Graig, you should read opamp datasheets. They are often as different to each other as you can imagine. The only thing they have in common is the case (and, come to think of it, even that is no longer the case).
Just as with discrete designs, there is a wide range design decisions made by the designer.
There's bound to be a difference in sound between different IC's and, more important, different implementations.
I find it curious that whenever someone is critical about purported audible differences between discrete and chip, that person is immediately labeled with: "he thinks all [insert your component] sound the same!".
Pretty immature if you realize that the only thing they/we say is: "show me"!.
jan didden
Just as it is utter crap to claim that all discrete circuits sound good. That's pathetic, given the slew of incompetently designed discretes floating around.
jan didden
I find it amazing the lack of accuracy and assumptions in people’s statements in this thread.
jan, what made you assume I do not read datasheets? If you read my comments before you started making comments on this subject matter you would find me making references to them and asking the IC proponents if the were what they considered properly implemented circuits.
I asked Bonsai also if he considered the power supply bypassing of a ceramic cap in parallel with an electrolytic, as recommended in a large number of manufacturer’s datasheets for audio IC op-amps, as proper implementation.
I find it amazing that people on an engineering forum can have such lack of precision with facts and the liberal use of assumption and sweeping generalisations.
Sure you may say my opinion and perceptions of IC op-amps sounding gritty are a generalisation, but it is not, in my experience it is fact. Even one of the best sounding IC op-amps IMO, the OPA627, sounds gritty or grainy. A significant part of that can be removed by biasing the output stage into class A with about 10mA or so. IMO that has to be a little bit of a clue about why they sound the way they do. However, I don't believe it is singular cause but more likely more than one. But what is the point of working out why when nothing can be done about it. Other than the pursuit of satisfying a curiosity.
The argument about different circuit implementation is not relevant to my statements. All of my findings were based on a singular optimised layout. The only change being the IC op-amp. Careful scoping of power supplies and outputs was done to confirm potential issues with transients and oscillation.
I didn't discover this by my own experience. I was heavily into trying all different types of IC op-amps and optimising them for best possible sonic performance. A person who I knew in the same field said to me "what are you wasting your time with those things for, they all sound the same, gritty and harsh. There is nothing you can do to get around it."
I persevered for a while longer trying everything I could, the AD811 power supplies was one of them. The first discrete design I did was for an I/V application (A feedback design) easily bested the best IC op-amp implementation. So much so that there was no point in returning to IC op-amps for serious audio reproduction or recording. I felt like I lost so much time in trying to make the things sound any good.
Hell, even some small signal transistors sound grainy, gritty, harsh, whatever you want to call it, compared to some other solid state devices. But that is much as I'll say about that. There are some people around that pick up on things easily, so I don't want to give away hard work for free. Probably said too much already.
Anyway, I am sure this is going to cause a whole lot more controversy and insults. Too much noise and not enough signal on these forums generally. Have to get on with life and stop wasting time here.
jan, what made you assume I do not read datasheets? If you read my comments before you started making comments on this subject matter you would find me making references to them and asking the IC proponents if the were what they considered properly implemented circuits.
I asked Bonsai also if he considered the power supply bypassing of a ceramic cap in parallel with an electrolytic, as recommended in a large number of manufacturer’s datasheets for audio IC op-amps, as proper implementation.
I find it amazing that people on an engineering forum can have such lack of precision with facts and the liberal use of assumption and sweeping generalisations.
Sure you may say my opinion and perceptions of IC op-amps sounding gritty are a generalisation, but it is not, in my experience it is fact. Even one of the best sounding IC op-amps IMO, the OPA627, sounds gritty or grainy. A significant part of that can be removed by biasing the output stage into class A with about 10mA or so. IMO that has to be a little bit of a clue about why they sound the way they do. However, I don't believe it is singular cause but more likely more than one. But what is the point of working out why when nothing can be done about it. Other than the pursuit of satisfying a curiosity.
The argument about different circuit implementation is not relevant to my statements. All of my findings were based on a singular optimised layout. The only change being the IC op-amp. Careful scoping of power supplies and outputs was done to confirm potential issues with transients and oscillation.
I didn't discover this by my own experience. I was heavily into trying all different types of IC op-amps and optimising them for best possible sonic performance. A person who I knew in the same field said to me "what are you wasting your time with those things for, they all sound the same, gritty and harsh. There is nothing you can do to get around it."
I persevered for a while longer trying everything I could, the AD811 power supplies was one of them. The first discrete design I did was for an I/V application (A feedback design) easily bested the best IC op-amp implementation. So much so that there was no point in returning to IC op-amps for serious audio reproduction or recording. I felt like I lost so much time in trying to make the things sound any good.
Hell, even some small signal transistors sound grainy, gritty, harsh, whatever you want to call it, compared to some other solid state devices. But that is much as I'll say about that. There are some people around that pick up on things easily, so I don't want to give away hard work for free. Probably said too much already.
Anyway, I am sure this is going to cause a whole lot more controversy and insults. Too much noise and not enough signal on these forums generally. Have to get on with life and stop wasting time here.
Yes I do decouple my designs at the PSU pins. Take a look on my website.
I have done some work on biasing the oupt stage into class class A. However even without this audio op amps sound pretty damn Awsome in my view and still deliver <1ppm distortion performance. Quite how anyone can then assert that they sound harsh and gritty is beyond me
You have not pointed us to any of your work (web site, publications etc), or even told us what system you are using or measuring equipment to assess your efforts and arrive at your conclusions.
I suspect that you realize you have in fact painted yourself into a corner on this one.
For my part, I will continue to use opamps, and discrete designs where I think they add value. However, I am not going to listen to or respond to unsubstantiated anti op amp bs on this thread anymore.
Over and Out.
I have done some work on biasing the oupt stage into class class A. However even without this audio op amps sound pretty damn Awsome in my view and still deliver <1ppm distortion performance. Quite how anyone can then assert that they sound harsh and gritty is beyond me
You have not pointed us to any of your work (web site, publications etc), or even told us what system you are using or measuring equipment to assess your efforts and arrive at your conclusions.
I suspect that you realize you have in fact painted yourself into a corner on this one.
For my part, I will continue to use opamps, and discrete designs where I think they add value. However, I am not going to listen to or respond to unsubstantiated anti op amp bs on this thread anymore.
Over and Out.
Last edited:
Bypassing opamps with ceramics (typical 100nF + 1nF optional c0g) is a must as we need to give them a lowimpedance powersupply up to 10 - 100MHz or further and that cannot be done with film caps or el caps! It has to be placed as close as possible also.
I would say less than 5mm and more than 20mil wide tracks.
I would say less than 5mm and more than 20mil wide tracks.
Yes Bonsai I already did have a look at your website and your designs. I made particular reference to your decoupling methods earlier. Textbook style or should I say datasheet style. If you do some more experimentation you will find there are some futher gains to be made in that respect.
WRT to the o/p biasing there are even more gains to be made with distortion generated at the input pair stage of the op-amp. I don't see it looking on these forums anywhere and it is certainly not commonly known or talked about.
I am only interested in pointing people with an open mind and who are concerned with the original question about the sound of discrete op-amps.
Anyone that wants to get personal or insult I suggest politely to not ruin a great resource for other interested people.
WRT to the o/p biasing there are even more gains to be made with distortion generated at the input pair stage of the op-amp. I don't see it looking on these forums anywhere and it is certainly not commonly known or talked about.
I am only interested in pointing people with an open mind and who are concerned with the original question about the sound of discrete op-amps.
Anyone that wants to get personal or insult I suggest politely to not ruin a great resource for other interested people.
Hi all.
I am going to make an attempt to squeeze a jfet input, folded cascode with diamond buffer down 500x500mil.
It does sound like a menu card - doesn't it?
The Jfet input has a noise level of 6nV at 100Hz.
I have not simulated it yet, but the biggest issue is to get the second gm stage high enough regarding impedance.
The capacitor for setting the 1st corner is going to be connected to one of the rails. So here the impedance of the powersupply will be very important. It also need to be lownoise/ripple.
It is going to be made as an replacement for any single opamp in and dip8 housing.
I have no analyzer at the moment, so maybe i could send it to bonsai for test?
I am going to make an attempt to squeeze a jfet input, folded cascode with diamond buffer down 500x500mil.
It does sound like a menu card - doesn't it?
The Jfet input has a noise level of 6nV at 100Hz.
I have not simulated it yet, but the biggest issue is to get the second gm stage high enough regarding impedance.
The capacitor for setting the 1st corner is going to be connected to one of the rails. So here the impedance of the powersupply will be very important. It also need to be lownoise/ripple.
It is going to be made as an replacement for any single opamp in and dip8 housing.
I have no analyzer at the moment, so maybe i could send it to bonsai for test?
Sonnya, the good thing about what you are doing is that it requires minimal mods to configure as an open loop amplifier with no feedback.
That way you can try both configurations (or something in between) and choose the one you prefer.
You can also choose to pick the feedback off somewhere before the output stage.
Lots of options that are for the most part not available with IC versions.
That way you can try both configurations (or something in between) and choose the one you prefer.
You can also choose to pick the feedback off somewhere before the output stage.
Lots of options that are for the most part not available with IC versions.
Yes that is right. But i have been down the path using construction like this one (http://www.diyaudio.com/forums/analog-line-level/189403-jfet-folded-cascode-diamond-buffer-opamp.html) without feedback, and i was not happy about it. After the "Wauv" factor has declined the sound was harsh.
On the other hand, my Mirand A1 which Tinitus and PIX owns is very neutral and described as "wire with gain". It has a high openloop gain and is very wideband (800KHz).
So i am not afraid of negative feedback anymore. On the other hand. Design like ZEN I/V converter is a beautiful flower in its simplicity and how well it performs.
On the other hand, my Mirand A1 which Tinitus and PIX owns is very neutral and described as "wire with gain". It has a high openloop gain and is very wideband (800KHz).
So i am not afraid of negative feedback anymore. On the other hand. Design like ZEN I/V converter is a beautiful flower in its simplicity and how well it performs.
But did you consider what was the cause of the harsh sound?
Was it because it didn't have feedback that it had harsh sound?
Or was it that the feedback helped correct the source(s) of harsh sound?
If the harsh sound was not due to lack of feedback and was addressed would the resulting non feedback design sound as good with feedback?
Was it because it didn't have feedback that it had harsh sound?
Or was it that the feedback helped correct the source(s) of harsh sound?
If the harsh sound was not due to lack of feedback and was addressed would the resulting non feedback design sound as good with feedback?
Hi Craig.
I would suspect the problem to be in the choice of output stage - Emitterfollower.
Bias them hard into class A would solve the problem.
Under all circumstances i would at all time go for at least a little feedback when we are talking heavy loads.
When i mean heavy load : Speaker
For linelevel or headphone amp:
I would suspect the problem to be in the choice of output stage - Emitterfollower.
Bias them hard into class A would solve the problem.
Under all circumstances i would at all time go for at least a little feedback when we are talking heavy loads.
When i mean heavy load : Speaker
For linelevel or headphone amp:
- A good diamond buffer as output with high idle current
- A mosfet, BJT, JFET attached to a current source like Pass constructions - also high idle current.
- a CFP pair output stage in class A.
Last edited:
and there's me thinking it was a simple question to start 
which now leads me to this....
If many here believe there is no audible difference between a well designed discrete opamp and a monolithic, then why are the same people designing discretes using different topologies, cascoding stuff et al?
Is there ANY audible difference in Samuel's LN opamp running at 24V and the Hardy JE990 running at 24v? And if not how can that be, because the transistors are different, and surely if you change all the variables then end result MUST be different.
Surely if one believes that all opamps 'sound' the same and there are no gains in going discrete OR a good discrete design is as good as a monolithic then it's time to forget designing opamps...
I'M CONFUSED...mainly because when I look at all the different opamp topologies that exist here on diyaudio, some of you are saying they should all sound the same! That cannot be right, and if that is the case then no 2 opamp designs should sound the same -which to me is right.
which now leads me to this....
If many here believe there is no audible difference between a well designed discrete opamp and a monolithic, then why are the same people designing discretes using different topologies, cascoding stuff et al?
Is there ANY audible difference in Samuel's LN opamp running at 24V and the Hardy JE990 running at 24v? And if not how can that be, because the transistors are different, and surely if you change all the variables then end result MUST be different.
Surely if one believes that all opamps 'sound' the same and there are no gains in going discrete OR a good discrete design is as good as a monolithic then it's time to forget designing opamps...
I'M CONFUSED...mainly because when I look at all the different opamp topologies that exist here on diyaudio, some of you are saying they should all sound the same! That cannot be right, and if that is the case then no 2 opamp designs should sound the same -which to me is right.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.