source has ultrasonic components there are many ultrasonic sources around us we are completely insensitive to. In any case, while I am a fan of extended bandwidth, the reality is that it contributes little and is usually not present in recordings.
I just want to thank moderators for finally banning petr_2009 and his other accounts (Alnikst and NikLik). I’m sure I was not the only one reporting him.
This is a step in the right direction as he was ruining many threads and I’m afraid that tolerating his blatant breaching of rules that long, to some extent contributed that such valuable member as Jan Didden closed his account three days ago.
It’s a shame that Jan’s last posts had to be about this despicable pigheaded troll:
https://www.diyaudio.com/community/threads/bob-cordells-power-amplifier-book.171159/post-7268582
This is a step in the right direction as he was ruining many threads and I’m afraid that tolerating his blatant breaching of rules that long, to some extent contributed that such valuable member as Jan Didden closed his account three days ago.
It’s a shame that Jan’s last posts had to be about this despicable pigheaded troll:
https://www.diyaudio.com/community/threads/bob-cordells-power-amplifier-book.171159/post-7268582