-290 dB Distortion?

Status
Not open for further replies.
Syn08, just stop.

Anyway I've yet to receive any direct counterpoints or suggestions to my main concerns, I continuously have to repeat myself, and any time I make a point or a response it is almost never directly responded to even after I keep answering peoples questions. I can't give information if people are not clear about what they want and what they are missing. If I'm not giving you the information you are seeking then you need to be specific. So can I assume this is as far as we go? It seems like this thread has predictably degraded to the point of being nonconstructive. That was fairly disappointing. Is there no way to find the true spec of the system?
 
Last edited:
“That was fairly disappointing. Is there no way to find the true spec of the system?”

Yes there is.

Stop the elusive ‘I can’t tell you what I’m measuring, or even how I’m measuring’ nonsense. State clearly your circuit and measuring setup

The only thing that’s disappointing is someone who seeks help, declines to give relevant information (even when it should be obvious that his questions can’t be answered without it) and then admits to coming to a DIY website for free help on a commercial venture 🙄
 
Jan, you are proving my point. You keep suggesting I provide you with information, and when I give you context about the information I've provided you stay silent. For example when you say you want clearer graphs with more objective reference points and I say it is inherently relative and my graphs are effectively equivalent I cannot proceed unless you clarify why I'm wrong and correct my misunderstandings if there are any. I've repeating the same things over and over and over again with no direct correction.

Stop the elusive ‘I can’t tell you what I’m measuring, or even how I’m measuring’ nonsense. State clearly your circuit and measuring setup
I went into extreme detail about how I'm measuring it. If it is unsatisfactory to you you need to explain why.

Again, if I have to repeat myself a million times you either aren't reading my words or you aren't correcting them.
 
Last edited:
Anyone with an audio amplifier with -150dB distortion would stop, as it is already negligible compare to say speaker distortion. So this discussion about -290dB (or not) is, if you don't mind, academic. Even of the -290dB distortion is real, it is most probably buried in the background noise anyway.

The key of the argument hinges on whether the claimed -140dB error correction (assuming real) still applies when the distortion is very small, namely -150dB. If error correction means detecting the error and then applying the correction, that means that the detection has also to be better than -290dB in the first place. What you cannot detect, you cannot correct.

Would the detection functions at such low level and free of noise and error ? That is the answer you need to find out yourself, since you want to keep the EC secret. No point asking people here about how a black box might or might not function.

Measuring mm's and correcting for it is not so difficult. Measuring pm's and then proving that the right amount of correction is applied is not so trivial. I hope you can see the analogy.

Best wishes,
Patrick
 
Thank you, that is the kind of clear discussion and refutation I'm looking for.
Would the detection functions at such low level and free of noise and error ?
Hmmmm. I'm not sure. Since the signal is burried in with the noise I assumed the signal would be corrected along with the noise because the EC is referenced to ground. That may explain why it is so low noise even though there are high parts count. I guess the EC noise may overtake this assumption at least partially. It can be used on a non inverting amplifier but I'm currently correcting the mid point between the input and feedback resistor of the inverting amplifier. at least for part of the circuit, there are other points of correction but this would probably be the part that matters in this context.

I don't know I think I'm over my head on this front. If only the world wasn't full of dirtbags I'd just provide a schematic and ask.

Let me ask you this, can any theoretical system correct for such levels of distortion whether it is EC, high gain NFB, or otherwise, assuming the system had real world noise levels?
 
Last edited:
Jan, you are proving my point. You keep suggesting I provide you with information, and when I give you context about the information I've provided you stay silent..

I don't ask for context, I ask for clear, step by step procedure what you do, what you see. You ONLY give context, nothing of substance.

Why can't you show a clear graph of your amp distortion, showing -150dB or nothing above the noise? A loopback of your measurement system to show baseline? A step at the time instead jumping all around with things nobody can follow? Don't forget, you want to convince us, not the other way around. Try.

Jan
 
Last edited:
Hmm, well the results of the DM are showing the harmonics down to -150db. These are clear harmonics, not noise. How am I supposed to interpret this if the DM cannot work below the noise floor of the analyzer? The DM manual even says it will allow sound cards and such to reach -140+db measurement capability.

I assume you can do the calculations yourself, at .18Hz per bin you get about 7.5dB improvement so -187.5 dB is 1nV at the input. A poweramp at say 26dB of gain with an RTI noise of 4nV (you need to include the amp and all noise contributions) would have 80nV at the output this is a noise floor of -149.5dB. I would consider a PA with around 1nV of noise including the gain network extraordinary.

Maybe you should replace the wallwart and show us what you mean.
 
I don't ask for context, I ask for clear, step by step procedure what you do, what you see. You ONLY give context, nothing of substance. Why can't you show a clear graph of your amp distortion, showing -150dB or nothing above the noise? A loopback of your measurement system to show baseline? A step at the time instead jumping all around with things nobody can follow?

Jan
That's where we differ in opinion. As far as I know I provided excessive details about my measurements. I don't know what I am not providing you that I haven't already said.
Don't forget, you want to convince us, not the other way around.
Not correct. I can care less whether anyone believes whatever. I don't want fame or infamy or any of that nonsense. It's ridiculous to have blind belief in an unverified product anyway. I want to know if I can find an official spec I can use for presentation to investors, that's all. So more accurately I'm hoping for you to convince me in one way or the other, or at least point me in the direction of how I can find out.

I assume you can do the calculations yourself, at .18Hz per bin you get about 7.5dB improvement so -187.5 dB is 1nV at the input. A poweramp at say 26dB of gain with an RTI noise of 4nV (you need to include the amp and all noise contributions) would have 80nV at the output this is a noise floor of -149.5dB. I would consider a PA with around 1nV of noise including the gain network extraordinary.

Maybe you should replace the wallwart and show us what you mean.
Actually I don't know of such calculations, my electronics learning has been highly particular to specific goals. There are a lot of odd gaps in my knowledge. If I recall the maximum gain I had on the full scale version of the circuit was not more than 2. It was extremely difficult for me to get stable at a higher gain than this. There were seemingly large variations in stability based upon how it was compensated so I think it may be possible to find a much more optimal compensation configuration than the one I use but it's beyond my experience at the moment.
 
Last edited:
You cannot hope to correct for noise, neither with lots of NFB, nor EC. So in the end noise will limit your overall performance, even if you have infinite EC or NFB. Your detection itself is also not noise free.

And then you have time delay and stability issues on top.

AND people will tell you that lowest THD does not mean best sound. Else there'll only be opamps or Class D with tons of NFB. But then it might well be true in a few decades from now.

Or people will just listen to Alexa.

Good luck,
Patrick
 
Okay that makes me feel much better 🙂 If noise cannot be corrected by any means then clearly the noise level of the circuit is truly the limiting factor, in which case it is effectively pointless to make such an advanced version of my EC. Thank you EUVL 🙂 I guess I'll just make my new goal to ensure that none of my amplifiers distortion goes below -150db within spec. As far as sound, I know full well that THD is practically meaningless. Harmonically speaking I've found that as long as all of the harmonics are below -120db there is no audible difference in a power amplifier. Although the effective threshold is probably lower than that.
 
Last edited:
It's ridiculous to have blind belief in an unverified product anyway. I want to know if I can find an official spec I can use for presentation to investors, that's all. So more accurately I'm hoping for you to convince me in one way or the other, or at least point me in the direction of how I can find out.

And you think anyone will do that unless you first convince them that you actually have something of interest? You think I want to fall on my face when I talk to my investor friend?

Jan

If the title had been -140dB Error Correction instead of -290dB distortion, the discussions might have been less emotional.

Patrick

But we still would have asked for some proof.

Jan

Dear Jan,

The -290dB was nothing more than just a hypothesis.

He didn't mean any harm. Just needs some pointing in the right direction.

Cheers,
Patrick
 
If the title had been -140dB Error Correction instead of -290dB distortion, the discussions might have been less emotional.

Patrick
Unfortunately a mod named the thread instead of me. I would never have named it something so blatant and boastful.

Dear Jan,

The -290dB was nothing more than just a hypothesis.

He didn't mean any harm. Just needs some pointing in the right direction.

Cheers,
Patrick
Finally someone that understands me :cheers:
 
I think what is happening here is the amplifier and/or the measurement system is clipping - all those horrendous harmonics.

When the distortion magnifier is plugged in at its lowering the input (perhaps to just below clipping) on the QA401 and viola suddenly the distortion has gone away and it looks as though the amplifier has good performance.

This is clearly a very, very flawed set-up.

Step 1 (I think Syn08 said this) is to make sure you test with exactly the same levels going into the QA401 - if the levels are even slightly different, all bets are off.

Step 2 - by driving the amp into clipping to try to create a 'reference' you are doing nothing of the sort - you are just creating harmonic hash. To test if you circuit is really bringing benefit, run it at 6-10 dB below full output.

Step 3 - I get similar looking outputs from the QA401 when the output is overdriven or the input is overdriven. Again, make absolutely sure you are well within the input range of the QA401 - I suggest the peak input voltage does not exceed say 1.5 to 1.8 Volts

Step 4 - you are getting a lot of help from some very seriously competent people here with decades of experience and some PhD's thrown in for good measure. Keep quiet, listen and do the tests properly - like me, you may then learn a little bit from them.
 
Last edited:
Status
Not open for further replies.