I have voted 
I think a real problem with this kind of test is simply that (to me at any rate) they are essentially all audibly the same.
If you took for example one of Pavels power amps and say Tom's LM3886 design then I'm certain they would be subjectively very different, even though the specs are beyond reproach. Those differences don't come across in the same way with small signal stuff in my experience.
So of these four indistinguishable opamps (lol) and in no particular order I find:
One of them is a little lifeless, another somewhat constrained, there is one that I keep coming back to as being the odd one out and one that I would choose as a preference.
If Pavel wishes then I will say in the thread which description applies to which.
I keep wondering over the source material. Its bloomin' good if its what it appears to be, or is it something else with a little niceness added not that it matters for the test as its a constant throughout.
(must remember to put Rice Krispies on the shopping list)
I think a real problem with this kind of test is simply that (to me at any rate) they are essentially all audibly the same.
If you took for example one of Pavels power amps and say Tom's LM3886 design then I'm certain they would be subjectively very different, even though the specs are beyond reproach. Those differences don't come across in the same way with small signal stuff in my experience.
So of these four indistinguishable opamps (lol) and in no particular order I find:
One of them is a little lifeless, another somewhat constrained, there is one that I keep coming back to as being the odd one out and one that I would choose as a preference.
If Pavel wishes then I will say in the thread which description applies to which.
I keep wondering over the source material. Its bloomin' good if its what it appears to be, or is it something else with a little niceness added
not that it matters for the test as its a constant throughout.(must remember to put Rice Krispies on the shopping list
)
Hi Karl,
thanks so much for your comments. Very interesting!
I would prefer if you did not assign your sound impressions to the file names at the moment. The poll is valid until Friday June 16, and though I do not expect to collect many more votes, your opinion linked to files might have had influence to the future voters. Of course, I will be very happy if you tell your opinion to the files in names after the poll is finished.
I agree with you that it is very difficult to tell the difference in this test. On the other hand, the choice of the opamps used is, IMO, very interesting and might be quite surprising for the voters.
Cheers,
Pavel
thanks so much for your comments. Very interesting!
I would prefer if you did not assign your sound impressions to the file names at the moment. The poll is valid until Friday June 16, and though I do not expect to collect many more votes, your opinion linked to files might have had influence to the future voters. Of course, I will be very happy if you tell your opinion to the files in names after the poll is finished.
I agree with you that it is very difficult to tell the difference in this test. On the other hand, the choice of the opamps used is, IMO, very interesting and might be quite surprising for the voters.
Cheers,
Pavel
Okay, understood. How about posting the source file, and maybe the source file only passing through your data converters, once the survey period has closed?
One reason I wonder about such things is because, suppose the source was a bit muffled sounding, and therefore people might prefer an opamp circuit that added a little brightness in the form of distortion? In that case, the survey might be effectively asking people how they would be inclined to adjust the source recording if they were mix or mastering engineers. In that case, one probably could not conclude that people always prefer added distortion. Of course, what I just described is completely hypothetical, but hopefully you see how something like that could provide it's own bias in choices of wav files, and also bias in how the results might be interpreted.
EDIT: Unless, perhaps, it is your intention to demonstrate such bias effects? Never mind, don't answer that.
Last edited:
Not at the moment. It would be one of the biggest mistakes I could make. The source file has, of course, a bit different volume level (remember we have a D/A ==> link stage ==> A/D chain) and of course some time/phase shift and some little difference in amplitude/phase response. I have learned, from my previous tests, that even a smallest level change and time shift makes the biggest difference in the ABX test, especially when performed with headphones.
Talking about opamps, all diyers MUST GO to this thread, with some opamps graph measures by jalejos.
-> http://www.diyaudio.com/forums/vendors-bazaar/281361-hypex-ncore-nc500-build-106.html#post4737441
Burson v5
-> http://www.diyaudio.com/forums/vendors-bazaar/281361-hypex-ncore-nc500-build-106.html#post4737441
Burson v5
Last edited:
Yes, understood how it could change survey responses. But, I was asking about after the ABX test is over, not now. So, I am confused by what you mean by, "not at the moment." Do you mean you don't want to post the files now, or do you mean at the moment you would not consider ever posting the files I requested. If you mean the latter, I can't imagine why that would be.
This is as reliable as feeling softness of a piece of cloth, through a brick wall.
Any picoscopic (because microscopic would be too much) difference between 4 *good* Op Amps used unity gain will be *crushed* by the orders of magnitude higher distortion/noise/nonlinearity/phase shift/you-name-it present between Pavel´s test preamp *there* and our ears *here*
That said, if anybody feels bored and lonely during this long weekend, guessing or simply flipping a coin and answering is as good as anything else to pass the time.
Any picoscopic (because microscopic would be too much) difference between 4 *good* Op Amps used unity gain will be *crushed* by the orders of magnitude higher distortion/noise/nonlinearity/phase shift/you-name-it present between Pavel´s test preamp *there* and our ears *here*
That said, if anybody feels bored and lonely during this long weekend, guessing or simply flipping a coin and answering is as good as anything else to pass the time.
Well, I meant I do not want to post the original now. It is possible to do it after the poll is over.
I don't know what was done for this test, but what if I have an *good* opamp and attenuate the input signal very low, then configure the opamp for a gain of 10,000, and load the output with 100 ohms? Still microscopic, picoscopic totally inaudible distortion?
Mooly,
Yes, exactly. And if we swing the output to 10 volts on the peaks, with 100 ohm load, we should be drawing 100 mA. Hopefully, that would be enough to have some noticeable effect too. If not, we could lower the resistance a bit more.
My point, of course, is that it is not uncommon for people to look at the published distortion number for an opamp and just go with that like it's all there is. It could be viewed as an example of what Kahneman dubbed WYSIATI (What You See Is All There Is), an important and ubiquitous bias of all human brains.
Also, in this case, we don't know what PMA is up to. So far, he isn't saying much.
Yes, exactly. And if we swing the output to 10 volts on the peaks, with 100 ohm load, we should be drawing 100 mA. Hopefully, that would be enough to have some noticeable effect too. If not, we could lower the resistance a bit more.
My point, of course, is that it is not uncommon for people to look at the published distortion number for an opamp and just go with that like it's all there is. It could be viewed as an example of what Kahneman dubbed WYSIATI (What You See Is All There Is), an important and ubiquitous bias of all human brains.
Also, in this case, we don't know what PMA is up to. So far, he isn't saying much.
Last edited:
This (signal gain 1 and noise gain 10,000) can easily be done and in fact it is a circuit in which I test the opamps. If it was used here, there would be great differences in noise, some differences in amplitude response and quite big differences in distortion.
This is not the case of this current test, I can disclose the test circuit (attached), the same circuit used just (undisclosed) opamps are swapped.
Attachments
You could achieve something similar by simply increasing the noise gain, that works even for a unity gain buffer. It still has a gain of 1 yet the opamp 'thinks' it is working with a gain of 1000, or 10,000, whatever you want.
At a gain of 10,000 the response would be plummeting from as little as 100Hz. Its not realistic as a test imo, not at that gain anyway.
Loading the output stage is another factor in itself, and although the two are linked (gain and loading) there is no easy correlation.
A couple of years back when I ran a large opamp comparison, we had two opamps as I recall in an attenuate and then amplify scenario, with heavy loading as well I think, but they were all realistic values.
Lets wait until Pavel reveals the results of this one and hopefully all this will encourage more to have a go.
At a gain of 10,000 the response would be plummeting from as little as 100Hz. Its not realistic as a test imo, not at that gain anyway.
Loading the output stage is another factor in itself, and although the two are linked (gain and loading) there is no easy correlation.
A couple of years back when I ran a large opamp comparison, we had two opamps as I recall in an attenuate and then amplify scenario, with heavy loading as well I think, but they were all realistic values.
Lets wait until Pavel reveals the results of this one
and hopefully all this will encourage more to have a go.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.