Rolling (replacing) Op-amps in Topping D10 DAC

[pavlo_y0l0s]

This is a study of whether changing op-amps in a DAC makes a measurable difference. Taking a lead form "rolling" tubes in tube products, a trend has started in replacing op-amps -- sometimes with much difficulty -- to improve performance of solid state products like DACs. A few months ago I tested the Topping D10 DAC and since it has a socketed op-amp, members asked if I could experiment with changing it to other pin-compatible parts. A couple of kind members supplied me with a good inventory of op-amps to test all the way up to esoteric discrete ones.

Here is the overall circuit diagram:

View attachment 15696​

You can see how the ESS DAC drives the OPA2134 op-amps that are soldered, providing current (I) to voltage (V) conversion. Their outputs in turn drive a pair of op-amps in the socketed Op-amp (circled) which are our output buffer/op-amps. That is the part we will be replacing.

EDIT: an earlier version of this review mentioned that we were changing the I/V op-amp due to misunderstanding of a statement from Topping. It is now corrected above.

The op-amps I had on hand were:

• OPA-2134 (shipped in the Topping D10)
• Muse 8820
• Muse 8920
• LME49720
• OPA-1612a (SMD part mounted to through-hole socket)
• 994Enh-Ticha Dual Matched Discrete Operational Amplifier. This one is quite fancy: http://www.sonicimagerylabs.com/products/Model994DiscreteOpAmp-Ticha.html

View attachment 15684

It retails for a whopping $94 each! The OPA2134 in contrast costs $4.79 in single units.

Measurements
For this testing, I thought any important difference would show up in THD+N (distortion + noise) versus frequency as this is what the data sheet for op-amps emphasizes a lot. We will see the effects of both noise and distortion in one measurement.

Here are the results:

View attachment 15683

This chart is hugely amplified (vertical axis) relative to what I normally show. Despite that, the differences are tiny. Assuming you want to go by that difference anyway, the best performance was achieved using the op-amp shipped with the unit, the TI OPA2134!!! In other words, the designer knows best.

I also looked at SINAD differences (not shown) and there was no difference to speak of there. The unit itself varied in performance by 1 dB there and that was what I observed as I changed op-amps.

Conclusions
With respect to Topping D10 DAC at least, rolling opamps is of no value at all. Fortunately nothing bad happens but nothing good either. I suggest leaving it alone as is.

I will try the same exercise with a different DAC in the future to see how consistent the results are.

-------------

As always, questions, comments, recommendations, etc. are welcome.

If you like this review, please consider donating using Patreon (https://www.patreon.com/audiosciencereview), or upgrading your membership here though PayPal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).

What you got here is the distortion profile of the OPA2134 that has been used as I/V converter. Anything you put after it will have those distortions just passed down the chain. What might really change the sound - is swapping those SMD soldered opamps. They are not socketed and require soldering which sucks.

 

[pavlo_y0l0s]

Changed the SOP8 OPA2134 OPAmps of the I/V-Stage to OPA1642 just for fun (the soldered ones, not the socketed!).
Quick listening Test, it sounds as good as before

Some more listening needed.

EDIT: LPF/Outputstage is an LME49720

You might hear a difference if you try lm4562 as I/V. Or OPA1656. Or OPA1612.

 

[SIY]

You might hear a difference if you will use lm4562 as I/V

They do tend to oscillate in the hands of novices. It's one of the few swaps that might make an actual difference, but a bad one.

 

[pavlo_y0l0s]

They do tend to oscillate in the hands of novices. It's one of the few swaps that might make an actual difference, but a bad one.

If I understand correctly, oscillations for lm4562 (or any IC with high slew rate) are usually because of the bad layout of the PCB, or omissions of the stability components like decoupling capacitors from ground to positive and negative supply.

My point was that LM4562 is very low on distortion in 20hz to 20khz full range. Vs what was installed in the factory (opa2134 IC) has more distortion starting from 1~2khz and up. Same with OPA1642 - it's distortion graph is very similar to opa2134.

 

[SIY]

If I understand correctly, oscillations for lm4562 (or any IC with high slew rate) are usually because of the bad layout of the PCB, or omissions of the stability components like decoupling capacitors from ground to positive and negative supply.

This particular op-amp takes more heroic measures than usual. Small (physically) resistors or inductors right at the input terminals are needed if you want to extract performance out of them. One guy I know actually hand-wound small air core inductors with fine wire over a pencil lead. When I used them, I had SMD inductors right at the pins.

Knowing the quirks of particular parts is what separates competent designers from hacks. And one more reason I tell people to not screw with this stuff unless there's a specific problem to be solved AND they are confident that they know more than the engineer who did the design in the first place.