Opamp Rolling, Does It Work?

[JohnYang1997]

Parallel opamps not only reduces distortion but also reduces voltage noise. But current noise doubles. For fet input opamps it's like a gospel. But only if you can make it stable with more than two of them. I can't get 3 opa827 working. But i may have got it wrong.

 

[JohnYang1997]

on the opa551 side. The noise is pretty high. Even though you can argue the noise is still not going to be audible but compared to paralleled 4556 it's much higher. Also we want lower distortion under load, better capacitance load stability and lower noise but not essentially higher power especially with higher noise. We hardly use 10mw for most headphones. At normal volume noise will be much higher than all the distortion harmonics.

 

[AnalogSteph]

Parallel opamps not only reduces distortion but also reduces voltage noise. But current noise doubles. For fet input opamps it's like a gospel. But only if you can make it stable with more than two of them. I can't get 3 opa827 working. But i may have got it wrong.

The problem is, as input current noise increases with multiple ICs in parallel, so does input capacitance. In a noninverting amplifier circuit, this is essentially in parallel to Rg and gives you a high-frequency resonance peak. You either have to include some capacitance in parallel to Rf (usually no more than a few pF) or reduce feedback network impedance in general. Adding some input series impedance may also be beneficial.

Stacking modern-day parts also may not be as easy as the same for TL07x with their high internal output impedance (hundreds of ohms if memory serves). External current sharing resistors may be needed.

 

[jackenhack]

Parallel opamps not only reduces distortion

Parallel opamps do nothing for distortion.

 

[JohnYang1997]

.

Well, real world measurements are the key. If it measures good there is nothing wrong.

 

[JohnYang1997]

Parallel opamps do nothing for distortion.

As in o2, output uses opamos. Paralleling allows sharing the current draw for two opamps, less current draw means less distortion.

 

[JohnYang1997]

You either have to include some capacitance in parallel to Rf (usually no more than a few pF) or reduce feedback network impedance in general. Adding some input series impedance may also be beneficial.

ok this part helps. So does directly connect output to inverting input cause instability? According to you, it's fine right? I just need to add resistor in front of non inverting input?
Or I'll include 47ohm to 220ohm resistor at feedback loop and in front of the noninv input.

 

[LTig]

So what is the next step? I need to find a better replacement for the output opamps. They need to be able to handle the same amount of output current as the NJM4556 but have better specs. If you have any suggestions, I'm all ears. It could be fun to see if there are further gains (pun intended) to be had with another opamp.

I think the THD results from common mode distortion. To improve it opamp rolling does not help a lot, you've already tried really good ones. Improving THD means getting rid of common mode swing, and there are several possibilities:

• let the output opamps amplify. Problem is that for parallel work the amplification factors must be extremely close (selected resistors).
• use the output opamps in inverse mode. Same problem as above, plus increased noise due to the resistors at the inputs.
• rail bootstrapping: move the supply voltages of the output opamps with the signal. See page 110 ff in Douglas Self's book "Small Signal Audio Design". Needs additional parts and a somewhat higher supply voltage.

 

[JohnYang1997]

I think the THD results from common mode distortion. To improve it opamp rolling does not help a lot, you've already tried really good ones. Improving THD means getting rid of common mode swing, and there are several possibilities:

• let the output opamps amplify. Problem is that for parallel work the amplification factors must be extremely close (selected resistors).
• use the output opamps in inverse mode. Same problem as above, plus increased noise due to the resistors at the inputs.
• rail bootstrapping: move the supply voltages of the output opamps with the signal. See page 110 ff in Douglas Self's book "Small Signal Audio Design". Needs additional parts and a somewhat higher supply voltage.

Why not just use jfet/cmosfet input opamps instead, and probably parallel more for less noise?

 

[bigx5murf]

I swapped out the op-amp on my Little Dot I headphone amp and thought it sounded better. Who knows, I was probably so surprised that I didn't break it that it had to be an improvement. IIRC, I settled on some kind of adapter that transformed a "stereo" op-amp into two mono parts. It was fun and maybe cost $10.

I did as well, and also ended up with a little collection of tubes as well. I enjoyed this unit when it was my only H amp. But after getting a JDS O2, all the flaws of the Little dot became noticeable. Especially it's tendency to pickup noise from every wireless device, and it's especially bad in high gain mode (which is a huge pain to switch from low/high gain). Mine's been gathering dust, and not doing much more than look pretty.

 

[noobie1]

I tried opamp rolling in my Creative X7. Only tried a few different opamps. It’s generally a pain unless u have the proper tools. Most resulted in non-discernible changes. There was one that did make a noticeable subjectively positive difference. There were some that degraded the sound significantly but I suspect it’s because they were fake (I ordered some from Chinese eBay sellers). Once I started ordering from Mouser, it was fine.

 

[LTig]

Why not just use jfet/cmosfet input opamps instead, and probably parallel more for less noise?

Because it does not improve common mode distortion?

 

[JohnYang1997]

Because it does not improve common mode distortion?

I thought it would? In the diyaudio forum, the opa1622 thread, people there was talking about wraping opa827 around opa1622 to minimize common mode distortion. Maybe I was mistaken?

 

[JohnYang1997]

some reference from John Caldwell
https://www.diyaudio.com/forums/vendor-s-bazaar/283672-audio-op-amp-opa1622-3.html

 

[LTig]

I tried opamp rolling in my Creative X7. Only tried a few different opamps. It’s generally a pain unless u have the proper tools. Most resulted in non-discernible changes. There was one that did make a noticeable subjectively positive difference. There were some that degraded the sound significantly but I suspect it’s because they were fake (I ordered some from Chinese eBay sellers). Once I started ordering from Mouser, it was fine.

In 1998 I did some opamp rolling in an Arcam Black Box 3 (Bitstream DAC from 1991). I replaced LM327 at the DAC chip output by OPA134PA and LM627 at the analog output stage also by OPA134PA. Wow, what an improvement! More dynamic, cleaner bass and highs, more room, more details - you name it.

The shock came on the next day when I performed a blind test against an unmodified Black Box 3 (had 2 at that time). I scored 8 correct from 10 tries but it was very difficult to here a difference at all. Since then I know how easily I can fool myself.

 

[LTig]

I thought it would? In the diyaudio forum, the opa1622 thread, people there was talking about wraping opa827 around opa1622 to minimize common mode distortion. Maybe I was mistaken?

Maybe not, I haven't read this thread. Do you have a link?

 

[noobie1]

In 1998 I did some opamp rolling in an Arcam Black Box 3 (Bitstream DAC from 1991). I replaced LM327 at the DAC chip output by OPA134PA and LM627 at the analog output stage also by OPA134PA. Wow, what an improvement! More dynamic, cleaner bass and highs, more room, more details - you name it.

The shock came on the next day when I performed a blind test against an unmodified Black Box 3 (had 2 at that time). I scored 8 correct from 10 tries but it was very difficult to here a difference at all. Since then I know how easily I can fool myself.

I don’t trust my ears. I make changes without telling my wife what I did. Only when she voluntarily makes a comment do I chalk it up to a sonic difference.

 

[JohnYang1997]

Maybe not, I haven't read this thread. Do you have a link?

Apart from the link above. This is the exact post I was referring to.
https://www.diyaudio.com/forums/vendor-s-bazaar/283672-audio-op-amp-opa1622-12.html

 

[LTig]

Apart from the link above. This is the exact post I was referring to.
https://www.diyaudio.com/forums/vendor-s-bazaar/283672-audio-op-amp-opa1622-12.html

Thanks for the link, had a quick look into it. If the stability problems can be solved and the OPA827 does not suffer from CM distortion, this solution could work. It would also improve crossover distortion. It depends on how much loop gain is available for correcting the output opamps.

Just looked into the data sheets: CMRR is 110-126 dB for the OPA827, but 110-137 for the LME49990, so the latter could be a better choice.

 

[DonH56]

Note CMOS input op-amps will typically be very noisy in the audio band. Chopper-stabilized op-amps can reduce 1/f noise but have other issues. JFETs are generally preferred as they have much lower low-frequency noise. If you have a low driving impedance, bipolar (BJT) inputs often have higher performance (with high driving impedance the current input noise of a BJT op-amp can be high).

FWIWFM - Don