Lars Clausen said:Differential tone Distortion.
Here are the results with 9 + 10 kHz, 4 Ohms load, and 10W power.
UCD400AD:
An externally hosted image should be here but it was not working when we last tested it.
ZAPpulse 2.3SE:
An externally hosted image should be here but it was not working when we last tested it.
ZAPpulse 2.3SE with simple THD optimization:
An externally hosted image should be here but it was not working when we last tested it.
As you can see the standard ZAPpulse 2.3SE has very high differential tone distortion compared to the neat results from UCD. However a small and simple modification will tune the amplifier towards lower distortion, and the result is a very low differential tone distortion. This optimizing is not the one proposed by Phase Accurate, it is a small change in the existing circuit.
We will test the mod for sound quality, and let you know if it also makes for a better sound. (never know ... ?)
Hello Lars,
Very nice results, is this with the same mod as those THD results that you posted in the other thread with Charles tweak? Or has this one additional tweaks as you suggest here (meaning more tweaks than the module for the other THD curves?)
Is this done with the low noise version of the 2.3Se or the normal version?
Best regards
Gertjan
Hi Lars,
congratulations on new results. It seems that properly implemented output filter (generously dimensioned discrete airgap inductor) is not source of distortion.
Did you measure idle current on UcD modules in factory standard and "tweaked" setting? According to Jan-Peter and Bruno factory setting should be 60-70mA. I wonder how much it was in "tweaked" setting ?
Best regards,
Jaka Racman
ritchie00boy: If there had been anything at 1 kHz, you could easily see it, the fft spikes are same width per Hz, and so at 1 kHz they are very wide. I think it's a 4096 point fft.
Ghemik: The graphs with Phase Accurate mods, also have the other THD optimization. But still Charles' circuit reduces THD. Of course not from 1.5% to 0.015%, most of it comes from the other mod. I will release it when i have made listening tests. I don't design for low THD, i design for good sound. If those are in conflict, i take good sound over low THD. (As you know )
Jaka Racman: I used the factory setting which was left on one module. Then i took another module and adjusted it to minimum THD at 10W / 4 Ohm / 1kHz. Then the THD went down quite a bit, over most of the frequency range, and started to resemble what was shown on the graph. Unfortunately the module got very hot after a few minutes, so i am afraid to run too much in that mode. I didn't measure the Idle current though. My guess is maybe 200 - 250 mA. ??
The software i am using can be downloaded here so if you have a good soundcard on your computer ( i use ESI Juli@ ) and a few cables, load resistor, and so on, you can make the measurement yourself.
I have an AES17 filter, but on the ESI Juli@ the antialias filter on the card is enough, so that you don't need an AES17 filter for most purposes.
Ghemik: The graphs with Phase Accurate mods, also have the other THD optimization. But still Charles' circuit reduces THD. Of course not from 1.5% to 0.015%, most of it comes from the other mod. I will release it when i have made listening tests. I don't design for low THD, i design for good sound. If those are in conflict, i take good sound over low THD. (As you know
)Jaka Racman: I used the factory setting which was left on one module. Then i took another module and adjusted it to minimum THD at 10W / 4 Ohm / 1kHz. Then the THD went down quite a bit, over most of the frequency range, and started to resemble what was shown on the graph. Unfortunately the module got very hot after a few minutes, so i am afraid to run too much in that mode. I didn't measure the Idle current though. My guess is maybe 200 - 250 mA. ??
The software i am using can be downloaded here so if you have a good soundcard on your computer ( i use ESI Juli@ ) and a few cables, load resistor, and so on, you can make the measurement yourself.
I have an AES17 filter, but on the ESI Juli@ the antialias filter on the card is enough, so that you don't need an AES17 filter for most purposes.
Test software
Lars,
Thanks for the software link. I think their interpretation is very optimistic with the bottom being -130db. Most sound cards have 110db capabilities or less so I would have to question this, looks very well done otherwise.
Really nice data on your new mod! Looking forward to your report on the sonics.
Roger
Lars,
Thanks for the software link. I think their interpretation is very optimistic with the bottom being -130db. Most sound cards have 110db capabilities or less so I would have to question this, looks very well done otherwise.
Really nice data on your new mod! Looking forward to your report on the sonics.
Roger
sx88: So right! And when you use the program with a normal soundcard, the noisefloor goes up significantly.
But still the -110 - -115dB of the sound card data is the total dynamic range (of all audio band) . Still you can have lower signals than that, when you divide them into discrete frequencies,
as is done in the FFT analyzer.
But still the -110 - -115dB of the sound card data is the total dynamic range (of all audio band) . Still you can have lower signals than that, when you divide them into discrete frequencies,
as is done in the FFT analyzer.
Hello,
Hereby our results of the 9+10kHz measurement of the both amps.
http://www.hypex.nl/classd/imd zap ucd 10w 2w.pdf
Measurement is done on the Audio Precision System 2.
I just picked up one of the UcD400 modules what were lying on the workbench and measure them on the AP. Nothing special....
Perhaps time to invest in an AP system....
Jan-Peter
Hereby our results of the 9+10kHz measurement of the both amps.
http://www.hypex.nl/classd/imd zap ucd 10w 2w.pdf
Measurement is done on the Audio Precision System 2.
I just picked up one of the UcD400 modules what were lying on the workbench and measure them on the AP. Nothing special....
Perhaps time to invest in an AP system....
Jan-Peter
In fact the THD performance of some of the better sound cards is good enough to make them usable for many measurement purposes. The problem is that THD performance is not enough. Sound cards have fixed input and output levels. If you want to build an audio analyser around a sound card you need to have calibrated gain/attenuator stages at both the inputs and outputs. This is the only way to get repeatable, absolute level settings. Floating outputs and differential inputs are helpful too, but not critical.Jan-Peter said:Perhaps time to invest in an AP system....
Trying to work around the level problem by adding manual attenuators on the inputs and outputs is very error-prone and can seriously invalidate a measurement, much more so than the THD of the sound card itself.
The Audio Precision ATS2 is such a device. It's little more than a pair of decent AD/DA chips coupled to the same output and input electronics as the SYS2. I can wholeheartedly recommend it to anyone who does not need the full SYS2 package but still wants to make trustworthy and repeatable measurements.
Lars, your THD mod is quite intriguing. Of course I'd like to test it myself, but I've got a more practical idea. I could send you a .wav file with test signals, which you could play through the amp, and record and return the result. If I sent you such a file, would you do the test?
Jan-Peter said:Hello,
Hereby our results of the 9+10kHz measurement of the both amps.
http://www.hypex.nl/classd/imd zap ucd 10w 2w.pdf
Measurement is done on the Audio Precision System 2.
I just picked up one of the UcD400 modules what were lying on the workbench and measure them on the AP. Nothing special....
Perhaps time to invest in an AP system....
Jan-Peter
With Lars's latest tweak, his results maybe better. Quite amazing is the difference in noise level, in the order of 18dB, I guess this is he normal 2.3SE version, not the low noise version, correct? The relatively high noise level of the ZAP was also a reason why I choose Ucd, with the standard ZAP, the noise level was too high to drive tweeters directly.
Best regards
Gertjan
Hello Lars,
In this post http://www.diyaudio.com/forums/showthread.php?postid=677977#post677977
You show us your measurements.
However these mesaurements shows a big difference with our mesaurements, http://www.hypex.nl/classd/imd zap ucd 10w 2w.pdf
Can you confirm that youhave measured correct?
Regards,
Jan-Peter
In this post http://www.diyaudio.com/forums/showthread.php?postid=677977#post677977
You show us your measurements.
However these mesaurements shows a big difference with our mesaurements, http://www.hypex.nl/classd/imd zap ucd 10w 2w.pdf
Can you confirm that youhave measured correct?
Regards,
Jan-Peter
Jan Peter: It's quite possible we are measuring apples against bananas
Here is what i did:
And the results were as you see.
About level calibration, You are right, if you really want to go into low THD figures with a soundcard, you have to hit the right level for the A/D converters. This i have solved in my breakout box, with an auto levelcontrol. And i have made sure the THD of the levelcontrol is well below that of the rest of the tester, so no matter which setting it is in, it will not affect the outcome of the test.
Anyway for this paticular test, it was not needed, since all the input levels for the A/D were the same.
As you can see on the 9 and 10 kHz peaks, all three measurements are made at the exact same level.
Now i am curious to see what your test was?
All the best from Lars
Here is what i did:
An externally hosted image should be here but it was not working when we last tested it.
And the results were as you see.
About level calibration, You are right, if you really want to go into low THD figures with a soundcard, you have to hit the right level for the A/D converters. This i have solved in my breakout box, with an auto levelcontrol. And i have made sure the THD of the levelcontrol is well below that of the rest of the tester, so no matter which setting it is in, it will not affect the outcome of the test.
Anyway for this paticular test, it was not needed, since all the input levels for the A/D were the same.
As you can see on the 9 and 10 kHz peaks, all three measurements are made at the exact same level.
Now i am curious to see what your test was?
All the best from Lars
I have used a Turtle Beach sound card and the problem is calibration. I have very little control over the input-output circuits.Bruno Putzeys said:
With the measurement below I get different levels of the curves and they are less than one step in the attentuation in the software(?) or the hardware(?).
http://www.sjostromaudio.com/hifi_files/qrv/qrv07_testreport/QRV07_HD545.htm
But... we do agree that your setup does not yield the same outcome as the AP2. Well I know which one I trust best.
Seriously, check how you measured the 6.3Vrms output level. I suspect the meter was so wideband that it measured the switching ripple as well, leading to a higher reading than the 9+10kHz tones alone. Just to be sure, you could also snap a plot of the signal at the input of the DUT to be sure there are no surprises there, but most probably it's the level reading which is at issue. In our setup, level is measured after the first half of the AES17 filter, insuring only the relevant signal is measured.
In another thread you are showing plots suggesting that Charles' mod improves your THD figures by a whopping 40dB at 7kHz. We understand you like to be seen showing nice THD plots as well (the "bad for sound" story having proven to be less than convincing), but 40dB is such a bold claim to make as to render it highly suspicious to put it mildly. As you may have understood by now, we know our maths inside-out here, and we can assure you that the simple feedback tweak does not improve loop gain by much more than 10dB, a far cry from 40dB. However, we do know how to make such good THD (and IMD) plots on the Zap modules, where the extra feedback loop allows a change in test conditions that would not work well on the unmodified module (but one that is not very representative of normal operation either). Out of courtesy we are not making this method public.
To put it simply: if you had claimed an improvement of 15dB, we'd have thought "OK Lars is making progress" but a sudden 40dB "improvement", especially using the simplest of circuit mods, is imho a dead giveaway for quite a different kind of move. We would like to advise you that such manoeuvres are not good for your personal reputation (much worse than your products having a reputation problem). For us, to keep engaging in discussions with you in these circumstances is also not a good idea, so if these are the conditions under which we are to have a discussion with you, we'd rather not, thank you very much.
Jan-Peter & Bruno
Seriously, check how you measured the 6.3Vrms output level. I suspect the meter was so wideband that it measured the switching ripple as well, leading to a higher reading than the 9+10kHz tones alone. Just to be sure, you could also snap a plot of the signal at the input of the DUT to be sure there are no surprises there, but most probably it's the level reading which is at issue. In our setup, level is measured after the first half of the AES17 filter, insuring only the relevant signal is measured.
In another thread you are showing plots suggesting that Charles' mod improves your THD figures by a whopping 40dB at 7kHz. We understand you like to be seen showing nice THD plots as well (the "bad for sound" story having proven to be less than convincing), but 40dB is such a bold claim to make as to render it highly suspicious to put it mildly. As you may have understood by now, we know our maths inside-out here, and we can assure you that the simple feedback tweak does not improve loop gain by much more than 10dB, a far cry from 40dB. However, we do know how to make such good THD (and IMD) plots on the Zap modules, where the extra feedback loop allows a change in test conditions that would not work well on the unmodified module (but one that is not very representative of normal operation either). Out of courtesy we are not making this method public.
To put it simply: if you had claimed an improvement of 15dB, we'd have thought "OK Lars is making progress" but a sudden 40dB "improvement", especially using the simplest of circuit mods, is imho a dead giveaway for quite a different kind of move. We would like to advise you that such manoeuvres are not good for your personal reputation (much worse than your products having a reputation problem). For us, to keep engaging in discussions with you in these circumstances is also not a good idea, so if these are the conditions under which we are to have a discussion with you, we'd rather not, thank you very much.
Jan-Peter & Bruno
Whoaa hold your horses Jan Peter!
If you read what that thread actually says, it says exactly that it is not only because of charles' circuit, but also another (rather simple) change on the module.Not at all! Nothing changed here,
i am just demonstrating that i can easily reduce the THD if i find it to be beneficial. However until now i have not found it to be beneficial for the sound, to go for the very low thd. That's also why (as you can read in that other thread) i am not releasing the simple thd optimizing mod, until i have found out if it will improve the sound or not.'First Half' Do you mean the 24 kHz 'brick wall' filter? I also used that one for my test.
It will probably not 'improve the loop gain' at all
No reason to get angry Jan-Peter, i think you have misunderstood my postings, as you have obviously not read them thoroughly.
Thanks for the personal advice, but please try to focus on kicking the ball, not the player
My postings are not maneuvres, but simply the measured results. You are right i am not using an AP2, like you, but i have been able to make many of the exact same results with this lower cost system. Just except those with THD vs. freq in 4 Ohms / 10W on your UCD400 with factory setting. In my test your amp got 0.01 to 0.03% THD (4-10 times higher than your result) depending on freq, and the 6 kHz peak of my amplifier didn't go to 2%, but only to 0.5%. (4 times lower than your result).On the other hand i have no good explanation for the rather big difference in what i can measure ct. what you can measure with the 9/10 kHz test. I think my test is reliable, even if does not have an AP logo in the corner.
I guess someone 'neutral' should make the measurements instead ... ?
Well, it's wholly up to you what you want to claim or publish. Frankly it's not our business. Nevertheless we are getting very curious as to how you're getting such numbers - or at least to see the miracle happen under our own eyes (and on an AP test set).
I'm aware that you've been saying in this thread that the mod was not Charles' tweak, but in the thread about Charles' tweak you are also posting plots showing 40dB reduced THD, supposedly purely through use of that tweak... Hence the loop gain remark. Of course, if your "low THD tweak" is not about loop gain, it must have been about speeding up switching a hundredfold. Take it the way you want it, there is no simple magic tweak that can do any of this, either way we're looking at a significant redesign. I've been in this business for too long to believe in magic.
We're all for having modules tested by a third party. If this third party is equally well equipped like us (ie. at least a recently calibrated ATS2), we have nothing against sending them some modules for them to test. Likewise you could send them unmodified and tweaked Zap modules. Especially the latter, since even your own measurements show the standard Zap modules to distort quite a bit more than UcD, so it's really the magic low THD modules that are of interest. If you insist on drawing it out this way, so be it and we'll have to find an uninterested 3rd party that we can agree on.
Yet, I still don't think there's much to be gained from this discussion. So far a large portion of the thread has been me explaining basic loop control theory to rectify the notion of "output filter delay", basic stuff really. If the next month will be about "how to properly conduct audio performance testing" this thread is turning into a soap opera. This is not something we can afford to spend much more time on, let's be frank about this. Honestly I hope you too are spending more time selling your modules than slugging it out verbally with other designers.
I think we both have made our positions clear, both with respect to sonic philosophy and design method. In listening we both quite strongly prefer our own modules over the other's so that's fine too. There are customers for neutral and for coloured amps so neither of us is going hungry. Neither does one party want the other to go hungry (or thirsty, to follow up on one of your own comments).
So there it is. If we must, we can mobilise a 3rd party. But preferably I'd call it quits because I've got work to do. It's up to you.
I'm aware that you've been saying in this thread that the mod was not Charles' tweak, but in the thread about Charles' tweak you are also posting plots showing 40dB reduced THD, supposedly purely through use of that tweak... Hence the loop gain remark. Of course, if your "low THD tweak" is not about loop gain, it must have been about speeding up switching a hundredfold. Take it the way you want it, there is no simple magic tweak that can do any of this, either way we're looking at a significant redesign. I've been in this business for too long to believe in magic.
We're all for having modules tested by a third party. If this third party is equally well equipped like us (ie. at least a recently calibrated ATS2), we have nothing against sending them some modules for them to test. Likewise you could send them unmodified and tweaked Zap modules. Especially the latter, since even your own measurements show the standard Zap modules to distort quite a bit more than UcD, so it's really the magic low THD modules that are of interest. If you insist on drawing it out this way, so be it and we'll have to find an uninterested 3rd party that we can agree on.
Yet, I still don't think there's much to be gained from this discussion. So far a large portion of the thread has been me explaining basic loop control theory to rectify the notion of "output filter delay", basic stuff really. If the next month will be about "how to properly conduct audio performance testing" this thread is turning into a soap opera. This is not something we can afford to spend much more time on, let's be frank about this. Honestly I hope you too are spending more time selling your modules than slugging it out verbally with other designers.
I think we both have made our positions clear, both with respect to sonic philosophy and design method. In listening we both quite strongly prefer our own modules over the other's so that's fine too. There are customers for neutral and for coloured amps so neither of us is going hungry. Neither does one party want the other to go hungry (or thirsty, to follow up on one of your own comments).
So there it is. If we must, we can mobilise a 3rd party. But preferably I'd call it quits because I've got work to do. It's up to you.
Bruno: I agree with you completely, so i hope we can agree to stop throwing curves and tables at each other anymore. The summer is too short (at least here in Denmark). Better spend it in company with a cold beer and some nice looking girls
Finally i will just comment on your last post, the special mod is - like you suggest - simply about shortening switching times. So no magic is involved
Finally i will just comment on your last post, the special mod is - like you suggest - simply about shortening switching times. So no magic is involved
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.