This is a study of measured differences if one swaps up the TI NE5532p dual IC op-amp in a Fosi V3 Mono amplifier with a "discrete" opamp from Sparkos. It is a test requested in the youtube comments of my video test of the same using Douk A5 and Sonic Imagery op-amp. The Sparkos SS3602 retails for US $79.80 but I found a discount coupon that brought the cost down to $74.33 with shipping included.
Per Fosi documentation, there are three op-amps. The one at the bottom is only for unbalanced/balanced conversion. That leaves two op-amps to swap. Company says if you have only one, to change the #2, middle right to impact the tonality. So that is what I did as you see in the above picture. For clarity, this is a mono amplifier so the two ICs are dedicated to the same channel.
The Fosi V3 Mono with its balanced input was quite stable in operation, allowing me to borrow some of my review measurements. I did however repeat a couple as well (which confirmed the same).
In this second assessment of opamp rolling, I tried to address some of the criticism of the last test as you see below. Hopefully it completes a picture as this is the forth or fifth such testing.
Sparkos SS3306 Opamp and Fosi V3 Measurements
For my first test, I ran a full power sweep with the stock NE5532p and then with SS306. This allows us to see the noise+distortion at all power levels, accomplishing a number of things at once:
As we see, the results (subject to run to run variation) is essentially identical. Performance is most likely dominated by the class D output stage, not the input buffer/gain stage.
Question was asked about frequency response, thinking that explains some of the perception of tonality difference in uncontrolled, subjective evaluations. Here is that:
Again, it is the output class D amplifier IC which generates this response, not the front-end op-amps. The latter likely has much flatter and wider response. With identical frequency response, no tonality difference can be explained or justified.
A common retort was running only "static" tests. Never mind that a sine wave is not a static signal but time varying by a mathematical formula. Regular readers of ASR know that we have a good stand in, in the form of 32-tone test signal. Most have seen those 32 spikes in frequency domain but not that many know what it looks like in time domain: the thing that the amplifier "sees:"
Let's agree that this is a highly complex signal. Not only that, it has ton more energy in higher frequencies than real music, with full amplitude tones going to 20 kHz. Any worry about "slew rate" then would be tested here. Let's overlay the stock 5532 opamp and Sparkos Labs SS3602:
The response is again, identical. Yes, there are individual noise/distortion spikes that are a hair different but they won't amount to anything and certainly won't explain any perceived audio difference.
Another question was that of ultrasonics and worries of oscillations. My audio analyzer has a bandwidth of 1 MHz so I tested that far:
I forgot to take out my high-resolution AES-17 filter so the energy is brought down gradually from 40 kHz on. Since both measurements are impacted the same way, we can still compare. Once more, we see no meaningful difference.
Discussion and Conclusions
All of this came about because I predicted and promised that op-amp swapping won't make a difference when I tested the Douk A5 amplifier. How did I know that? For the following reasons:
1. The op-amp stage uses a ton of feedback. Since they have an easy time versus the amplifier output stage, it is trivial to get this pre-amp portion to far exceed the performance that is needed in terms of noise, distortion and bandwidth.
2. Some replacement op-amps may be designed as to produce better performance. But that requires circuit modification (e.g. to change level of feedback). Without it, you won't see that impact.
3. Turns out the NE5532p, despite its age, is an excellent opamp for audio applications. While there are thousand of opamps out there, they are solving different problems than the plain application we have.
As to outcomes of different sound people hear in swapping op-amps, basic science of controlled testing says a) these tests have to be fully controlled (blind, level matched) and b) run instantaneously. You can't listen to an amp, tear it down, put in a different opamp and expect to have any kind of valid test. Our knowledge of engineering, confirmed by measurements and science of psychoacoustics says there is no audible difference.
Note that mistakes can be made that do cause audible differences. If the new opamp causes oscillation for example, that could be audible. If you change volume, as a pair of youtubers did, of course you can perceive a difference. So it is not that it is impossible to hear something. It is that you need to do your homework in any such swaps and very preferably, measure to see what has happened. This rules out any audiophile testing, no matter how golden you think your ears are.
Finally, our instrumentation as always is exceedingly accurate, able to dig way, way below threshold of hearing to find differences. So it is not that our measurements are blind to differences. They are not. We can see and evaluate the changes and confidently confirm lack of audibility in such swaps.
As to Sparkos SS3602, it works and doesn't harm anything. But doesn't improve your amplifier either for reasons stated above.
-----------
As always, questions, comments, recommendations, etc. are welcome.
Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
Per Fosi documentation, there are three op-amps. The one at the bottom is only for unbalanced/balanced conversion. That leaves two op-amps to swap. Company says if you have only one, to change the #2, middle right to impact the tonality. So that is what I did as you see in the above picture. For clarity, this is a mono amplifier so the two ICs are dedicated to the same channel.
The Fosi V3 Mono with its balanced input was quite stable in operation, allowing me to borrow some of my review measurements. I did however repeat a couple as well (which confirmed the same).
In this second assessment of opamp rolling, I tried to address some of the criticism of the last test as you see below. Hopefully it completes a picture as this is the forth or fifth such testing.
Sparkos SS3306 Opamp and Fosi V3 Measurements
For my first test, I ran a full power sweep with the stock NE5532p and then with SS306. This allows us to see the noise+distortion at all power levels, accomplishing a number of things at once:
As we see, the results (subject to run to run variation) is essentially identical. Performance is most likely dominated by the class D output stage, not the input buffer/gain stage.
Question was asked about frequency response, thinking that explains some of the perception of tonality difference in uncontrolled, subjective evaluations. Here is that:
Again, it is the output class D amplifier IC which generates this response, not the front-end op-amps. The latter likely has much flatter and wider response. With identical frequency response, no tonality difference can be explained or justified.
A common retort was running only "static" tests. Never mind that a sine wave is not a static signal but time varying by a mathematical formula. Regular readers of ASR know that we have a good stand in, in the form of 32-tone test signal. Most have seen those 32 spikes in frequency domain but not that many know what it looks like in time domain: the thing that the amplifier "sees:"
Let's agree that this is a highly complex signal. Not only that, it has ton more energy in higher frequencies than real music, with full amplitude tones going to 20 kHz. Any worry about "slew rate" then would be tested here. Let's overlay the stock 5532 opamp and Sparkos Labs SS3602:
The response is again, identical. Yes, there are individual noise/distortion spikes that are a hair different but they won't amount to anything and certainly won't explain any perceived audio difference.
Another question was that of ultrasonics and worries of oscillations. My audio analyzer has a bandwidth of 1 MHz so I tested that far:
I forgot to take out my high-resolution AES-17 filter so the energy is brought down gradually from 40 kHz on. Since both measurements are impacted the same way, we can still compare. Once more, we see no meaningful difference.
Discussion and Conclusions
All of this came about because I predicted and promised that op-amp swapping won't make a difference when I tested the Douk A5 amplifier. How did I know that? For the following reasons:
1. The op-amp stage uses a ton of feedback. Since they have an easy time versus the amplifier output stage, it is trivial to get this pre-amp portion to far exceed the performance that is needed in terms of noise, distortion and bandwidth.
2. Some replacement op-amps may be designed as to produce better performance. But that requires circuit modification (e.g. to change level of feedback). Without it, you won't see that impact.
3. Turns out the NE5532p, despite its age, is an excellent opamp for audio applications. While there are thousand of opamps out there, they are solving different problems than the plain application we have.
As to outcomes of different sound people hear in swapping op-amps, basic science of controlled testing says a) these tests have to be fully controlled (blind, level matched) and b) run instantaneously. You can't listen to an amp, tear it down, put in a different opamp and expect to have any kind of valid test. Our knowledge of engineering, confirmed by measurements and science of psychoacoustics says there is no audible difference.
Note that mistakes can be made that do cause audible differences. If the new opamp causes oscillation for example, that could be audible. If you change volume, as a pair of youtubers did, of course you can perceive a difference. So it is not that it is impossible to hear something. It is that you need to do your homework in any such swaps and very preferably, measure to see what has happened. This rules out any audiophile testing, no matter how golden you think your ears are.
Finally, our instrumentation as always is exceedingly accurate, able to dig way, way below threshold of hearing to find differences. So it is not that our measurements are blind to differences. They are not. We can see and evaluate the changes and confidently confirm lack of audibility in such swaps.
As to Sparkos SS3602, it works and doesn't harm anything. But doesn't improve your amplifier either for reasons stated above.
-----------
As always, questions, comments, recommendations, etc. are welcome.
Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
