Hi,
I was reading another forum and found an interesting subject discussing opamps used for unity gain.
The general comments were opamps when used for unity gain sound bad. The possible cause, I guess, is the largest amount of feedback used.
Note that I am only talking about using unity-gain-stable opamps. I use opamps in active crossover network. Shallen key is common place.
So, is this right? Does it mean we should always add some gains, despite higher noise? High frequency gain or low frequency gain? How much gain? Would it make the opamp clip easily? Any opamps that sound the best for unity gain buffers? opa627, opa2134? ad825? What is your solution to this?
Regards,
Bill
I was reading another forum and found an interesting subject discussing opamps used for unity gain.
The general comments were opamps when used for unity gain sound bad. The possible cause, I guess, is the largest amount of feedback used.
Note that I am only talking about using unity-gain-stable opamps. I use opamps in active crossover network. Shallen key is common place.
So, is this right? Does it mean we should always add some gains, despite higher noise? High frequency gain or low frequency gain? How much gain? Would it make the opamp clip easily? Any opamps that sound the best for unity gain buffers? opa627, opa2134? ad825? What is your solution to this?
Regards,
Bill
A post by Nelson Pass here can give a different POV of the problem : http://www.diyaudio.com/forums/showthread.php?postid=606437#post606437
I'd say th e comments about stability ar e valid in many cases . . . . but then read the app note and take th e necessary precautions to prevent oscillation/misbehaviour:-
Dont operate non-unity gain stable op-amps at gains lower than specified by the manufacturer
Make sure you isolate capacivie loads adequately - these can cause a pole in the op-amps response and this leads to instability
Decouple thoroughly and close to th e op-amp
Make sure your grounds are clean and are low inductance
For summing circuits (or non-inverting feedback nodes), place the feedback resistors AS CLOSE TO THE INVERTING INPUT PIN AS POSSIBLE. I've seen this rule broken many times and always with the same results - noise, EMC pickup, instability
Finally, to all those people that complain about th e 'op-amp' sound, remember next time you listen to one of your favourite CD's (or maybe vinyl recording) that the signal chain from mike through to cutting head has almost certainly gone through about quite a few op-amps along the way. Welcome to the 'op-amp' sound!
Dont operate non-unity gain stable op-amps at gains lower than specified by the manufacturer
Make sure you isolate capacivie loads adequately - these can cause a pole in the op-amps response and this leads to instability
Decouple thoroughly and close to th e op-amp
Make sure your grounds are clean and are low inductance
For summing circuits (or non-inverting feedback nodes), place the feedback resistors AS CLOSE TO THE INVERTING INPUT PIN AS POSSIBLE. I've seen this rule broken many times and always with the same results - noise, EMC pickup, instability
Finally, to all those people that complain about th e 'op-amp' sound, remember next time you listen to one of your favourite CD's (or maybe vinyl recording) that the signal chain from mike through to cutting head has almost certainly gone through about quite a few op-amps along the way. Welcome to the 'op-amp' sound!
Hi,
Unity gain op-amps used in high gain circuits do not sound so good.
Non-unity gain op-amps used at unity gain oscillate, not too good.
The idea that high gain op-amps "slugged" for unity gain sound poor
is somewhat of a fallacy but an advanced, (compared to the normal
6dB/octave scheme), 12dB/octave compensation scheme can improve
high frequencies (y filters) by maintaining internal loop gain and
consequently feedback factor at high frequencies.
The same is true for audio gain stages ......
I'd say this is the "golden ears" imagining things. It is only possibly
true is stabililty is marginal and improved by higher gain. As an idea
it is overly simplistic as the "amount of feeddback" i.e. the feedback
factor is largely constant for a given stability. By increasing gain and
reducing feedback factor (increasing stability) distortion is increased.
Unless of course the op-amp sounds fabulous with the lowest possible
feedback, i.e. max gain and attenuating the output for your required gain.
🙂/sreten.
Unity gain op-amps used in high gain circuits do not sound so good.
Non-unity gain op-amps used at unity gain oscillate, not too good.
The idea that high gain op-amps "slugged" for unity gain sound poor
is somewhat of a fallacy but an advanced, (compared to the normal
6dB/octave scheme), 12dB/octave compensation scheme can improve
high frequencies (y filters) by maintaining internal loop gain and
consequently feedback factor at high frequencies.
The same is true for audio gain stages ......
I'd say this is the "golden ears" imagining things. It is only possibly
true is stabililty is marginal and improved by higher gain. As an idea
it is overly simplistic as the "amount of feeddback" i.e. the feedback
factor is largely constant for a given stability. By increasing gain and
reducing feedback factor (increasing stability) distortion is increased.
Unless of course the op-amp sounds fabulous with the lowest possible
feedback, i.e. max gain and attenuating the output for your required gain.
🙂/sreten.
I've got some confusing experience with this.
When one amplifier is nearly perfect (technically), the sound is said "dull, dead, thin, bad".
When there's another amp that has a not so good clipping behavior plus quite high harmonic distortion (2nd, 3rd), the sound is said "good".
Chipamp with PS cap only nF (while the manual suggest about 1000uf/more) and omits the zobel stabilisator is said "very realistic", but I'm afraid that there's a marginally oscilation threshold and the PS junk is actually entering the sound. Confusing for me 😀
I suspect not. 😀
I know, I know, it's a dull way to look at things, but wouldn't it be useful if, in subjective evaluation, people stopped talking about a box of gain "sounding good" or "sounding bad," but instead talked about whether or not the output sounded like the input?
I didn't want to offer the "boring" view, but I just can't help myself any longer ;-) It baffles me why people think they can just pop one of a dozen opamps into a circuit, then listen and declare they know something about the way opamps sound. Where is it written that all those opamps operate at their best in the same circuit? Why do people think all those different opamps exist in the first place? Sometimes the tests are done without a scope, often without a schematic or knowledge of the actual circuit configuration (typically in CD players). "Rolling" opamps is a joke. IMO, almost any of the popular high performance opamps can be stabilized and configured to be indistinguishable in audio applications, if they have sufficient gain bandwidth product, low distortion, and aren't asked to deliver excessive output current or drive an unsuitable load. IMO, opamps are viewed as really simple to use, and the truth is that a certain amount of knowledge and skill is required to get the most out of 'em.
Conrad, your thoughts are always in my mind but I have long time stopped arguing about it.
Picture this: Not 100% good recording => MP3 coded => average sound card or even bad => not optimal pcb in general and not in particular for the opamp => bad or average headphones
This is a common setup to judge the characteristics of an opamp.
Picture this: Not 100% good recording => MP3 coded => average sound card or even bad => not optimal pcb in general and not in particular for the opamp => bad or average headphones
This is a common setup to judge the characteristics of an opamp.
Hi, SY,
That would be very nice, and will save alot of reviews 😀
What makes me confuse is I like the sound of those "faulty" amps. When I tweaked both the "not so good clipping behavior plus quite high harmonic distortion (2nd, 3rd)" and "Chipamp with PS cap only nF (while the manual suggest about 1000uf/more) and omits the zobel stabilisator" to meet the technical POV, ie: stable towards R//C load, have the best square wave response and best residual (in my distortion meter+scope), the sound becomes "worse". (oops..sorry, I use this kind of term again 😀)
That would be very nice, and will save alot of reviews 😀
What makes me confuse is I like the sound of those "faulty" amps. When I tweaked both the "not so good clipping behavior plus quite high harmonic distortion (2nd, 3rd)" and "Chipamp with PS cap only nF (while the manual suggest about 1000uf/more) and omits the zobel stabilisator" to meet the technical POV, ie: stable towards R//C load, have the best square wave response and best residual (in my distortion meter+scope), the sound becomes "worse". (oops..sorry, I use this kind of term again 😀)
NFB is not the cause of 'sound' differences in unity gain op-amps.
whether 120dB of NFB or 130 dB NFB wont matter to the sound.
Stability may be an issue.
whether 120dB of NFB or 130 dB NFB wont matter to the sound.
Stability may be an issue.
AD797 - properly implemented, it's the best sounding opamp in unity gain configuration (chosen among 20-30 most popular and best specified opamps made by TI, Analog, Linear).
It even sounds better than many discrete buffers.
Don't worry too much about NFB. It's not good or bad by itself , it will just reveal topology behind it.
http://www.diyaudio.com/forums/showthread.php?postid=980927#post980927
It even sounds better than many discrete buffers.
Don't worry too much about NFB. It's not good or bad by itself , it will just reveal topology behind it.
http://www.diyaudio.com/forums/showthread.php?postid=980927#post980927
Of course, you can build a unity gain stable circuit with *any* opamp, regardless of its internal compensation. If I'm limited to two resistors and two bypass caps, then my options are a bit more limited ;-)
True enough, and it might not be my first choice for a unity gain buffer, though it can be forced to do the job. IMO, the 797 has a great data sheet, and op amp "rollers" would be wise to read it carefully. It ain't necessarily plug & play with a lot of other parts, unless you mod the circuit accordingly, which sort of makes my original point.
PMA said:
Sorry PMA, but according to my extensive experience with this chip - what you are saying is simply not true.
AD797 is absolutely stable at unity gain - I've built numerous circuits with it and there was no sign of instability at all.
I don't wish any further argument on this subject - I'm talking out of experience, not waving with the paper.
Yes, Scott Wurcer is the man who designed AD797 and he is the member of diyaudio.
He might say a word or two about it 🙂
He might say a word or two about it 🙂
Thanks, guys. Please keep your input coming and let us newbies be educated.
Yes, to me the comments I read from the other forum were more of a myth. I have used opa2134 mainly, most in unity gain and have not subjectively found any audible problem. I'd say the system is pretty much of high fidelity, although I must confess that opa627 was subjectively better in one of my experiments.
So, is AD797 the best for unity gain? Juma, would you mind sharing on how to properly implement it?
Yes, to me the comments I read from the other forum were more of a myth. I have used opa2134 mainly, most in unity gain and have not subjectively found any audible problem. I'd say the system is pretty much of high fidelity, although I must confess that opa627 was subjectively better in one of my experiments.
So, is AD797 the best for unity gain? Juma, would you mind sharing on how to properly implement it?
- Status
- Not open for further replies.