Does Op-amp Rolling Work?

[PlanarSearcher]

Ignore your scope measurements it seems to be full of coupled (common mode) noise and 12 bit is not accurate enough.

Also ignore the 'claims' of improved sound quality.
Changing op-amps is NOT going to do that, regardless how many people claim to hear definite differences (ranging from subtle to night and day).
Well.... unless you put a really crappy in or one that is not suitable for the design/application in there.

I also had a similar thought: that an “out-of-the-box” solution could be some kind of “plug-in” component which, when replaced, produces a clearly noticeable effect.
I was also surprised (and even impressed) that they even provide tables showing how combinations of LPF and operational amplifiers (as the amplifier stage) can improve sound quality, add detail, and completely change the sound signature — and, of course, without any measurements as proof.
That really feels like “magic,” or something almost too good to be true.

 

[MAB]

In my humble opinion (although I only have an engineering degree in Computer Science — logic-level design rather than electronics), probably yes.

I am attaching photos, including annotated PCB images with labeled key components. The device is a Hires (at least according to the product description, typical Chi-Fi) portable audio player with a fairly simple architecture:

DAC ES9018Q2 → LPF (OP275, soldered) → OU (amplifier, DIP-8, replaceable).

As far as I understand the power path (please correct me if I am mistaken):
3.7 V → 13 V boost converter → 2× LDOs (DAC and OU supply rails).

I assume the PCB has a maximum of 4 layers.

I am also attaching some very primitive measurements taken with a 12-bit oscilloscope (tip = probe, sleeve = ground), without load (16 Ω resistor disconnected), mainly in the hope of showing a general “noise picture.”
Sorry if this is not particularly relevant. I still have the device (bought it for researching) and could redo the measurements more properly if further investigation is required.

The background story: I found several modding communities (I could post a summary of the most common modifications for this device if needed; replacing the LPF-stage OP275 (typically OP1612 or muses89xx, sometimes even opa1656) is one of the key recommendations) claiming that this modification transforms the device from a ~$2x player into something comparable to a ~$5xx one.
From my skeptical perspective, such claims sound rather extraordinary, especially because they are usually based on subjective impressions and are never supported by objective measurements.

Hard to mess up a low-pass filter with an OP275. If so, it is likely messed up with a 1612 too, or any other OpAmp that is useful for audio-frequency LPF. If it's messed up, changing the OpAmp won't change the design and/or layout, or the designer who messed it up.

I find modding communities to be unreliable, they typically lack any circuit or measurement or design knowledge. They are about rando-swap. Also, I like <$100 DACs, the expensive ones are most often not good performers. $2k players are absurd, unless they have a dozen channels of DSP. $5k is more like a Server, which is so much more.

 

[RandomEar]

I also had a similar thought: that an “out-of-the-box” solution could be some kind of “plug-in” component which, when replaced, produces a clearly noticeable effect.
I was also surprised (and even impressed) that they even provide tables showing how combinations of LPF and operational amplifiers (as the amplifier stage) can improve sound quality, add detail, and completely change the sound signature — and, of course, without any measurements as proof.
That really feels like “magic,” or something almost too good to be true.

Regular scopes are useless for these measurements: Amir does ther 50 mV test for most headphone amplifiers. A typical SNR result is above 80 dB, meaning the noise is below 5 µV. The lowest scale for the DHO804 is 500 µV/div. Also, the scope is 12 bit, whereas the audio output is at least 16 bit. All you'll ever see without a signal and load present is HF noise from the cables acting as antennas. And with a signal and load, you can only have a rough look and verify that the frequency you expect is present. You'll need a decent ADC and REW to get a grasp of THD and N.

All that stuff about "upgrades" and "improved sound quality" is typically just perceptive bias in action. Unless backed by some valid measurement, I would not believe any report of "improved XYZ" in audio circles. Also, I wonder why they never report worse sound after "upgrading"?

 

[PlanarSearcher]

Hard to mess up a low-pass filter with an OP275. If so, it is likely messed up with a 1612 too, or any other OpAmp that is useful for audio-frequency LPF. If it's messed up, changing the OpAmp won't change the design and/or layout, or the designer who messed it up.

I find modding communities to be unreliable, they typically lack any circuit or measurement or design knowledge. They are about rando-swap. Also, I like <$100 DACs, the expensive ones are most often not good performers. $2k players are absurd, unless they have a dozen channels of DSP. $5k is more like a Server, which is so much more.

Maybe I am mistaken (I have not measured the capacitors), but from a first quick look (based on the PCB photo), the LPF appears to be a 4th-order MFB filter with a cutoff around 22 kHz. In this design, it is placed after the ES9018Q2M DAC, and the output stage uses an OPA1622.


I have attached a photo (in this case, the OP275 was replaced with an OPA1612). HOWEVER, it seems that (ranging from subtle to night-and-day differences) there is essentially no reason to change the op-amp at all. It would probably make more sense to redesign the LPF cutoff frequency for this case to around 30–40 kHz. However, there are many other questions regarding this PCB’s quality, including ground layout and the placement of key components (DAC, LPF, op-amp stage, etc.). This board appears to be quite far from reference designs.

Because of this, it may have more impact on listening to FLAC files at 96 kHz and above, although I am not at all sure that the difference would actually be perceptible.

;-) provocative moment is:
what if we speak not about "day and night" but about very very small difference in case if LPF will be recalculated and all its components will be replaced with proper ones and then you placed exactly OPA1612 - and then
this could be (not even sure) at the level of “it seems like it,” rather than a guaranteed or consistently reproducible difference.

If it is audible at all, it would likely be perceived as:

• slightly more “dry / precise”
• slightly sharper or more clearly defined attacks
• slightly more “open” top end

And the main question is: can this kind of small “it seems…” difference be captured in measurements in order to evaluate it more technically? Or is it impossible to measure it because it does not actually exist?

 

[CedarX]

I am attaching photos, including annotated PCB images with labeled key components.

Based on your pictures, this is a Zishan Z2.
Zishan’s were not the best design around, but were cheap and very “mod-friendly”. You could get a lot of fun and significant improvements by modding the HW—like going from “it works fine” to “it smoked & no longer works”… That is until Zishan introduced their latest players, which were mostly ok (still terrible SW though), but also much more expensive… and the mod frenzy died.

 

[PlanarSearcher]

Based on your pictures, this is a Zishan Z2.
Zishan’s were not the best design around, but were cheap and very “mod-friendly”. You could get a lot of fun and significant improvements by modding the HW—like going from “it works fine” to “it smoked & no longer works”… That is until Zishan introduced their latest players, which were mostly ok (still terrible SW though), but also much more expensive… and the mod frenzy died.

Yes, it is the Zishan Z2, based on the photo — the v2 edition.

Pls see my post above: Core: Nevertheless, the main question remains the same: is there a measurable difference when changing the LPF op-amp in this circuit? It was clarified above that there is no “day and night” difference, but if a small one does exist, could such a hypothetical difference be captured and verified through measurements?

 

[MAB]

Maybe I am mistaken (I have not measured the capacitors), but from a first quick look (based on the PCB photo), the LPF appears to be a 4th-order MFB filter with a cutoff around 22 kHz. In this design, it is placed after the ES9018Q2M DAC, and the output stage uses an OPA1622.


I have attached a photo (in this case, the OP275 was replaced with an OPA1612). HOWEVER, it seems that (ranging from subtle to night-and-day differences) there is essentially no reason to change the op-amp at all. It would probably make more sense to redesign the LPF cutoff frequency for this case to around 30–40 kHz. However, there are many other questions regarding this PCB’s quality, including ground layout and the placement of key components (DAC, LPF, op-amp stage, etc.). This board appears to be quite far from reference designs.

Because of this, it may have more impact on listening to FLAC files at 96 kHz and above, although I am not at all sure that the difference would actually be perceptible.

;-) provocative moment is:
what if we speak not about "day and night" but about very very small difference in case if LPF will be recalculated and all its components will be replaced with proper ones and then you placed exactly OPA1612 - and then
this could be (not even sure) at the level of “it seems like it,” rather than a guaranteed or consistently reproducible difference.

Lots of changes.
Did you read my earlier post? Yes, rolling OpAmps will likely change the distortion and noise if you find a suitably high resolution analyzer. More likely you mess up the SMT and end up with a broken player. Your ears are orders of magnitude less sensitive than any audio analyzer, so no you won't hear it unless you end up breaking it.

If it is audible at all, it would likely be perceived as:

• slightly more “dry / precise”
• slightly sharper or more clearly defined attacks
• slightly more “open” top end

Why would the above perceptions be likely? Changes in signals are more like different frequency response, noise, distortion. The above somewhat flowery descriptions are the unfortunate byproduct of the HiFi industry, selling snake-oil by the gallon. They are hard to avoid, even on ASR. They are utterly devoid of meaning. We are all inundated by them, they make the changes sound irresistible, even if the changes are non-existstent.

And the main question is: can this kind of small “it seems…” difference be captured in measurements in order to evaluate it more technically? Or is it impossible to measure it because it does not actually exist?

These changes are measurable. I showed earlier. I can show you other things about this preamp I built, like there is a difference in Left and Right channel feedback resistor values for the input stage. The changes in feedback interact differently with each OpAmp; some see small benefit, some struggle, some don't care about the lower resistor values. All of this is below human audibility by orders of magnitude.

Our ears really suck. We should be happy, I personally don't want to have to listen to everybody in a restaurant chew their food.

 

[antcollinet]

How do I determine whether it makes sense to replace a soldered OP275 when it is used as a low-pass filter (LPF) after an ES9018Q2 chip?

One of the easiest questions to answer. You determine by listening to the very talented engineers who have taken time out of their day to advise you - and simply..... do not.

You can't make any audible improvement, you can only make things worse - or break your player.

Core: Nevertheless, the main question remains the same: is there a measurable difference

We can measure all that we can hear. But we can't hear all that we can measure - not by orders of magnitude.

 

[PlanarSearcher]

You determine by listening

Well, I am just trying to compare two sides’ positions. Maybe I also lost the topic: OP275 (not a discrete op-amp) vs. the top discrete op-amps mentioned in this video (hope, may be mistaken, “discrete” is not meaningful in this case).

I found this "Hell Yeah Rolling Opamps Makes Your Amps Sound Better! (Hear It For Yourself!)"

from GR-Research.

as the quintessence of what hundreds of modders communities across different geographical regions claim.
(unfortunately, 24 minutes of words, no fairly real proofs but in this case there was something more).
To save the time, some useful hints: skip the first 7 minutes of blabla, then listen to declarations about discrete op-amps in this case, and from 16:30 begins something like a “measurement approach/methodology”.

Some claims from this video:

"there's a lot of those flat Earth guys out there that think because they measure the same that they've got to sound the same.
They could not be more wrong."

"just because you can go in and measure the distortion levels or the noise floors and think, "Well, if it measures the same, it's going to sound the same."

"They're not looking at everything else that is involved in how those things are made, because what they're doing is they're basically a voltage stepping device.
They're stepping up voltage. So it's adding gain. So there's opportunity there for a lot of things to happen and a lot of ways to change the way it sounds,
by the way it's adding that gain and the quality of everything in the path."
...
and finally: "there is a huge variance in sound between all those different op-amps"

In this video were considered: Burson Vivid 7, Sparkos (dual discrete op-amp SS3602), etc...

Small humble analysis:

"Higher resolution" – I hope at least this could be measured in the range 20–20 kHz?

"Smoother, lusher, more richness" = I have no clue if it is even possible to calculate somehow.

"More harmonic structure" – this one I guess is possible to capture and reflect?

"More involvement" – also no clue what it is.

Accordingly to their statements: Final declaration about concrete discrete op-amp (but it is still an op-amp!):


one amp: "a little tighter control in the bottom end, a little more definition there and control down low"

another amp:
"a little more relaxed" – no clue what that is

"the vocals were richer" – this I suppose is in audible range and could be reflected / taken into account

"the sound stage was bigger" – AFAIK only indirectly via a bundle of parameters, could also be somehow compared.


Regarding experts – he mentioned himself, Hobbs, Ron – frankly I do not know anyone, including their labs, but they make it public to discuss.


What they did in the video from 16:30 onward is that they try to estimate perception from the listener’s point of view in a test audio room.

 

[Sokel]

Well, I am just trying to compare two sides’ positions. Maybe I also lost the topic: OP275 (not a discrete op-amp) vs. the top discrete op-amps mentioned in this video (hope, may be mistaken, “discrete” is not meaningful in this case).

I found this "Hell Yeah Rolling Opamps Makes Your Amps Sound Better! (Hear It For Yourself!)"

from GR-Research.

as the quintessence of what hundreds of modders communities across different geographical regions claim.
(unfortunately, 24 minutes of words, no fairly real proofs but in this case there was something more).
To save the time, some useful hints: skip the first 7 minutes of blabla, then listen to declarations about discrete op-amps in this case, and from 16:30 begins something like a “measurement approach/methodology”.

Some claims from this video:

"there's a lot of those flat Earth guys out there that think because they measure the same that they've got to sound the same.
They could not be more wrong."

"just because you can go in and measure the distortion levels or the noise floors and think, "Well, if it measures the same, it's going to sound the same."

"They're not looking at everything else that is involved in how those things are made, because what they're doing is they're basically a voltage stepping device.
They're stepping up voltage. So it's adding gain. So there's opportunity there for a lot of things to happen and a lot of ways to change the way it sounds,
by the way it's adding that gain and the quality of everything in the path."
...
and finally: "there is a huge variance in sound between all those different op-amps"

In this video were considered: Burson Vivid 7, Sparkos (dual discrete op-amp SS3602), etc...

Small humble analysis:

"Higher resolution" – I hope at least this could be measured in the range 20–20 kHz?

"Smoother, lusher, more richness" = I have no clue if it is even possible to calculate somehow.

"More harmonic structure" – this one I guess is possible to capture and reflect?

"More involvement" – also no clue what it is.

Accordingly there statements: Final declaration about concrete discrete op-amp (but it is still an op-amp!):


one amp: "a little tighter control in the bottom end, a little more definition there and control down low"

another amp:
"a little more relaxed" – no clue what that is

"the vocals were richer" – this I suppose is in audible range and could be reflected / taken into account

"the sound stage was bigger" – AFAIK only indirectly via a bundle of parameters, could also be somehow compared.


Regarding experts – he mentioned himself, Hobbs, Ron – frankly I do not know anyone, including their labs, but they make it public to discuss.


What they did in the video from 16:30 onward is that they try to estimate perception from the listener’s point of view in a test audio room.

[Cello eerie-ominous in D-minor/C-sharp minor]

 

[PlanarSearcher]

[Cello eerie-ominous in D-minor/C-sharp minor]

Sorry, I just missed the core of the message — did you mean that their “test clips” are not relevant?

I personally missed what exactly they did with microphone measurements in the test audio room and the reported values. However, above I have listed my thoughts on what could potentially be measured additionally based on his statements. So I am mainly interested in this: what if we take those ideas and enhance the measurement methodology, while also trying to take the listener into account?

I find it strange that so many people, independently of location all over the world, report the same impressions — even in different languages, and often without much overlap.

From a statistical point of view it cannot simply be a random coincidence. In my humble opinion, I am not claiming anything, but rather asking whether it would be possible to improve the measurement methodology and whether such effects can actually be measured at all — this is the main question.

In the video above, it is also interesting that, besides loudspeakers, high-resolution headphones were mentioned, on which, according to their conclusion, the difference would be audibly distinguishable.

 

[LTig]

Well, I am just trying to compare two sides’ positions. Maybe I also lost the topic: OP275 (not a discrete op-amp) vs. the top discrete op-amps mentioned in this video (hope, may be mistaken, “discrete” is not meaningful in this case).

I found this "Hell Yeah Rolling Opamps Makes Your Amps Sound Better! (Hear It For Yourself!)"

from GR-Research.

You are new to ASR, so you may have missed all the science based critic on Danny's videos. All his listening is without proper controls and hence baseless. 'Nough said ...

 

[antcollinet]

Well, I am just trying to compare two sides’ positions. Maybe I also lost the topic: OP275 (not a discrete op-amp) vs. the top discrete op-amps mentioned in this video (hope, may be mistaken, “discrete” is not meaningful in this case).

I found this "Hell Yeah Rolling Opamps Makes Your Amps Sound Better! (Hear It For Yourself!)"

from GR-Research.

as the quintessence of what hundreds of modders communities across different geographical regions claim.
(unfortunately, 24 minutes of words, no fairly real proofs but in this case there was something more).
To save the time, some useful hints: skip the first 7 minutes of blabla, then listen to declarations about discrete op-amps in this case, and from 16:30 begins something like a “measurement approach/methodology”.

Some claims from this video:

"there's a lot of those flat Earth guys out there that think because they measure the same that they've got to sound the same.
They could not be more wrong."

"just because you can go in and measure the distortion levels or the noise floors and think, "Well, if it measures the same, it's going to sound the same."

"They're not looking at everything else that is involved in how those things are made, because what they're doing is they're basically a voltage stepping device.
They're stepping up voltage. So it's adding gain. So there's opportunity there for a lot of things to happen and a lot of ways to change the way it sounds,
by the way it's adding that gain and the quality of everything in the path."
...
and finally: "there is a huge variance in sound between all those different op-amps"

In this video were considered: Burson Vivid 7, Sparkos (dual discrete op-amp SS3602), etc...

Small humble analysis:

"Higher resolution" – I hope at least this could be measured in the range 20–20 kHz?

"Smoother, lusher, more richness" = I have no clue if it is even possible to calculate somehow.

"More harmonic structure" – this one I guess is possible to capture and reflect?

"More involvement" – also no clue what it is.

Accordingly to their statements: Final declaration about concrete discrete op-amp (but it is still an op-amp!):


one amp: "a little tighter control in the bottom end, a little more definition there and control down low"

another amp:
"a little more relaxed" – no clue what that is

"the vocals were richer" – this I suppose is in audible range and could be reflected / taken into account

"the sound stage was bigger" – AFAIK only indirectly via a bundle of parameters, could also be somehow compared.


Regarding experts – he mentioned himself, Hobbs, Ron – frankly I do not know anyone, including their labs, but they make it public to discuss.


What they did in the video from 16:30 onward is that they try to estimate perception from the listener’s point of view in a test audio room.

Please correct the partial quote of my post that completely changes the meaning of what I wrote. If you are going to quote that sentence, include it all:

You determine by listening to the very talented engineers who have taken time out of their day to advise you

 

[DSJR]

You really would be amazed how half a DB difference in gain, would subliminally change one's perception of 'detail' in a piece of music. Alter the level half a dB on one channel only and it's even more confusing, as our crude hearing and brain processing may tell there's a difference, but cannot work out what it is. been there, done it and basically gave up after. Confirmed hobbyist audiophiles would probably refuse to believe it as all they trust are their ears, which are actually awful in the animal world... Our brains are supposedly far more biased to sight I gather...

 

[PlanarSearcher]

You are new to ASR, so you may have missed all the science based critic on Danny's videos. All his listening is without proper controls and hence baseless. 'Nough said ...

Surely so (i am a beginner), but could i please reask: is any way (or may be it was also already done?) to transform his (actually even their - it is a Lab) words into some measurement:

1. "the sound stage was bigger" - did someone measured this while Op amps comparison?
2. "the vocals were richer" - it is in audio range, so could it be somehow traced? Harmonic-to-noise ratio (HNR), spectral shape (brightness vs. fullness), RMS variation over time, etc... ?

Sorry if I missed smth and it was already measured and there was no any difference. But if it was, could you please point one example or even more - is it possible to take his sample and to measure again?

 

[antcollinet]

Surely so (i am a beginner), but could i please reask: is any way (or may be it was also already done?) to transform his (actually even their - it is a Lab) words into some measurement:

1. "the sound stage was bigger" - did someone measured this while Op amps comparison?
2. "the vocals were richer" - it is in audio range, so could it be somehow traced? Harmonic-to-noise ratio (HNR), spectral shape (brightness vs. fullness), RMS variation over time — and vocals, etc... ?

Sorry if I missed smth and it was already measured and there was no any difference. But if it was, could you please point one example or even more - is it possible to take his sample and to measure again?

If those were real differences in the sound waves, it is possible to measure difference. The difference would be in the form of distortion, or frequency phase response (or noise but that won't cause those described differences) There is nothing else. When op amps changes are measured there is never any audible difference - unless it has "broken" the circuit. (there will be measurable differences but those differences are too small for human hearing to detect)

But here is a very important point. Soundstage is in the recording and is brought out by the speakers, and the way they interact with the room (reflections etc). Amps/Dacs/other electronics don't influence that in any way unless their distortion is so bad it destroys the cues in the signal. I don't think you will find anything bad enough to do that. So when you see someone claim "better soundstage" when discussing electronics, you know they are talking out of their ass.

But you are going round in circles. If you come here, you are going to get the science and engineering based answer, and you've had plenty already. If you don't want that - if you want the word of subjectivist guru's or (frankly) snake oil salesmen - then you are in the wrong place.

From a statistical point of view it cannot simply be a random coincidence. In my humble opinion

Perhaps not - but it can be lack of level matching, or it can be perceptive or confirmation bias. If everything you have read or watched tells you will hear something, then if you compare sighted so you know what you are listening to - you often will hear what you've been told you will hear. It is an unconcious process to do with how our senses work. We can't turn it off even if we know it is happening.
This is why if someone tries to convince you something is audible - but they haven't done a properly controlled blinded comparison, you can safely ignore them.

 

[LTig]

Surely so (i am a beginner), but could i please reask: is any way (or may be it was also already done?) to transform his (actually even their - it is a Lab) words into some measurement:

1. "the sound stage was bigger" - did someone measured this while Op amps comparison?
2. "the vocals were richer" - it is in audio range, so could it be somehow traced? Harmonic-to-noise ratio (HNR), spectral shape (brightness vs. fullness), RMS variation over time, etc... ?

I have to repeat the common sense here again. As long as listening impressions are gathered without proper controls (I repeat again: blind or double blind, level matched to less than 0.1 dB) people will hear differences all over the place, influenced by unconscious bias and level differences. The latter can be measured but they didn't. Been there, done that.

Sorry if I missed smth and it was already measured and there was no any difference. But if it was, could you please point one example or even more - is it possible to take his sample and to measure again?

You can almost always measure a tiny difference in output when rolling opamps. The difference is in almost all cases inaudible though, as 80+ years of psychoacoustic research have shown. Exception are those cases where opamp rolling broke the intended design of the circuit (e.g. oscillating amplifier stage).

 

[PlanarSearcher]

You can almost always measure a tiny difference in output when rolling opamps. The difference is in almost all cases inaudible though, as 80+ years of psychoacoustic research have shown.

A thought I mentioned in my first post: could it be that all of you have such high-quality hardware that this factor simply does not matter in your case? A different op-amp can tolerate (or expose) PCB/layout imperfections differently, which might sound like a “routing improvement.”

Yes, this is definitely outside the main scope, but it could correlate with low-cost devices, where such contradictory statements usually appear. Just a thought…

 

[MaxwellsEq]

Sorry, I just missed the core of the message — did you mean that their “test clips” are not relevant?

I personally missed what exactly they did with microphone measurements in the test audio room and the reported values. However, above I have listed my thoughts on what could potentially be measured additionally based on his statements. So I am mainly interested in this: what if we take those ideas and enhance the measurement methodology, while also trying to take the listener into account?

I find it strange that so many people, independently of location all over the world, report the same impressions — even in different languages, and often without much overlap.

From a statistical point of view it cannot simply be a random coincidence. In my humble opinion, I am not claiming anything, but rather asking whether it would be possible to improve the measurement methodology and whether such effects can actually be measured at all — this is the main question.

In the video above, it is also interesting that, besides loudspeakers, high-resolution headphones were mentioned, on which, according to their conclusion, the difference would be audibly distinguishable.

Have you encountered Cochrane Reviews?

https://www.cochranelibrary.com/

They independently look into all available academic research into a specific medical subject such as deep brain stimulation. They then look at methodology and controls. They reject any research (no matter by whom, size etc.) where controls are inadequate; where some data was "ignored" because it didn't fit the conjecture; where summaries don't match the data. This often leaves a smaller set of rigorous, repeatable research. Quite often the results contradict "popular" views and even some big, expensive research.

Now imagine you applied this rigour to all the things you are talking about.
Did they NOT do before and after measurements (including checking for very high frequency oscillations) - bin that opinion.
Did they NOT do double-blinded listening tests with at least 20 iterations and applied probability - bin that opinion.
Did they NOT ensure all testing is done with all signals level-matched to 0.01V - bin that opinion.

If you do this you will find little or no quality evidence supporting audible benefits of op-amp swapping.

 

[antcollinet]

A different op-amp can tolerate (or expose) PCB/layout imperfections differently, which might sound like a “routing improvement.”

The opposite would normally be the case. A poor quality design will have high noise and distortion that will form the bottleneck for performance. Any change from opamp swaps is so infinitiesimal compared to this that it is essentially meaningless. This is in fact the case even in quality electronics. The overall performance of the device is worse than that of the op amp - so tiny changes from op amps are pointless.

This is all assuming that by swapping op amps you don't make the circuit unstable and cause a dramatic reduction in quality.