From what I've understood, settling time and slew rate are very important for I/V, and these don't come close to the 826.
Cheers George
10s of ns settling time doesn't matter for audio
the point of I/V is to linearly turn the DAC Iout into a audio
the feedback C should only be the 1st pole of a analog low pass reconstruction filter
from a specs perspective a better I/V op amp would be the ADA4898
the 'highly linear input stage" is poorly advertised but is exactly what's needed in I/V to maintain linearity in the face of the mV switching glitches and over the feared settling time dynamics
you could sum the DAC Iout into one I/V or use the 2 DAC, one I/V each for balanced output
but as mentioned summing the 2 DAC into 2 I/V isn't a simple, safe option
the ADA4899 specs look even better but I would want 'scope and active probes costing as much as a (cheap) new car to see that it ain't misbehaving at 600 MHz
the point of I/V is to linearly turn the DAC Iout into a audio
the feedback C should only be the 1st pole of a analog low pass reconstruction filter
from a specs perspective a better I/V op amp would be the ADA4898
the 'highly linear input stage" is poorly advertised but is exactly what's needed in I/V to maintain linearity in the face of the mV switching glitches and over the feared settling time dynamics
you could sum the DAC Iout into one I/V or use the 2 DAC, one I/V each for balanced output
but as mentioned summing the 2 DAC into 2 I/V isn't a simple, safe option
the ADA4899 specs look even better but I would want 'scope and active probes costing as much as a (cheap) new car to see that it ain't misbehaving at 600 MHz
Last edited:
Well I had an AD811 (and AD829) on a PCM63. I even had it powered by AD811 wideband voltage regulators. The OPA627 was definitely better. I think it's something to do with the CB DI process, but I have no proof as to why the 627 sounds better than everything else.
I'm with you, with the OPA627 sounding better than any other opamp, but for me it was as the output buffer after the I/V stage where it shines, not the I/V itself.
Cheers George
Yes, I/V is demanding position due to the mixed analog and RF signal. Although the 627 still beat the 811 in my results.
I do believe from memory I had a 1000pF WIMA FKP shunt at the I/V i/p to GND.
I do believe also I had a matching 1000pF and R2R and FB resistor equivalent resistance in the +ve input to ground on the VFB op-amps.
It helps to balance the impedances at the op-amp inputs to minimise common mode to differential conversion distortion and IM products.
Question: What's best for the input of an I/V opamp, bi-polar or fet??
It seems Jocko Homo who used to frequent these pages before I think getting banned for being so affable, uses bi-polar in his "best ever" I/V stage he designed, which no one I know has ever seen the circuit of.
Cheers George
It seems Jocko Homo who used to frequent these pages before I think getting banned for being so affable, uses bi-polar in his "best ever" I/V stage he designed, which no one I know has ever seen the circuit of.
Cheers George
Last edited:
Settling time and slew rate do not make an op-amp "fast" necessarily. I would heed the advice of jcx, who really knows what he's talking about. Even if you don't care about the input stage of the ADA4898 and its possible benefits for this application, it has good specs for this anyway and is unity gain stable.
Well a transistor beats all op-amps quit easily. But knowing which ones and how to implement for best results is the secret.
Contrary to another post you received the op-amp does not need to be unity gain stable in the I/V position.
You can use an OPA637 if you understand how to keep the noise gain above 5 from DC to the HF corner where the forward gain drops below 5.
once again the 'highly linear input stage' is poorly advertised but is exactly what's needed in I/V to maintain linearity in the face of the mV switching glitches and over the feared settling time dynamics'
the ADA4898 'highly linear input stage' obsoletes any older comparison using ordinary BJT diff pair inputs
just look at the HD3 plots in the DS (the HD2 is likely limited by the power pin location and Class AB output stage )
the ADA4898 'highly linear input stage' obsoletes any older comparison using ordinary BJT diff pair inputs
just look at the HD3 plots in the DS (the HD2 is likely limited by the power pin location and Class AB output stage )
I did not say that it must be unity gain stable, first of all. It just makes it easy to apply without having to consider the noise gain, as you have pointed out.
What is a transistor? What is an op-amp? Why can't you build an op-amp out of discrete transistors? You talk about this discrete transistor "sound" superiority as if it's engineering canon, except it's not. You should let the OP know that this is your own opinion based on pseudoscience.
Apologies, I don't know you or what you know so nothing personal. I just took it you implied an op-amp needed to be unity gain stable for I/V based upon you mentioning that the op-amp was.
What is a transistor?
That's sort of like asking the same question about two different op-amps that have virtually identical performance characteristics, built on the same processes but when put on a spectrum analyser have different harmonic sideband amplitudes and proportion. So you measured the THD+N and they were the same for both. But the spectrum analyser gives a different picture. Which one is correct, which one is wrong? If you lived in a day when you only had THD+N measurement and you had some quack who developed his own spectrum analyser, and he was telling the world they were not the same, all you would say his opinion is based on pseudoscience because current day measurements don't show a difference. Correct?
Have you investigated any distortion mechanisms that are never mentioned or measured in the current state of the art world of audio engineering?
There's mention in op-amp design white papers about thermal gradients caused by varying power dissipation in the output stage when driving loads. How designers are careful about layout to minimise its affects on other gain stages. An op-amp usually has one die with all of the transistors on it.
Does anyone measure for this in any tests?
That's a lot different to a discrete design that has a lot more thermal insulation between gain stages. I.e. PCB with copper tracks and relatively large physical space between components, usually air.
That's just one parameter. There's more.
"You talk about this discrete transistor "sound" superiority as if it's engineering canon, except it's not." Is that your opinion?
"You should let the OP know that this is your own opinion based on pseudoscience." Do you have proof that it is?
pseudoscience can be just that to people who don't understand the science.
Last edited:
Good enough for DSL but not good enough for line level audio apparently.
The thermal gradient "issue" you mention can be addressed in several ways, composite amplifiers for one. There are circuits on this forum comprised of IC op-amps and no discrete transistors that have incredible levels of performance to the point where there barely exists measurement equipment capable of characterizing them.
It is a logical fallacy to ask someone to prove non-existence. You're the one making the claims, so proof of existence of these phenomena needs to come from you. I will be happy to read any halfway legitimate paper you can link establishing the audibility of these top secret parameters or distortions of yours.
Not holding my breath.
Even if we play along, do you really think you've stumbled on something that the rest of the engineering world hasn't already considered? I am sure there are aerospace, defense, medical guys that would love to know about these inherent issues with IC op-amps that they aren't currently considering that you can hear with the human ear. Shades of Bybee. Yeah... I have these super awesome quantum purifiers, top secret tech, I just prefer to sell to audiophiles instead of license it for a bazillion dollars to industries that actually matter.
Anyway, the OP has ordered his perfectly suitable op-amps for job.
The thermal gradient "issue" you mention can be addressed in several ways, composite amplifiers for one. There are circuits on this forum comprised of IC op-amps and no discrete transistors that have incredible levels of performance to the point where there barely exists measurement equipment capable of characterizing them.
It is a logical fallacy to ask someone to prove non-existence. You're the one making the claims, so proof of existence of these phenomena needs to come from you. I will be happy to read any halfway legitimate paper you can link establishing the audibility of these top secret parameters or distortions of yours.
Not holding my breath.
Even if we play along, do you really think you've stumbled on something that the rest of the engineering world hasn't already considered? I am sure there are aerospace, defense, medical guys that would love to know about these inherent issues with IC op-amps that they aren't currently considering that you can hear with the human ear. Shades of Bybee. Yeah... I have these super awesome quantum purifiers, top secret tech, I just prefer to sell to audiophiles instead of license it for a bazillion dollars to industries that actually matter.
Anyway, the OP has ordered his perfectly suitable op-amps for job.
Last edited:
Not sure what DSL has to do with audio. People here seem to go off on all of these irrelevant tangents such as the other day, LIGO, op-amps and audio being mentioned and some comparisons drawn. If they are good enough for LIGO they are good enough for audio implication. I mean, what nonsense.
And now you go on about addressing the thermal gradient issue, totally irrelevant to the discussion. My question was, does anyone test for it?
There is the allegation by you that I am involved in pseudoscience and that "You talk about this discrete transistor "sound" superiority as if it's engineering canon, except it's not."
So I go on to explain a mechanism in IC op-amps that is not tested for and impacts upon them much more so than discrete designs, and you come back with irrelevant off topic post about how to design around it. So by implication you are confirming it may be an issue that impacts upon IC op-amps therefore contradicting yourself as to the differences, or lack thereof between discrete and IC op-amp performance.
Regardless of your contradictions, there are white papers on this subject by IC op-amps designers. It is never tested for. So it becomes an unknown variable as to how well the IC designer considered and addressed it and how much variation is there across op-amps designs on implementation effectiveness.
Then you claim "You're the one making the claims, so proof of existence of these phenomena needs to come from you. Does it? Why? Is there a law that requires me to back up my assertions with proof? Will I be arrested and tortured if I make an assertion without providing evidence?
I find your arguments raised and communication very anti-intellectual, petulant and extremely poorly formed in logic. I'm afraid that we communicate at different levels and there may not be much common ground to be gained by my further discussion with you.
I am glad we have reached a mutually acceptable agreement on something. I am also very pleased with the undertaking you have made and hope that you will honour that commitment.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.