I can sample in real time up to 10Gs/sec, that is 500,000 samples per 20KHz cycle, would that be good enough for our dear petr_2009? 
I would think it's time for him to yet again shift his moving target to nowhere.
Hans, don't hold your breath waiting for instructions, we still need you around
.
I would think it's time for him to yet again shift his moving target to nowhere.
Hans, don't hold your breath waiting for instructions, we still need you around
.Speed (velocity) distortion occurs at any deviation of the signal from the sinusoid of a given frequency. In this case, the burst is passed through a 100 kHz low-pass filter.
Hans, to begin with, measure the second error of your amplifier at a frequency of 10 kHz, otherwise some authors with 30 years of experience strained with this
"Now, please post the link to the article with the noise intermodulation test, the one where the tube amp got a high score. That's interesting. "
журнал "Радио" 1999-04 page 29
Hans, to begin with, measure the second error of your amplifier at a frequency of 10 kHz, otherwise some authors with 30 years of experience strained with this
"Now, please post the link to the article with the noise intermodulation test, the one where the tube amp got a high score. That's interesting. "
журнал "Радио" 1999-04 page 29
Attachments
Last edited:
nonlinear distortion...
linear "distortion", 5 times over audioband.. Speaker box or your ears are also "LP filter"..
Why is it interesting? You are superbat or dolphin?
But in simulation it is no problem even to 100MHz Delay and LP filtering ( way over audioband) are not distortions
Last edited:
I’m out of popcorn
This is …. a pinnacle!
Petr, in the whole history of your trolling at diyaudio, you have never ever shown any your measurement. Just simulations.
Do you know that simulations and real circuit results could be very different? Do you?
In your "native" language: 당신은 당신의 완전한 무지에 한심합니다
This is …. a pinnacle!
Petr, in the whole history of your trolling at diyaudio, you have never ever shown any your measurement. Just simulations.
Do you know that simulations and real circuit results could be very different? Do you?
In your "native"
language: 당신은 당신의 완전한 무지에 한심합니다
Here's the link
Стр. 29 журнала <<Радио>> № 4 за 1999 год
Machine translation
What he says about signal probability density is interesting, it's similar to GedLee metric. Also similar to what I've been doing in the output stage measurement topic.
However, it seems the amps clipped during measurements, which means the results measure clipping behavior more than any other stuff. That's sad, because in normal hifi use an amp should rarely (if ever) clip. The technique is interesting though. I might try it.
Стр. 29 журнала <<Радио>> № 4 за 1999 год
Machine translation
What he says about signal probability density is interesting, it's similar to GedLee metric. Also similar to what I've been doing in the output stage measurement topic.
However, it seems the amps clipped during measurements, which means the results measure clipping behavior more than any other stuff. That's sad, because in normal hifi use an amp should rarely (if ever) clip. The technique is interesting though. I might try it.
I used the soundcard to acquire LRCK at the SPDIF transmitter and LRCK at the SPDIF receiver. Phase shift between these two gives the delay (and its variations known as "jitter") introduced by the SPDIF chain. Then I used a divider to do the same with sender MCLK vs receiver recovered MCLK. It was around 25MHz, so that's 5x more than the 5MHz you asked
1ps is 0.000017° phase shift on a 48kHz signal, which is measurable with a soundcard if noise is low enough.
For the group delay of an amp I would use the network analyzer, but that's not an interesting measurement: open loop phase shift is much more interesting for stability.
The Bode100 in my soundcard-based lab would measure group delay directly, but I have never bothered to measure it in amplifiers. With electronics-based group delays several orders of magnitude below that of the acoustical part of the reproduction chain, who would bother?
Jan
would measure group delay directly, but I have never bothered to measure it in amplifiers. With electronics-based group delays several orders of magnitude below that of the acoustical part of the reproduction chain, who would bother?Jan
Attachments
Last edited:
Petr,
I need a more precise definition of what to do that really makes sense.
It is easy to generate a three cycle sine wave at 10Khz with 1000 point per cycle and offering this to an amp, see image below.
And also subtracting input from output after gain correction is easy to perform, although rather questionable other than to show that there is a difference between the two.
But then comes the third step which is subtracting the input signal from the (gain + delay) corrected output signal.
And here comes my question:
Do you use a fixed value for this time delay, then it makes no sense at all because GD depends on frequency.
Especially at the start and end of the 3 cycle sinewave, there is a wild spread of frequencies, so you will have to restrict the input signal to a frequency band where the Amp's GD is constant.
Not doing that will give an outcome that can't be used for any prove of distortion.
Hans
P.S. I asked for a link or a copy of the test in #632, but none of both was given.
I just saw that PeuFeu gave the link.
.
Attachments
Last edited:
No problem at all with the VNA software using the same Picoscope.
Everything up to 100Mhz will be a hit of a button, but what's the point, why are you asking ?
Is this to have an excuse why you are never measuring anything at all ?
Hans
Thx for supplying the link + translation.
The clipping issue is rather masking the results, however I'm anxious to see any results from your side.
Hans
Apparently it is worth repeating again - you have function breaks in the places you are talking about. And at the same time, a signal with a wide spectrum is received at the input of the amplifier, which is not limited by anything, since you do not put additional filters in the simulator.
Explain how noise-like signals differ from multi-tone ones in your opinion? The waveforms of a multi-tone signal were given above in the topic and it looks quite noise-like. I know the answer, but I want to hear your version so that we can compare where the difference is hidden.
Note that even in the Syritso article, the white noise signal was limited to a bandpass filter. And in the same way, the obtained difference components were usually filtered to isolate the part of the signal that falls into the sound range (or, in general, into the operating range of the device under test). Thus, no one tries to evaluate the results in the entire spectral axis, since this is not necessary.
No, I don't think so. I have not tried to prick you personally anywhere, but you persistently turn to personalities and express your hostile attitude. My questions and theses are strictly within the framework of the topic discussed here. If you are not satisfied with my answers to your personal attacks, then maybe you should not make them? Behave with dignity, including with the rest of the forum participants, and then the communication will be much more neutral and even pleasant, despite the possible previous negative experience.
Yes, it's pretty sad.
I think I will try it if I have time. My "two sinewaves" test can pinpoint a trouble spot according to output current, so if something happens when one device turns off, there is a blip on the curve at exactly that spot. That's very useful. But the drawback is that noise floor is pretty high because it is equivalent to stepping DC voltage and doing a FFT run for each point, which means the number of samples in each bin has to be low otherwise it takes forever.
The noise test from the article has the opposite advantages and drawbacks.
I like the fact that its signal spends more time around zero, just like normal music. I tweaked my signal from the beginning to do that also.
BTW We have also touched upon acoustic measurements using a sound card and REW. Just found out that there is a TI app note describing just that: https://www.ti.com/lit/an/slaa641/slaa641.pdf
They use my favorite, a Scarlet Focusrite 2i2 in some of the measurements. To drive the speaker they use a TPA3116D2 eval module.
Worth a read.
Jan
They use my favorite, a Scarlet Focusrite 2i2 in some of the measurements. To drive the speaker they use a TPA3116D2 eval module.
Worth a read.
Jan
Last edited:
Just out of curiosity I did both a Sim and a real measurement of my Amplifier's GD from 5 Khz to 5 Mhz.
It's obvious that those two curves are not the same, so simulations are helpful but not always perfect.
GD is apparently constant up to just ca. 150Khz.
So subtracting a (gain+10Khz delay) corrected output from a sudden starting and stopping input signal, that's going way beyond 150Khz, will produce the sort of non-sense pictures with so called "Speed Errors".
Not shown here, but after offering a 100Khz limited input signal composed of just three 10Khz cycles and then subtracting it from time and gain corrected output, resulted as expected in a flat line !
Hans
.
Attachments
