Does Op-amp Rolling Work?

[Deleted member 88295]

I am extremely suspicious to opamp parts offered by Audiophonics. The prices do not make sense and the photos (OPA627) are suspicious again.

They are refurbished parts as stated in description. Can you say what looks off about them?

 

[Deleted member 88295]

Yesterday I rolled some op amps (ic) in the Fosi ZD3. This is all subjective but I am posting it anyway. The lme49720 (stock) was the best so I kept it it. What surprised me was that opa1612 and opa1656 sounded identical in this dac. Biggest difference was with the muses8920, it sounded very congested.

There is one thing I want to try next and that is a class A output buffer with the lme49720, is that considered op amp rolling too?

 

[Buckeye Amps]

Biggest difference was with the muses8920, it sounded very congested.

Sudafed works wonders for congestion.

 

[pma]

They are refurbished parts as stated in description.

Maybe that is enough for an amateur.

Otherwise, unknown history, unknown potential stress they were exposed to, unpredictable reliability. At best. And at some of the photos do not look like genuine parts. My final comment to this.

 

[Digital_Thor]

Yesterday I rolled some op amps (ic) in the Fosi ZD3. This is all subjective but I am posting it anyway. The lme49720 (stock) was the best so I kept it it. What surprised me was that opa1612 and opa1656 sounded identical in this dac. Biggest difference was with the muses8920, it sounded very congested.

There is one thing I want to try next and that is a class A output buffer with the lme49720, is that considered op amp rolling too?

Again, ask yourself why. What are you missing?
Designing electronics is the other way around. You dont pick components that you think look cool, then try to make everything around it fit.
You set up parametres and a plan with a given input and a given output, and then work from there. Whatever components inside your box might be, doesn't matter, as long as the box does what it needs to do.
When you change an opamp, you disregard everything else in that circuitry, and you become the designer, which also means that you take responsibility for everything that potentially changes with that opamp-swap.
So, was original product faulty? Was working less efficiencient? Did you have any trouble with all external components connected to the product, leaving you with the suspicion that another opamp could make anything better?
Is a training/learning excersice?
Just for fun?
Sorry to boil it down, but you most likely confuse yourself, if you don't understand the basic operations of the circuit you are working on, leaving you with nothing else, but to believe whatever someone is selling you

 

[JSmith]

This is all subjective

Then there is no interest in that here, no need to post it.


JSmith

 

[theREALdotnet]

Op-amp rolling can be beneficial, if there is a specific problem to be addressed. Years back I used to own a Rotel RB991 amp and needed to use its balanced inputs. Unfortunately, the opamps used for making the balanced signal single-ended (only to feed it into the RCA input and turn it balanced again using a long-tailed pair ) were AD711. They produced an audible hiss. I replaced them with OPA627 and the hiss was gone.

 

[Mnyb]

Op-amp rolling can be beneficial, if there is a specific problem to be addressed. Years back I used to own a Rotel RB991 amp and needed to use its balanced inputs. Unfortunately, the opamps used for making the balanced signal single-ended (only to feed it into the RCA input and turn it balanced again using a long-tailed pair ) were AD711. They produced an audible hiss. I replaced them with OPA627 and the hiss was gone.

I also had a 991 decent amp but it had some problems apparently, mine could demodulate AM radio ?

 

[theREALdotnet]

I also had a 991 decent amp but it had some problems apparently, mine could demodulate AM radio ?

I reckon any diode junction can do that, if you can catch a strong enough signal

 

[pma]

All well known for decades - BJT vs. JFET opamp susceptibility to RFI.

https://www.analog.com/media/en/training-seminars/tutorials/mt-096.pdf

 

[Deleted member 88295]

Here are the two measurements side-by-side. The original amp review measurements had a slightly higher H3 (~ -92 dB @ 3 kHz) that accounted for the difference. The measurements after the 1 channel op-amp swap resulted in a ~ 2 dB reduction in the H3 spikes, but added some AC main spikes that didn't affect the aggregated SINAD number.

This underscore the highly sensitive nature of measurements at these low SINAD levels. Any seemingly minor changes may result in some "measurable" differences. They may be the run-to-run variations, differences in the time and number of reading averaging, moving some wires around, component temperature differences, etc.

View attachment 436925

I did not read every post in this thread so this might have been mentioned. I find this interesting, what I think happens is because of the discrete op amp drawing a more consistent load the Sinad goes up. It would probably be even better with 2 discretes.

 

[MAB]

I did not read every post in this thread so this might have been mentioned. I find this interesting, what I think happens is because of the discrete op amp drawing a more consistent load the Sinad goes up. It would probably be even better with 2 discretes.

Did you at least read the first post?

 

[antcollinet]

I did not read every post in this thread so this might have been mentioned. I find this interesting, what I think happens is because of the discrete op amp drawing a more consistent load the Sinad goes up. It would probably be even better with 2 discretes.

There are two questions that stem from this.

1 - Why do you think discrete op amps draw a more consistent load?

2 - Why do you think (if they do) that will result in a better SINAD, when the OP shows that the discrete op amp (at best) is not better than the integrated IC?

 

[Deleted member 88295]

There are two questions that stem from this.

1 - Why do you think discrete op amps draw a more consistent load?

2 - Why do you think (if they do) that will result in a better SINAD, when the OP shows that the discrete op amp (at best) is not better than the integrated IC?

Because the transistors are always on.

And I do see a difference:

 

[antcollinet]

And I do see a difference:

90.3 TO 90.4 is not a relevant difference - it is less than run to run variation, and is equally likely to be due to component tolerance differences in the two channels. If you read Amir's commentary he points out the lack of difference.

There is certainly nothing that can result in any audible difference.

 

[Deleted member 88295]

90.3 TO 90.4 is not a relevant difference - it is less than run to run variation, and is equally likely to be due to component tolerance differences in the two channels. If you read Amir's commentary he points out the lack of difference.

There is certainly nothing that can result in any audible difference.

Afaik it is the same amp tested. On the left is first test (baseline) second test shows that channel two goes from 86 90 Sinad with the discrete and channel 1 improves too.

 

[antcollinet]

Afaik it is the same amp tested. On the left is first test (baseline) second test shows that channel two goes from 86 90 Sinad with the discrete and channel 1 improves too.

You don't know set up differences between the two tests - so comparing them is not valid. It is not even stated if it was the same physical amp that was tested at the two times.

The only valid comparison is in this specific test - set up to compare stock op amp against discrete by changing one channel - in that case they are (obviously) compared under the exact same conditions.

Which is why in this OP all the discussion is about comparing the two channels - not referring back to an earlier test.

 

[MAB]

Afaik it is the same amp tested. On the left is first test (baseline) second test shows that channel two goes from 86 90 Sinad with the discrete and channel 1 improves too.

The original OpAmp for this Douk A5 amp is NE5532. Your comparison is NE5532 vs. NE5532.

You are comparing two different experimental runs to each other, using the same OpAmp model.

First, do you think you can hear the difference between 88.5 dB and 90.3 dB? Most people can hear about 30 - 40 dB if listening to a pure 1 kHz tone, much less if listening to music.

Also, this Douk amp has sockets to enable OpAmp rolling, the sockets introduce noise and distortion compared to proper soldered components. Just breaking the contact on the OpAmps in the socket is worth a couple dB of noise and distortion. 1-2 dB is normally run to run differences, interesting that after OpAmp swap the unit was at 90.3-90.4 dB SINAD. Possible that repocketing broke the oxide barrier in the pin socket. Which brings me to another point...

It so interesting to see people buy a device with OpAmp rolling sockets, which introduce more noise and distortion than the differences between OpAmps themselves. The people then crow crow about the differences in distortion when the crappy socket is likely a bigger factor.

If you were really able to hear or worried about this level of distortion, you would ditch the sockets and buy a device with good design, PCB layout, and no sockets since the Nose and Distortion they introduce is larger than the typical differences people see between OpAmps.

 

[Deleted member 88295]

You don't know set up differences between the two tests - so comparing them is not valid. It is not even stated if it was the same physical amp that was tested at the two times.

The only valid comparison is in this specific test - set up to compare stock op amp against discrete by changing one channel - in that case they are (obviously) compared under the exact same conditions.

Which is why in this OP all the discussion is about comparing the two channels - not referring back to an earlier test.

I disagree, without a proper baseline test i.e. stock, we can't rule out the discrete changed the Sinad. You can see by the 60Hz noise that the discrete changes something.

 

[MAB]

I disagree, without a proper baseline test i.e. stock, we can't rule out the discrete changed the Sinad. You can see by the 60Hz noise that the discrete changes something.

You are 100% incorrect.
The 60 Hz noise changed in both channels.
Test setup changed from the original to the OpAmp swap run, and you can't make comparisons.

Like I said, I expect run to run differences, especially with gear using OpAmp sockets (sockets are flakey), leading to measurements like the -105 dB noise signal that you try to identify as a gotcha.

You seriously need to stop navel-gazing, learn what human thresholds are, and learn to make some measurements to see how external factors affect noise and distortion. At 100dB, turning on/off LED lighting in my shop affects measured performance. Glad I can't hear it. Good news, I also can't hear the effect of gravitational waves affecting space time in my shop.