It certainy has to be done in a way such that the listener has no clues whatsoever about the device being listened to. Opinions may differ, but I profoundly believe this to be imperative. It is also why I don't count the two sighted devices in Lvandoorn's results as objective.
What bothers me about opamp rolling is that circuits are designed with the chosen opamp in mind. Then comes someone along and swaps in an opamp with dfifferent needs, maybe with one that isn't unity gain stable, or one that needs compensation or whatever you can think of that makes it behave differently.
I was (actually still am) temped to try something like test this myself. But then I have to admit to myself that if I can't identify a NOS DAC from an OS DAC without the use of an oscilloscope (and these are huge measurable differences), why then would I be able to tell the difference from a NE5532 and two OPA627s on a converter PCB? (This one I tried years ago, and no, I couldn't tell for sure if there was a difference).
What bothers me about opamp rolling is that circuits are designed with the chosen opamp in mind. Then comes someone along and swaps in an opamp with dfifferent needs, maybe with one that isn't unity gain stable, or one that needs compensation or whatever you can think of that makes it behave differently.
I was (actually still am) temped to try something like test this myself. But then I have to admit to myself that if I can't identify a NOS DAC from an OS DAC without the use of an oscilloscope (and these are huge measurable differences), why then would I be able to tell the difference from a NE5532 and two OPA627s on a converter PCB? (This one I tried years ago, and no, I couldn't tell for sure if there was a difference).
Last edited:
Agreed. I don't think humans posess a detailed memory for sound, just a very coarse one.
I think over time, you are conditioning yourself to a new situation by repetition.
I usually can tell if something sounds wrong or bad, has a problem like a dry capacitor in a speaker, a bad fake Opamp or too low MP3 data rate, after very short listening. Same is to gear in a chain that is simply of low quality. For example a pickup, CD player, tuner or amp. No problem to pinpoint such problems. IMO only a matter of training your ears and some experience. Technical understanding helps, otherwise you are endangered to talk voodoo.
When it comes to "what is better" within high quality gear, it can take hours of listening, sometimes on different days and in the end may deliver no conclusive results. Any background noise can make serious listening impossible, what is why I prefer to do so late in the evening. Don't drink and listen!
This second kind of finding a better or worse, may often only manifest it self in very short passages of the listening material. If this event is repeated serveral times, it finaly get's clearer and can be isolated. To not fool my self I use an A-B switching device that makes changing out the DUT's a push of a button only. With a CD player, repeating critical, short passages is a good way to find out.
When it comes to "what is better" within high quality gear, it can take hours of listening, sometimes on different days and in the end may deliver no conclusive results. Any background noise can make serious listening impossible, what is why I prefer to do so late in the evening. Don't drink and listen!
This second kind of finding a better or worse, may often only manifest it self in very short passages of the listening material. If this event is repeated serveral times, it finaly get's clearer and can be isolated. To not fool my self I use an A-B switching device that makes changing out the DUT's a push of a button only. With a CD player, repeating critical, short passages is a good way to find out.
Just a remark, the two 'sighted devices' in the range I tested were not 'known' by my three fellow testers. So not the double blind gold standard but not too far from that. And the three 'best' opamps in our test were compared for quite a few hours. Not weeks for sure, but not short burst comparisons either. I could not bring myself to do this for the ones we liked less....
Some lucky ones possess this ability from birth. The rest of us have to painstakingly develop it and it often takes work and persistence. And also a state of mind: not everyone is inclined to audition after every minute change.
I know if you have absolute hearing it means you can tell a note without having to hear a reference note for comparison first. I did not know "photographic" audio memory was a thing.
In my younger days I returned a Philips DCC900 days after purchase because I thought it sounded defective. This device used a lossy compression algorithm called PASC (nowadays linked to Covid, but it used to mean Precision Adaptive Subband Coding) and could be considered equal to MPEG1 at 384 kpbs.
I wonder if I would have if those days were now.
In my younger days I returned a Philips DCC900 days after purchase because I thought it sounded defective. This device used a lossy compression algorithm called PASC (nowadays linked to Covid, but it used to mean Precision Adaptive Subband Coding) and could be considered equal to MPEG1 at 384 kpbs.
I wonder if I would have if those days were now.
I watched this from afar as someone interested but who does not use vinyl. I guess I was surprised at the results but also in a way not that surprised. I have long championed the TL071 series but credibility can evaporate when you suggest them. When you really are listening blind you have to base all impressions on what you hear and there is also much truth in comments that you generally have to listen and live with something for quite a while before forming a definitive view.
Getting mildly interested for a rainy Saturday afternoon I simulated the Pearl and created an inverse RIAA using component values from Stanley Lipshitz and Walt Jung taken from an old Wireless World magazine. Now the fun starts, recording the output of the Pearl (the simulation). With hindsight I should have fed more signal in and attenuated the main output but as it is the output levels were just within the -/+1 volt pk/pk output .wav limit of LTspice. The circuit was duplicated and ran in 'stereo'. E1 is set at just 0.06 meaning the output is attenuated by that factor.
For anyone interested here is the result and a 320kbps MP3 file is attached at the end. One channel output shown here.
There is an educational 741 model in LTspice that uses all discrete transistors based on the actual 741 topology, now that could be a challenge to use that but I suspect the sim would take forever to run. The standard opamp models work totally differently.
Anyhoo... this took about 25 minutes to run and complete.
One channel before duplication.
Getting mildly interested for a rainy Saturday afternoon I simulated the Pearl and created an inverse RIAA using component values from Stanley Lipshitz and Walt Jung taken from an old Wireless World magazine. Now the fun starts, recording the output of the Pearl (the simulation). With hindsight I should have fed more signal in and attenuated the main output but as it is the output levels were just within the -/+1 volt pk/pk output .wav limit of LTspice. The circuit was duplicated and ran in 'stereo'. E1 is set at just 0.06 meaning the output is attenuated by that factor.
For anyone interested here is the result and a 320kbps MP3 file is attached at the end. One channel output shown here.
There is an educational 741 model in LTspice that uses all discrete transistors based on the actual 741 topology, now that could be a challenge to use that but I suspect the sim would take forever to run. The standard opamp models work totally differently.
Anyhoo... this took about 25 minutes to run and complete.
One channel before duplication.
Thanks for all that work @Mooly! I'm in a similar situation, interested but not a vinyl listener. I put together a spreadsheet with some specs different from those in @Mark Johnson's spreadsheet. There's a correlation between input capacitance and preference but with only 5 of 9 op-amps having input capacitance specs, 2 of which only provide differential mode, I'm not drawing any firm conclusions. It would be nice if JRC put a little more data in the datasheet but so it goes...
Thanks Monk55, doctor visit went beautifully and I'm back at full strength. You'll notice that for this experiment, I chose not to include duplicate opamp-pairs in the blind listening test. If the first pair of WXYZ opamps sounded much much better than the second pair of identical WXYZ opamps, that might be noteworthy (so said a few armchair quarterbacks). But I ran out of patience after sourcing, assembling, and anonymizing fourteen SOIC opamps on DIP adapters. Seven unique pairs of jolly interesting opamp chips was about my limit.