Averaging does not lower the noise floor. Synchronous averaging does.
Not that I expect you to understand the difference 😀.
I think there was an article about resistor distortion in Linear Audio many years ago and there is Cherry's AES paper about magnetic coupling of distorted output transistor currents to the amplifier input. I haven't got access to those at the moment, but I'll look them up when I have the chance.
Edit: this is the Linear Audio article:
Linear Audio | your tech audio resource
By the way, see post 245 of this thread for a way to minimize it.
Last edited:
If I look at these cases, the impression I get is as follows. In the first few cases, I see an EC which lowers the distortion up to about 60dB. That in itself is quite a feat, but it is at 1kHz. How about 10kHz?
The later cases seem to run into the floor of the QA. There may or may not be anything down there, hard to say.
It would be of interest if you could find someone with an APx555 in your area, or even with an AP SYS2722, to do measurements. I know a guy in Germany who does this on an APx55 for $ 200 but that's probably not realistic.
The thing is, we all are sure that -290dB is pie in the sky, and it's not very useful for us to keep on banging on that ad nauseatum. But if you could convincingly show a reduction of 60dB over the audio band, that would put you into -150dB territory and earn you lots of respect for sure.
So how can we do that, without you disclosing the circuit? Are you able to travel with an implementation in a black box to somewhere to measure?
Jan
I think there’s so much going on here, it’s hard to try to understand what you’ve done. The high harmonics in the plot don’t look like normal distortion. It looks like the amp has clipped, or the QA401 is overloaded.
Why not redo the tests with s block diagram of what/ how you’ve connected it.
I would start with the EC connected and the best reading you can get (normal operation)
Then insert the noise gain resistor again making sure nothing is clipping.
Use the same scales/sampling etc as the previous test.
With these two readings, you can then calculate the EC effectiveness by looking at the reduction in loop gain, ergo increase in distortion a la Gronner.
Consider this thread a ‘Dragons Den’ schooling. If you can get it past these guys, you’ll be pretty good.
I think you biggest problem has been your less than rigorous description of your methodology and inconsistent test regimes (take this this as advice - not a jab in the ribs).
Anyway, if you intend to find investors, they’ll have their guys grill you as well. Probably worse.
I recall 10khz tests always being below noise floor. The analyzer doesn't have any inherent distortion above the noise floor at 10khz that I recall so it just looks flat. But the last time I did a 10khz test was a long time ago and I don't know if I ever tested it on a lesser version of my EC to show relative improvements. Most of my serious tests have been at 1khz. I will do some 10khz tests when I get the time.
So my problem with that suggestion is I don't know how they are going to measure it. I could make each level of my EC in a separate black box but I can't measure it if the main amplifier is good to any real degree, which is the entire reason I made the main amplifier distort to begin with.
So what sort of test circuit would I bring exactly? A simple breadboard opamp example circuit with the full scale circuit? A full scale power amplifier on a PCB?
Yes, I described many times that I push the amp into clipping and then apply the EC in order to correct the clipping. The EC cannot correct for clipping above the rails so I make it go very very close to absolute clipping instead of true clipping. This makes the output stage in the main amplifier distort highly.
Well, the circuit shown in the resistor test is the circuit I've been using the entire time, only without the resistor in my method and with a higher output voltage in my method. I'm willing to do whatever tests in whatever way that people want me to do, but I'm too busy to do it for another week or two.
I did these things as shown in the resistor method results. Although I did not post the EC with the opamp functioning normally, at least not that I recall, because it will never show any distortion unless the amplifier is artificially distorting already, which is the whole problem.
Okay so, I'll explain it once more. This is the my test circuit in all of these tests
The only difference between my method and the resistor method is that I removed the 10 ohm resistor and applied a greater output voltage in order to make the amplifier distort.
Last edited:
That's not your problem. You make a box with your best circuit inside with only the inputs and output accessible (and power supply if not internal). Then get it measured.
If the best minds and equipment on this planet cannot see any distortion above -200dB, Bob's your uncle.
If they measure -118dB, you've failed. Simple as that. They don't have to know what's inside, you don't have to know how they measure.
I would give it a shot myself but you're on the wrong side of the pond ;-)
Jan
Dude I would fly over to you if I could finally resolve this issue.
So I think what I'll do then is make a not too expensive PCB version of my best circuit and enclose it.
Do I want to have it be high power? Is using an opamp as a main amplifier acceptable?
Does it matter?
So I think what I'll do then is make a not too expensive PCB version of my best circuit and enclose it.
Do I want to have it be high power? Is using an opamp as a main amplifier acceptable?
Does it matter?
“I described many times that I push the amp into clipping and then apply the EC in order to correct the clipping”
Be careful. Is the fundamental at exactly the same level it was before you applied EC? This is important because if it is reduced the amp comes out of clipping and that’s not the EC working.
Naturally, all your readings should be at some absolute reference like dBV or dBU.
Be careful. Is the fundamental at exactly the same level it was before you applied EC? This is important because if it is reduced the amp comes out of clipping and that’s not the EC working.
Naturally, all your readings should be at some absolute reference like dBV or dBU.
Yeah, they are all exactly at the same point within each test. I measure the circuit without the EC first. I leave everything the same and apply the EC and re-measure. On the last test I did using my method I even let the circuit burn in for exactly 5 minutes on each test in order to get exact relative measurements.
You can see on the screenshot where the fundamental is marked as "M0", it's above the 0db point but still marked. Also the display shows the relative voltages between tests where it says "Peak R" assuming I averaged them in the analyzer for the same period of time. I only averaged them for the same period in the last test though, in order to show that the levels were exactly the same without screenshotting my scope as well because a member had an issue with that.
Also I showed in those tests that even if I back off the output voltage to 4.32V instead of 12.9v the main amplifier still distorts without the EC so a small discrepancy wouldn't matter anyway. I'm pushing the amplifier way beyond its intended limits in the tests using my method. The absolute values shown for "Peak R" aren't correct though, they are different from the scope. I specified the exact output voltages on each test. I'm only repeating things I've already said in those posts I linked. It's all there just read my words carefully.
You can see on the screenshot where the fundamental is marked as "M0", it's above the 0db point but still marked. Also the display shows the relative voltages between tests where it says "Peak R" assuming I averaged them in the analyzer for the same period of time. I only averaged them for the same period in the last test though, in order to show that the levels were exactly the same without screenshotting my scope as well because a member had an issue with that.
Also I showed in those tests that even if I back off the output voltage to 4.32V instead of 12.9v the main amplifier still distorts without the EC so a small discrepancy wouldn't matter anyway. I'm pushing the amplifier way beyond its intended limits in the tests using my method. The absolute values shown for "Peak R" aren't correct though, they are different from the scope. I specified the exact output voltages on each test. I'm only repeating things I've already said in those posts I linked. It's all there just read my words carefully.
Last edited:
You should make it such that it clearly shows the effect of the EC and the lowest distortion you can get. It's the principle that counts, if it is established that that is sound, it can be applied to other circuits. And if you take it somewhere, no need to pot it; nobody will find out how the smart EC works from just looking at a full PCB!
Jan
So if I made the circuit with the intention of showing the improvement of the EC itself then I would have to make sure the main amplifier is distorting and then show an equivalent circuit without the EC. In this case I won't be able to show the full performance.
Or, I can just make it as good as I can and let the person apply different loads to it as they see fit.
So I'm guessing you mean option 2?
Or maybe I combine option 1 and 2 and give them an equivalent main amplifier without EC as well as another main amplifier with EC for comparisons. Or maybe I can have a PCB that contains each stage of sophistication of my EC in order to show the incremental improvement as well.
Or, I can just make it as good as I can and let the person apply different loads to it as they see fit.
So I'm guessing you mean option 2?
Or maybe I combine option 1 and 2 and give them an equivalent main amplifier without EC as well as another main amplifier with EC for comparisons. Or maybe I can have a PCB that contains each stage of sophistication of my EC in order to show the incremental improvement as well.
Last edited:
I think this is where you need to take some decisions on your own. The overall aim is to convincingly show that you are on to something. We have no idea how your stuff works, so difficult to advise you here. Maybe a version where the EC can be switched off and on without any other changes.
Jan
Jan
I question whether I will be able to maximize performance of either circuit this way. I want to keep traces as short as possible. I think what I'll do is just use my LT1364 opamp as a main amplifier since I have a bunch lying around and they seem to be good for this test so far. I'll make a quality test PCB for it and have an increasingly potent version of my EC from left to right of the PCB starting from the base main amplifier and ending at my best version of EC.
I doubt you will find someone to work on a hidden wonder.
This is too much known as a basis of scams. We had color rays that cure cancer, airplanes that detect oil deposits.
This is too much known as a basis of scams. We had color rays that cure cancer, airplanes that detect oil deposits.
Is the QA401 Melt Noise mode synchronous averaging? I don't own one but I thought it was based on a post I had read.
I didn't look so I don't know.
EDIT - There was a mention of an experimental menu I found maybe someone can help the OP use it? Though there is still the THD of the source and the posted plots didn't look like it would be enough help..
EDIT - There was a mention of an experimental menu I found maybe someone can help the OP use it? Though there is still the THD of the source and the posted plots didn't look like it would be enough help..
Last edited:
- Status
- Not open for further replies.