Fool`s gold indeed. The unsure mind is addicted to clear, black and white answers.
Measurements can provide these with ease, if needed. Just look at figure X on the page and you`ll see what I mean.
With critical listening it`s the other way around, the error, "noise", is buried in the signal and hard to point out: it`s all mixed up with music.
The LM6171 is really a video/instrumentation chip---not really suited for audio. Its ridiculously high slew rate (3600V/uSec) is way overkill for audio, and achieves this at the expense of high voltage noise (12nV√Hz) and THD (~-90 db).
You missed the point. It's not about sounding different, it's about proper circuit layout to avoid signal degradation. Every circuit will have different requirements for how it's laid out based on physical as well as electrical specs. I usually go with whatever the chip manufacturer recommends for physical layout , power supply decoupling, impedance scaling, etc. After I'm done building a circuit I'll test it for any anomalies and fix any I find. There isn't any magic one size fits all answer to your question.
Mike
Once I built a headphone amp on a prototype board with a careful layout and bypass caps right across every power in. It used a BUF634 as an output buffer. Swapped around different opamps and found the LM6171 was oscillating at around 100Mhz! Decided to settle on the much slower AD744 😀
Interesting comment from a person who's addicted to the pure black and white position that measurements are the only thing that matters. As you say, its pure fools gold. 😛
IMO much of it has to do with operating point. For example if we look at the THD+N vs output voltage charts for a LM4562 we see V shaped charts. Not all parts have the same plots in this respect, so unless both parts have similar sweet spots and the circuit is optimized accordingly, good chance the end result sounds different.
Of course if you've followed the mudfest of the previous several pages, prepare for an onslaught by the malcontents for your blasphemy in claiming that you hear differences, lol.
When you retrofitted the E-MU 1820m what capacitor series, brand and values did you use for ADC FILT caps?
What about the DAC FILT+, VREF and the two AOUT+/- caps?
Any chance that you measured the performance and could post the spectrum? I ask since another forum member used OPA1612 and Nichicon FPcap family and achieved really good results with an E-MU 1616m: #7104
So I am trying to refine my choices of capacitors and op-amps before I make further modifications to mine. LM4562 is attractive cost-wise. The OPA1612 is expensive but the results look fantastic.
What about the DAC FILT+, VREF and the two AOUT+/- caps?
Any chance that you measured the performance and could post the spectrum? I ask since another forum member used OPA1612 and Nichicon FPcap family and achieved really good results with an E-MU 1616m: #7104
So I am trying to refine my choices of capacitors and op-amps before I make further modifications to mine. LM4562 is attractive cost-wise. The OPA1612 is expensive but the results look fantastic.
I suspect strongly that your problem could have been solved by connecting a small capacitor (100pF to 1nF) from the output of the LM6171 to its inverting input.
Read the following:
https://www.analog.com/media/en/technical-documentation/application-notes/an47fa.pdf
https://www.analog.com/media/en/technical-documentation/application-notes/an21f.pdf
https://www.analog.com/media/en/technical-documentation/application-notes/an18f.pdf
That`s an excellent example of why one should always double check with a measurement, before critical listening to make an opinion which opamp sounds best.
Let`s not to leave out that one can not hear the oscillation itself of course, but sq might very well degrade nonetheless.
ymmv etc
Which begs the question: what does oscillation sound like? One cannot hear supersonic oscillation fundamentals, but surely they must cause some sort of audible artifact: what is it?
We had an LPUHP oscillating at ~10MHz lately. Was fixed by very small caps parallel to the feedback resistors. I can't tell you how it sounds, because we fixed it before listening. But I can at least tell you that measured harmonics wen up by 20-30dB.
FPCAP catalog available at: https://www.nichicon.co.jp/english/products/pdf/2012fpcap_catalog_all.pdf
The Precautions section toward the beginning of the catalog is rather interesting.
HF oscillation on an opamp or linear reg almost always manifests itself as a white or pink noise sound when heard over speakers. I don’t quite know the mechanism that causes a 2MHz 1V pk-pk oscillation to sound like this, but that’s my experience. Ditto 4 MHz and 300 kHz oscillation.
Bonsai, thanks for that.
Hope the usual stirrers don't accuse you of claiming to hear 4MHz since no human on earth can hear over 20kHz.
Hope the usual stirrers don't accuse you of claiming to hear 4MHz since no human on earth can hear over 20kHz.
Well, it is in the audio path. C0G seems like a pretty good choice, although I ran across some AP guy's article that advised to use 50v or 100v NPO/C0G because lower voltage were not as linear.
Last edited:
On a previous version of my headphone amp one side was oscillating slightly and one was not. The oscillating side was quiet but had slightly lower output. It wasn't a crazy high frequency either, way less than 1Mhz if I recall.