I cant believe we are still debating op amps like 20 years later...
LM4562 is good as far as I am concerned.
AD7...is very good, better but it is colored.
ad8599 is my preferred of the group of the best LM, BEST OPA 21XX series NE, etc including opa627.
There are other AD opamp that are very good to the 8597, but it is a matter of my preference. I can't remember which one but is in a box in the basement
The ad8597ARZ is very nice to use, input 1k Z max, not a problem with V out dacs. plenty of power to drive a 100k input impedance volume control directly. Sound is as good as it gets with opamps, quest over. https://www.analog.com/media/en/technical-documentation/data-sheets/AD8597_8599.pdf
Edit: I remember, the ad797, it is very strong, very well sounding, but to me I prefer as a matter of taste only the ad8597.
If I could live with only 2 opamps it would be ad797 and ad8597, depending of the music playing, the ad8597 is more even no matter what is going on with the music, the ad797 is ultra focused, high detail, and can affect some music in some way, but it handles it well, just not as neutral/involved as the ad8697
LM4562 is good as far as I am concerned.
AD7...is very good, better but it is colored.
ad8599 is my preferred of the group of the best LM, BEST OPA 21XX series NE, etc including opa627.
There are other AD opamp that are very good to the 8597, but it is a matter of my preference. I can't remember which one but is in a box in the basement
The ad8597ARZ is very nice to use, input 1k Z max, not a problem with V out dacs. plenty of power to drive a 100k input impedance volume control directly. Sound is as good as it gets with opamps, quest over. https://www.analog.com/media/en/technical-documentation/data-sheets/AD8597_8599.pdf
Edit: I remember, the ad797, it is very strong, very well sounding, but to me I prefer as a matter of taste only the ad8597.
If I could live with only 2 opamps it would be ad797 and ad8597, depending of the music playing, the ad8597 is more even no matter what is going on with the music, the ad797 is ultra focused, high detail, and can affect some music in some way, but it handles it well, just not as neutral/involved as the ad8697
Last edited:
To destroy some dreams here: The AD797 ($10) definitely has bigger distortions than the LM4562 ($1.60). Distortions must be paid.
Are you comparing 'specs' or how it sounds?
Because 'specs' means nothing, feedback can correct the THD and any thd under 80 db is impossible to hear.
I have the LM4562, it doesn't sound better than any other regular opamp.
If I purchased the ad797 it was with careful analytics, actually i never look at thd figures
To destroy some dreams here: The AD797 ($10) definitely has bigger distortions than the LM4562 ($1.60) in the 10-20kHz region. Distortions must be paid.
Seriously, what is wrong with you?
Every one of your posts is about how good the 5534 or 4562 is, and yet you couldn’t even get an LM4562 stable despite the fact that it is very easy.
The AD797 is lower distortion in several circuits. There is a best fit for every application. Please just drop it, because you’re wrong.
I do not compare listening results or specs, but measurement results of course.
LM4562: 0.0007% @20kHz
AD797: 0.0015% @20kHz
To destroy some dreams here: The AD797 ($10) definitely has bigger distortions than the LM4562 ($1.60) in the 10-20kHz region. Distortions must be paid.
I've found the opposite on an APx-555. The APx-555 is not the cleanest possible analyzer available around 10-20kHz (some custom built fixed frequency oscillators and notch filters can surpass it), but it's probably the lowest residual programmable test set. Using this analyzer, the LM4562 can be shown to exhibit some thermally induced (post warmup) high order distortion that the AD797 does not seem to exhibit - thermally load the 4562 package and this problem goes away.
Overall, the AD797 seems to have less distortion than any monolithic amplifier I've measured, basically buried in the APx residual to some extent or another. This analyzer gets sloppy above 10kHz, so both amplifiers seem to be buried in analyzer residual, which is -140dBc or less. So, aside from the thermally induced high order distortion effects of the 4562, it's not so easy to reliably distinguish these amplifiers, since they're both basically cancelling analyzer residual and yielding false numbers.
What are you using to measure this? Don't tell me "listening tests" as these distortion levels are well below -140dBc, and I find it difficult to imagine that this can be heard easily. I personally would like a better analyzer setup just to separate the op amp from the analyzer residual but then again, the APx-555 does pretty well around 2kHz, basically about -155dBc to -160dBc for 2nd and 3rd harmonic - pretty far down there.
I submit you're comparing circuit noise. 0.0015% is 15 parts per million, around 40dB higher than the actual distortion levels of an AD797 or LM4562 as observed using an FFT of an analyzer residual.
THD+N is a pretty useless test for actual distortion - get a setup that can take an FFT of the analyzer residual so you can look at the actual distortion products, and not the total noise power. On modern amplifiers, the distortion is buried deep into the noise, and can be separated by our ears and an FFT, but will not be shown by a THD+N number. That became obsolete 20-30 years ago.
The APx-555 is newer. Not sure what they're using inside but I suspect there are some composites.
Last edited:
I don't have the guts to open it up
Still, there are some external fixed frequency oscillators and notch filters that can beat the APx-555, but they're a pain in the rear to use. You have to do your own calibration for everything, including each harmonic number. The Linear AN-67 10kHz oscillator and Sam Groner's notch filter are on my 'to do' list but still, the APx-555 is pretty good. I also have a suspicion that if you're scraping that residual, your circuit is pretty close to "perfect" anyway.
They have a new upgrade that is supposed to fix up the HF ranges, so I might spring for that. Not sure how much nicer it'll be, but it will be an improvement.
Still, there are some external fixed frequency oscillators and notch filters that can beat the APx-555, but they're a pain in the rear to use. You have to do your own calibration for everything, including each harmonic number. The Linear AN-67 10kHz oscillator and Sam Groner's notch filter are on my 'to do' list but still, the APx-555 is pretty good. I also have a suspicion that if you're scraping that residual, your circuit is pretty close to "perfect" anyway.
They have a new upgrade that is supposed to fix up the HF ranges, so I might spring for that. Not sure how much nicer it'll be, but it will be an improvement.
You don;t need more than 1/10th of what your analyzer can do to be able to tell if any audio circuit on Earth is good enough for the most difficult audio test.It's just the industry set some standards which are completely useless.
I never needed a thd analyzer to make a circuit sound well.An oscilloscope is enough.
I never needed a thd analyzer to make a circuit sound well.An oscilloscope is enough.
Last edited:
That is probably true, although when measuring distortion near the analyzer residual, the DUT harmonics can cancel or add to the analyzer residual and make the numbers 'wrong'. So, for simple no-brainer results, it's really nice to have an analyzer that's about 10dB or more better than your circuits.
And, one should keep in mind that these are harmonic distortion numbers measured with sine waves - not something that makes sense in terns of audibility, but what makes sense in terms of how well we can measure things. IM distortion and harmonic distortion are inseparable, but it's hard to measure IM with a low residual. So, you can measure HD instead and assume (correctly) that IM will also be low. However, it's completely reasonable to expect higher IM numbers from a circuit that produces low HD numbers... thus it seems somewhat reasonable to worry about HD numbers that seem to be inaudible - none of this happens with complex signals and I suspect the actual degradation is worse than with a simple sine wave.
And, one should keep in mind that these are harmonic distortion numbers measured with sine waves - not something that makes sense in terns of audibility, but what makes sense in terms of how well we can measure things. IM distortion and harmonic distortion are inseparable, but it's hard to measure IM with a low residual. So, you can measure HD instead and assume (correctly) that IM will also be low. However, it's completely reasonable to expect higher IM numbers from a circuit that produces low HD numbers... thus it seems somewhat reasonable to worry about HD numbers that seem to be inaudible - none of this happens with complex signals and I suspect the actual degradation is worse than with a simple sine wave.
I've found the opposite. I've optimized an old mike amp (the Symetrix SX202) using an older AP-2322 analyzer to measure each change, and while each small adjustment didn't amount to much, repeating the process over a period of months with numeric verification along the way to verify whether a change was beneficial or not ended up in a clearly audible improvement of the amplifier.
I was able to re-compensate the SSM2015 amplifier to move the knee where distortion rises with frequency from 3kHz to about 30kHz, as well as reducing distortion levels overall. This materially improved the sound of the circuit, and i'm quite sure that if I had to rely only on listening tests at each step, I would have never been able to fully optimize the design.
The AP-2322 is pretty primitive compared to modern analyzers, but still, the principle still applies - being able to measure a circuit can allow you to see what is important and what is not, and can help guide design decisions.
An AP is useful, but yes, not cheap. Still, you can get close (and actually surpass one number-wise) with fixed oscillators and fixed notch filters as well as a soundcard system to do an FFT of the residual. Not sure what people find most pleasant, but it's worth checking out.
Compared to the old days, time marches on. Back in the '60s, 1% distortion was "inaudible". This has progressed, probably by 20dB per calendar decade, and we're at a place where our source material is good enough that the rest of the chain will sound different if it's not high performance. 20 years ago, I never saw a circuit that beat a 1 ppm (120dBc) distortion floor, but these days, if a circuit I build performs that poorly, I look for something that has blown up, or wonder whether the PCB was static zapped.
Life moves forward, and while it seems convenient to say that "enough is enough", incremental improvements still seem to provide material benefits. I still think we're not done yet, and more can be improved. Or, this "perfection" could perhaps be done more cheaply... More audio circuit design work and improvements are needed IMHO!
Compared to the old days, time marches on. Back in the '60s, 1% distortion was "inaudible". This has progressed, probably by 20dB per calendar decade, and we're at a place where our source material is good enough that the rest of the chain will sound different if it's not high performance. 20 years ago, I never saw a circuit that beat a 1 ppm (120dBc) distortion floor, but these days, if a circuit I build performs that poorly, I look for something that has blown up, or wonder whether the PCB was static zapped.
Life moves forward, and while it seems convenient to say that "enough is enough", incremental improvements still seem to provide material benefits. I still think we're not done yet, and more can be improved. Or, this "perfection" could perhaps be done more cheaply... More audio circuit design work and improvements are needed IMHO!