Well ok, but what you describe here is caused by (the typically as cheap as possible) power supplies in a computer. There are several other situations when current is drawn, this is not specific to fragmentation. Asynchronous transfer and not using the computers clock at all might be preferable.
(I use a RME Digi96 + ADI2 combination for example that allows for the DAC being the master clock.)
(I use a RME Digi96 + ADI2 combination for example that allows for the DAC being the master clock.)
On "hybrid" op amp circuits:
This is the most basic form I’ve ever seen.
The same, but a little more ellegant. As shown, a variant of this is used in say Revox 226 CD player.
We could call this „Wayne’s Way“, because, with a right selection of the output devices, you could be pumping Amperes of current and the op amp would never know it.
I have tried in real life all of these at one time or another, and all I can say is that one’s op amp will sound better than you ever knew it could. The third, most complex version, makes a great headphone amp with a minimum of fuss.
Two notes.
1. As ever, it is NOT all the same which op amp you use. My experience suggests that of the Old Guard, OP 37 (OPA 37) by Analog gives a very good sound (and it’s safe to use as long as your gain is 6:1 or better – it’s decompensated). The NE series I was never quite happy with. Of the newer generation, Analog’s 829 gives what I would honestly call High Fidelity in the true meaning of the word, and it’s lowish noise, too, and
2. If ever the power supplies played a critical role, this would be it. These circuits react unusually strongly to power supply quality, and I would strongly encourage people to use proper shunt supplies to get the best of it all. At this level, they are not too complex and not too expensive, in case anyone cares about the price.
I can’t really explain why, but by trial and error, I have come to the conclusion that shunt PSU is of critical value in circuits such as those shown above. Sure, they always play a very important part, but no other type of ciruit I have ever worked with has exibited such a sensitivity to power supplies. And if naything, I don’t skimp but overkill most of the time, and especially in case of power supplies; if theory and practice show that a say 10 VA toroid would do, I put in a 30 VA toroid. If they show a 10W transistor is quite enough, I put in a 50W transistor with a double sized heat sink. The worst case safety margin I have ever done was +30%, I typically do +50% and over.
In my neck of the woods, we tend to be very conservative that way. We’re way behind world trends and are still building them to last a long, long time. I guess that’s because we’re too poor to buy cheap junk.
I should also add that I have seen far more complex circuits built around op amps for line use, mostly from Scandinavia. However, I think of them as somewhat silly and disorientated - an op amp actually acts as a simple differential pair in them, backed up by 10+ discrete trannies, plus 10+ discrete diodes, and so forth. I suppose it makes economic sense, given that a decent op amp costs less to far less these days than an ultra matched transistor pair, while op amps are, by and large, cheap enough, especially in quantitty.
This is the most basic form I’ve ever seen.
The same, but a little more ellegant. As shown, a variant of this is used in say Revox 226 CD player.
We could call this „Wayne’s Way“, because, with a right selection of the output devices, you could be pumping Amperes of current and the op amp would never know it.
I have tried in real life all of these at one time or another, and all I can say is that one’s op amp will sound better than you ever knew it could. The third, most complex version, makes a great headphone amp with a minimum of fuss.
Two notes.
1. As ever, it is NOT all the same which op amp you use. My experience suggests that of the Old Guard, OP 37 (OPA 37) by Analog gives a very good sound (and it’s safe to use as long as your gain is 6:1 or better – it’s decompensated). The NE series I was never quite happy with. Of the newer generation, Analog’s 829 gives what I would honestly call High Fidelity in the true meaning of the word, and it’s lowish noise, too, and
2. If ever the power supplies played a critical role, this would be it. These circuits react unusually strongly to power supply quality, and I would strongly encourage people to use proper shunt supplies to get the best of it all. At this level, they are not too complex and not too expensive, in case anyone cares about the price.
I can’t really explain why, but by trial and error, I have come to the conclusion that shunt PSU is of critical value in circuits such as those shown above. Sure, they always play a very important part, but no other type of ciruit I have ever worked with has exibited such a sensitivity to power supplies. And if naything, I don’t skimp but overkill most of the time, and especially in case of power supplies; if theory and practice show that a say 10 VA toroid would do, I put in a 30 VA toroid. If they show a 10W transistor is quite enough, I put in a 50W transistor with a double sized heat sink. The worst case safety margin I have ever done was +30%, I typically do +50% and over.
In my neck of the woods, we tend to be very conservative that way. We’re way behind world trends and are still building them to last a long, long time. I guess that’s because we’re too poor to buy cheap junk.
I should also add that I have seen far more complex circuits built around op amps for line use, mostly from Scandinavia. However, I think of them as somewhat silly and disorientated - an op amp actually acts as a simple differential pair in them, backed up by 10+ discrete trannies, plus 10+ discrete diodes, and so forth. I suppose it makes economic sense, given that a decent op amp costs less to far less these days than an ultra matched transistor pair, while op amps are, by and large, cheap enough, especially in quantitty.
Last edited:
...and refused to do a file compare when that was suggested.
I don't think any of that's relevant anymore though. Originally I assumed this was a one-off thing, and thought it might be caused by data error, fragmentation or something else to do with how the file is stored.
However it's not just one file. Apparently several people have reported improvements ripping different tracks on different computers in safe mode. Now it's being claimed that you can even email bit-identical (but sonically different files) to someone else, and they will hear the difference.
I give up looking for a rational explanation. Given the OP's attitude, I don't think he ever wanted one, but prefers to believe in magic.
Hi,
Based on my testing most of the noise sources are on the Motherboard, but the actual PSU does not help either. And I found that "expensive" PSU's usually are no better regarding noise and modulation than cheap ones.
I agree, however fragmentation is one of the things that can be controlled and where this makes a difference. CPU activity is another.
Sure, however this forms currently still the exception and not the standard.
As Mr. One Stone once is reported to have said, every problem has a solution, if there is no solution there was no problem to start with.
Having a PC and a PSU are not negotiable in the context and unless you build your ATX (or whatever the latest standard is called) compatible low noise supply and clean up the noise on the motherboard, they are not the problem, but a condition.
In the context of this pre-existing condition which we decline to change things like de-fragmented disks, shut down of non-essential system services to reduce CPU loading, under-clocking and undervolting changing RAM refresh cyle setting sin the bios and so on and so on become valid solutions to the problem "on my PC fragmented files sound different from un-fragmented ones" and related ones...
In this case you have a problem how to signal the DAC the changing clock rate if you play anything else than CD...
I use a DAC I designed, which uses a self-powered (not powered from the PC) asynchronous US Audio Class 2 interface that is electrically isolated from the DAC (and fully 24/192 capable).
The DAC has a further Jitter suppression system, memory buffer and programmable very low jitter local clock - this is used mainly because the DAC is SPDIF equipped in addition to USB, a USB only version could dispense with this.
While this arrangements makes sure pretty much all possible routes in which noise or jitter may enter the DAC from the PC are blocked, it is extremely rare and first required me to work out what the problems where first. And yes, using this DAC I find no differences from tweaks that previously had substantial sonic impact.
Previous I used a modified SPDIF Out from the Motherboard with a separate clock/supply and correct driver plus a Jocko Homo/Jim Hagerman (and a bit of secret sauce from me) based frontend driving a CS8414...
The prior arrangement with the right tweaks sounded very good, it was easy to "hear" for example extra disk activity from a background copy or such...
Ciao T
PS, PC uses a passively cooled ECS Atom 330 Motherboard with a Pico-PSU (plus heavily modified 12V/10A "Brick" with much increased PSU capacitances), ATI passively cooled 4350 Graphics (as this unit is "AV") and multiple SATA disks (music is on a completely separate one not shared with anything else)
Based on my testing most of the noise sources are on the Motherboard, but the actual PSU does not help either. And I found that "expensive" PSU's usually are no better regarding noise and modulation than cheap ones.
I agree, however fragmentation is one of the things that can be controlled and where this makes a difference. CPU activity is another.
Sure, however this forms currently still the exception and not the standard.
As Mr. One Stone once is reported to have said, every problem has a solution, if there is no solution there was no problem to start with.
Having a PC and a PSU are not negotiable in the context and unless you build your ATX (or whatever the latest standard is called) compatible low noise supply and clean up the noise on the motherboard, they are not the problem, but a condition.
In the context of this pre-existing condition which we decline to change things like de-fragmented disks, shut down of non-essential system services to reduce CPU loading, under-clocking and undervolting changing RAM refresh cyle setting sin the bios and so on and so on become valid solutions to the problem "on my PC fragmented files sound different from un-fragmented ones" and related ones...
In this case you have a problem how to signal the DAC the changing clock rate if you play anything else than CD...
I use a DAC I designed, which uses a self-powered (not powered from the PC) asynchronous US Audio Class 2 interface that is electrically isolated from the DAC (and fully 24/192 capable).
The DAC has a further Jitter suppression system, memory buffer and programmable very low jitter local clock - this is used mainly because the DAC is SPDIF equipped in addition to USB, a USB only version could dispense with this.
While this arrangements makes sure pretty much all possible routes in which noise or jitter may enter the DAC from the PC are blocked, it is extremely rare and first required me to work out what the problems where first. And yes, using this DAC I find no differences from tweaks that previously had substantial sonic impact.
Previous I used a modified SPDIF Out from the Motherboard with a separate clock/supply and correct driver plus a Jocko Homo/Jim Hagerman (and a bit of secret sauce from me) based frontend driving a CS8414...
The prior arrangement with the right tweaks sounded very good, it was easy to "hear" for example extra disk activity from a background copy or such...
Ciao T
PS, PC uses a passively cooled ECS Atom 330 Motherboard with a Pico-PSU (plus heavily modified 12V/10A "Brick" with much increased PSU capacitances), ATI passively cooled 4350 Graphics (as this unit is "AV") and multiple SATA disks (music is on a completely separate one not shared with anything else)
Wow, becoming heavy handed here ...
One could also argue that those who can, design and build, while those who can't theorize and live off finding faults and "faults" in other people's designs ...
Come on, this kind of "discussion" should be beneath all of us.
Speaking strictly for myself, I am actually grateful to those prescious few who take a look at whatever I am doing, and then tell me, point by point, what I did wrong or what could be done much better with a few mods, and then suggest a few quickly, hand jotted schematics to make themselves perfectly clear. Thorsten's comments on my proposed power amp, in essence and form, PERFECTLY illustrate the point.
Just proclaiming me to be wrong on something, without stating how or why, is hardly constructive, as I would expect on a forum named as this one is.
The biggest problem we all face is when a $5 parts is said to work as well as a say $200 circuit board. Although I know this is theoretically possible, I guess I never ran into such a bad designer as to make a $200 PC board sound so much below par.
But I have seen and heard $5 parts sound as well as $200 PC boards, the only trouble being that by the time you manage that, the whole kaboodle costs way more than $5 and also requires a whole PC board.
Don't get me wrong, I'd kill myself without ICs, I rely on them far too much by now there's no way I could go on without them. They take care of my DC Servo, protection, display ciruitry, and I agree they do the job just as well as far more complex and costly discrete circuits would, at a sane price and with very good reliability.
But an IC is a closed book, it is what it is, take it or leave it. Whereas a discrete circuit allows the designer to control each and every aspect of it, and is, in that sense, far more preferable if one is serious about his work and is not on a shoe string budget.
There is a REASON why Dan d'Agostino's voltage amp, in his FPB series, uses up no less than 112 discrete transistors per channel. You don't necessarily have to agree with him, but you cannot deny the quality of his work, even if you'd rather buy a Levinson. If you think you can, all the better, let's hear your work in a head to head comparison. Bear in mind, though, that all of those 112 trannies are your common off-the-shelf MPS series. Now, THAT is THE trick to pull off.
I can't really comment on John's work, because I have seriously auditioned only one of his products, the 1205 power amp; what I thought of it is best seen by the fact that I tried to buy it (it was offered for sale), and didn't only because somebody was just that little bit faster than I was. And anyway, one swallow doth not a spring make. But I have passively followed John's progress over the years.
I have never heard one bit of Thorsten's work, but I have taken his advice on a number of occasions and was the better off for it. Not surprising, given that we've hung around on forums for some 11 or 12 years now.
Trust is a funny thing - it takes ages to build up, and only seconds to lose. And it's best built up with deeds rather than words.
Hi,
All of them WAY TOO COMPLEX.
I prefer a simple, low HFE VERY HIGH Ft PNP or NPN transistor (which sex depends which way the output stage of the OPA should be class A biased) with a CCS (a 317 plus single resistor serves very well here). Done.
Okay, only SE Class A so we dissipate more power, not very green and all, but so what?
Subjectively this "brute force but simple" output sounds better than all the more complex push-pull options I tried. I'd still rather use tubes though.
Ciao T
All of them WAY TOO COMPLEX.
I prefer a simple, low HFE VERY HIGH Ft PNP or NPN transistor (which sex depends which way the output stage of the OPA should be class A biased) with a CCS (a 317 plus single resistor serves very well here). Done.
Okay, only SE Class A so we dissipate more power, not very green and all, but so what?
Subjectively this "brute force but simple" output sounds better than all the more complex push-pull options I tried. I'd still rather use tubes though.
Ciao T
Indeed so, that configuration seems to be rather popular in Germany, I have a few schematics from Germany and all of them use it. Typically Burr-Brown FET input op amps driving a MOSFET output trannie.
I haven't tried that yet, so I can't comment.
Hi,
Massive parallel transistors are like making "discrete" ring emitter transistors.
You can parallel (and cascode) a large number of small signal parts and end up with something that has much lower parasitics, higher Ft and better Beta linearity than a single higher power part, at lower cost.
Though the situation is changing, as most discretes now are aimed at switched mode something, which means high voltages and fast operation are becoming more common.
I usually find that seriously Munz'ing stuff intelligently is well worth it though.
Ciao T
Massive parallel transistors are like making "discrete" ring emitter transistors.
You can parallel (and cascode) a large number of small signal parts and end up with something that has much lower parasitics, higher Ft and better Beta linearity than a single higher power part, at lower cost.
Though the situation is changing, as most discretes now are aimed at switched mode something, which means high voltages and fast operation are becoming more common.
I usually find that seriously Munz'ing stuff intelligently is well worth it though.
Ciao T
Why you need an OpAMP to deliver amperes at output?
OpAmp's have their role and power stages in NOT what they are made for. A "normal" design should never load an OpAmp output more than 1-2 kohm. I/V stage, filters... they don't need buffers like above.
Newer OpAmps have 20-50mA of clean output (to as low as 600ohm). The problem apears when they are used to do drive 32 ohm headphones - so in the end it is BAD usage of an electronic component. It's user error.
OpAmp's have their role and power stages in NOT what they are made for. A "normal" design should never load an OpAmp output more than 1-2 kohm. I/V stage, filters... they don't need buffers like above.
Newer OpAmps have 20-50mA of clean output (to as low as 600ohm). The problem apears when they are used to do drive 32 ohm headphones - so in the end it is BAD usage of an electronic component. It's user error.
Speaking of listening to vinyl, did anyone read the recent Stereophile review of the Wadax PRE1? (I figure Scott has to have read it, right? 
) What puzzled me was why Mr Fremer chose to listen to an LP made from DSD master tapes and pronounce it to be only a hairsbreadth away from 'real' analog I mean why not find an LP which is analog through and through?
) What puzzled me was why Mr Fremer chose to listen to an LP made from DSD master tapes and pronounce it to be only a hairsbreadth away from 'real' analog I mean why not find an LP which is analog through and through?
Speaking of listening to vinyl, did anyone read the recent Stereophile review of the Wadax PRE1? (I figure Scott has to have read it, right?
Well, maybe he got tired of his direct-to-disc and other audiophile recordings. He's mentioned some recent 45 RPM remastered things, a couple of which I've acquired and have been very happy with, but there's no getting past the tape hiss and scrape flutter.
Brad
Speaking of listening to vinyl, did anyone read the recent Stereophile review of the Wadax PRE1? (I figure Scott has to have read it, right?
I can't find a link to this, I hope they didn't use one of my goshawful IC's.
- Status
- Not open for further replies.