I'm quite sure it is not about stability...
Have you ever heard about psycho-acoustics????
Hi Alex,
Haha, I know that was a jibe and a rhetorical question, but for the record:
I'm trained as an engineer and therefore torn between wanting to believe the flat-earth-don't-bother-listening-the-instruments-tell-all argument that is convenient for engineers to rely on, and what my ears are telling me. In my opinion there ought to be a lot more correlation between the two if the hard-core "rationalists" were to be justified in maintaining their position so arrogantly.
The reason I'm still flogging a dead horse with the Self preamp is that I've invested quite a bit of time and money into building it and would therefore like to at least improve its SQ if not understand why it doesn't live up to expectations. But it seems the former is not very likely and the latter is extremely unlikely to happen at this rate! I certainly don't have the time or equipment available to carry out extensive research to find out why, and it seems the rest of the contributors here either don't believe there could be a problem or have already decided they know why it doesn't sound good despite relying on pure speculation. I suppose I'm still hanging on to the hope that it is possible to design good audio gear objectively if only THD could be banished as a useful measure of a design's performance and more work be done by those in a position to do so to come up with a new standard for measuring audio performance. The weighted-THD curve seems like a good step in this direction.
So maybe the problem is instead that designers shouldn't be allowed to keep using >>35 year-old test methods as a means of proving their design's success. Consistent improvements in THD+N curves or figures over the years has arguably not improved equipment sound quality and yet this still seems to be regarded as the only measurement worth showing or stating.
Haha, I know that was a jibe and a rhetorical question, but for the record:
I'm trained as an engineer and therefore torn between wanting to believe the flat-earth-don't-bother-listening-the-instruments-tell-all argument that is convenient for engineers to rely on, and what my ears are telling me. In my opinion there ought to be a lot more correlation between the two if the hard-core "rationalists" were to be justified in maintaining their position so arrogantly.
The reason I'm still flogging a dead horse with the Self preamp is that I've invested quite a bit of time and money into building it and would therefore like to at least improve its SQ if not understand why it doesn't live up to expectations. But it seems the former is not very likely and the latter is extremely unlikely to happen at this rate! I certainly don't have the time or equipment available to carry out extensive research to find out why, and it seems the rest of the contributors here either don't believe there could be a problem or have already decided they know why it doesn't sound good despite relying on pure speculation. I suppose I'm still hanging on to the hope that it is possible to design good audio gear objectively if only THD could be banished as a useful measure of a design's performance and more work be done by those in a position to do so to come up with a new standard for measuring audio performance. The weighted-THD curve seems like a good step in this direction.
So maybe the problem is instead that designers shouldn't be allowed to keep using >>35 year-old test methods as a means of proving their design's success. Consistent improvements in THD+N curves or figures over the years has arguably not improved equipment sound quality and yet this still seems to be regarded as the only measurement worth showing or stating.
I'm quite sure it is not about stability...
Have you ever heard about psycho-acoustics????
No never heard of it
What's your point?
Why dont you see if there's anyone here on the forum who has access to an AP - then ask them to test your pre. You will need to ship it to them. I used to have access to a Sys272 in Tokyo, but unfortunately no longer. At least that way, we can assess whether it is some sort of measurable problem, or whether there is an issue with its sonics.
In these matters, I like to consider myself a 'pragmatist'. I do believe equipment sounds different, but I am not inclined in any way to assign those differences to anything but hard engineering facts - I dont believe in snake oil bs. Period. However, its clear you are not 100% happy with this thing, which is a pity - but you need to get to the root cause. Maybe you will want to try my new pre out when its finished - I've taken a slightly more conventional path but focused on making everying run in class A - there is also the option of a discrete gain block as well.
No never heard of it
My point is the same that all "objectivists" have: you are hearing what you want to hear.
It is impossible to discern "by using the ears" among different OAs... unless they are oscillating but, as you have said, it is not the case. I know it is hard to believe that our ears are "not telling the truth", but it is what happen.
Anyway, it has been investigated and proved a very long time ago, and if you claim that the XYZ OA "sounds bad"... well, an ABX test will show a closer approach to the truth.
There's a reasoning flaw - permit me to point it out. AP does not incorporate all possible measurements, just permits some popular ones to be done fairly easily and other, less popular ones with a fair bit more effort. So no, a clean bill of health from an AP (or AP2) (which I fully expect) does not eliminate measurable problems.
Why would (just to take one example) a multitone IMD test be 'snake oil' ? Looks to me like a red herring. A true pragmatist incidentally would test his hypothesis rather than introduce deflections like 'snake oil' - deflections are non-pragmatic in that they serve no practical purpose.
There was no malice intended in my post - I was actually trying to be helpful. The snakeoil comments were not aimed at you -it was a general comment.
I have used the 5532, 5534, 4562 and the 49710 and have not experienced the kind of problems you mention (I also used the PGA2320 - nice sounding chip). I've had great results with all these chips.
THD tests (any type) are an accepted method of measuring amplifer performance and allows one to quickly guage whther there are any major problems - its a very useful tool, but of course its not the b all and end all and below 0.1% there is little return on design effort (Nelson Pass has made some succint comments on this point)
If you are getting such a bad sound, then my advice remains the same, get it measured to see if it meets spec and you dont have some gross problem. 2% distortion on a sine wave is pretty much invisible on a scope, but could have a major impact on the sound if it was high order harmonics.
I have used the 5532, 5534, 4562 and the 49710 and have not experienced the kind of problems you mention (I also used the PGA2320 - nice sounding chip). I've had great results with all these chips.
THD tests (any type) are an accepted method of measuring amplifer performance and allows one to quickly guage whther there are any major problems - its a very useful tool, but of course its not the b all and end all and below 0.1% there is little return on design effort (Nelson Pass has made some succint comments on this point)
If you are getting such a bad sound, then my advice remains the same, get it measured to see if it meets spec and you dont have some gross problem. 2% distortion on a sine wave is pretty much invisible on a scope, but could have a major impact on the sound if it was high order harmonics.
Last edited:
Hi owedo,
You make a good point about your experience with the newer LM4562, but I think it also makes my point.
That is, you can plainly hear the difference between two different op amps that have superb measurements. In this case, it was the 35-year-old 5532 that sounded better to you in a particular preamp.
Thus, the point you are making is that op amps can sound very different even when the specs and measured performance don't seem to justify the difference. By the same token, a modern op amp could sound dramatically better than the old 5532. In fact, many have said over the years that the 5532 measured extremely well but did not sound as good as many newer op amps. I realize that this may not be the case with the LM4562.
It is also possible that the circuit design of the Precision Preamp is not LM4562-friendly for some reason. For example, did you measure the distortion and other performance in the Precision Preamp with both 5532 and LM4562 devices? Any chance the 4562's might have been oscillating? Who knows? The point here is that just because the spec sheets say both op amps are very good, does not mean they will both be very good in a given circuit - e.g., when that circuit was designed and measured with the 5532, but 4562s were just dropped in.
Seems like there is a lot of food for though and wiggle room here.
Cheers,
Bob
I would say, given that both these opamps by any measure are very good, you should not be hearing huge differences. It's nuances and not night and day. If the differences are as stark as you say, then I would have to concur there is a problem.
I had a 180 MHz oscillation problem on my e-Amp cascode stage, and a few months ago on a headphone buffer breadboard a 50 MHz 2V pk to pk problem (4562 driving a discrete EF). A ferrite fixed it. Point I am making is its very easy in these high performance circuits to pick up some strange stuff that is likely to affect the sound. I'd look for these kinds of issues before blaming the opamp.
I had a 180 MHz oscillation problem on my e-Amp cascode stage, and a few months ago on a headphone buffer breadboard a 50 MHz 2V pk to pk problem (4562 driving a discrete EF). A ferrite fixed it. Point I am making is its very easy in these high performance circuits to pick up some strange stuff that is likely to affect the sound. I'd look for these kinds of issues before blaming the opamp.
I've actually used LM4562's in Dougs precision preamp (the single sided PCB version I made up to evaluate it)
http://www.diyaudio.com/forums/soli...ned-d-selfs-precision-preamp.html#post2028844
I'm hard pushed to say that there is any sonic difference between the 5532 and the 4562 although I have since altered the design slightly to accommodate lower impedance circuitry, not in the pursuit of noise but to push inter channel crosstalk ever lower (again as an experiment because I want to incorporate this into a new integrated amp at some point). I also tried the TLE2072 FET opamp because of its (claimed) high drive ability. Although I would need to do more detailed listening tests, I think I actually preferred the bjts, something which I am surprised at. There did seem some milage in using both 4562 and TLE2072 at differing locations in the circuit.
When the bass and treble are centered correctly on the precision preamp you are essentially listening to just an inverting (Gain -1) opamp.
Does the finger of suspicion come down to the active volume control again ???
There is certainly no oscillation detectable using a 100Mhz scope.
http://www.diyaudio.com/forums/soli...ned-d-selfs-precision-preamp.html#post2028844
I'm hard pushed to say that there is any sonic difference between the 5532 and the 4562 although I have since altered the design slightly to accommodate lower impedance circuitry, not in the pursuit of noise but to push inter channel crosstalk ever lower (again as an experiment because I want to incorporate this into a new integrated amp at some point). I also tried the TLE2072 FET opamp because of its (claimed) high drive ability. Although I would need to do more detailed listening tests, I think I actually preferred the bjts, something which I am surprised at. There did seem some milage in using both 4562 and TLE2072 at differing locations in the circuit.
When the bass and treble are centered correctly on the precision preamp you are essentially listening to just an inverting (Gain -1) opamp.
Does the finger of suspicion come down to the active volume control again ???
There is certainly no oscillation detectable using a 100Mhz scope.
I bought 5pcs of LM4562 for evaluation and possible application. I've had great expectations with this opamp knowing it's specs and that it is bipolar, which I like. I used it for several months hoping that it will break in. I also changed gain trying to find out why it turned out to be disappointment. No way, it never sounded quite right from the day one. I lost interest in them and now they are gathering dust. I doubt that in owdeo circuit it's oscillating problem. They are simply unmusical. It is not something particular that is wrong with them, something that you can detect and name. The whole impression is not satisfying. I am inclined to think that circuit designer had to surpass what is available at the market and that he included too many enhancements in the circuit making it overdone. He simply traded musicality for impressive specs. Gentlemen, I really think that as far as the sound is concerned, the whole audio opamp chapter is closed long time ago. In my humble opinion, the only breakthrough that is possible from now on is to design specialized current feedback opamps for audio (not just using AD844 for audio). VFB opamp chapter is closed.
I can see now why specialized audio companies like Profusion plc are offering only a very limited number of opamp types, and almost all are old. It is very difficult to produce something that will give better subjective results. The only way to get something different, at least subjectively, is to use either CFB or video opams and to make circuit friendly for these types (no oscillation, etc.)
If that could be shown to better correlate with listening tests then sure, but the normal two-tone IMD test as built into audio test gear probably wouldn't cut it - more sophisticated signals and interpretation of results along the lines of abraxilito's suggestion sounds promising.
Complete nonsense. If you really believe that why bother reading this discussion? You haven't proved anything by stating that boring and ridiculous argument, and certainly not that I can't hear any difference. I don't want to go into a philosophical discussion on the meaning of truth, but your idea of it is too depressing to contemplate further. Stay in your black-and-white fantasy world and get your computer to listen to music for you - it will do it more accurately
Owedo, you need to 'qualify and quantify' your observations of the new Self pre. You say it kills the music but you need to compare it t your previous preamp and comment on specifics viz
Mid band
Bass
Top end
Imaging left to right
Imaging front to back
Overall character - is it forward sounding or relaxed
Sound 'taste- ' is it dark and smooth, or lively and edgy
Etc
Mid band
Bass
Top end
Imaging left to right
Imaging front to back
Overall character - is it forward sounding or relaxed
Sound 'taste- ' is it dark and smooth, or lively and edgy
Etc
Hi Bob,
Excellent points thanks, it seems silly to generalise about the sound of a particular opamp given its performance will depend strongly on the particular circuit and implementation. Having said that, I do believe they each a have their own "flavour" that is independent of it. The circuit and implementation only seem to affect how strongly the particular subjective character manifests.
This particular design does seem to emphasize the opamp character as evident by how different it sounds even with different brands of 5532. But Self has presented his update in Linear Audio with the same basic configuration and using LM4562s throughout, so from this I would assume the LM4562 should work fine and measure slightly better for THD+N. There is no sign of instability on my 'scope, however I lack the test equipment to measure distortion low enough unfortunately. At this stage my (obvously subjective) conclusion is that the LM4562 is an unpleasant device to listen to, but may be less annoying in other circuit configurations.
Cheers,
owdeo
Hi Mooly,
Half my LM4562s were sourced as samples from Nat Semi a few years back when I was an Application Engineer for a distributor, and the remainder were purchased from Element14 (Farnell) so I'm confident they're genuine.
Yes I do think the active gain control stage is the most likely culprit at this point. Did you build this into your prototype also, or was it just the buffer and tone stage?
My guess is that the "nested" feedback where there is a unity gain buffer with 100% feedback inside the gain control feedback loop is somehow responsible for the subjective effect. Dunno why. I was going to try configuring the stage as a simple inverting gain stage without the buffer but it requires quite messy mods on this board and renders the preamp useless other than as a tone control so thus far I have resisted...
Cheers,
owdeo
Neither you by saying what you (believe to) hear . If you need a prove, you should look at your loudspeakers, which have a distortion (tipically) greater than 1% and then you should ask yourself: Is it possible to discern a 0.001% distortion buried below the loudspeaker's distortion?
Ok! Let's kill the messenger...I don't want to go into a philosophical discussion on the meaning of truth, but your idea of it is too depressing to contemplate further. Stay in your black-and-white fantasy world and get your computer to listen to music for you - it will do it more accurately