Burr Brown OPA 2143 vs NJM2068

Generally, not a good sub. OPA2134 is FET-input (higher noise voltage, lower noise current) and NJM2068 is bipolar (higher noise current, lower noise voltage). This means that they're optimized for different circuit impedances.

What's wrong with using the original part?

I was referring to the original poster's question - the OPA134 family is excellent, and I can see no reason to sub . . . other than the convienence of what parts one has lying around.

While I don't think the reason of "it's newer!" is in and of itself a valid reason for swapping opamps around, I am also very much NOT a fan of the TL0xx devices - they have an internal resistor in series with the output that can in many applications severely affect phase margin. But it's still a good idea to consider why they were chosen for the original design - and the same reasoning applies to the TL0xx opamps - they're FET-input, and the LM4562 is . . . not. So the current noise for an LM4562 will be MUCH higher than the TL0. A much better choice would be something like an AD845.

Also, with single opamps you generally have to be more careful with subbing - because of those extra pins. In the case of the TL071, it's still internally compensated, and the extra pins are just offset null or something - almost always unused in an audio application. But the PCB designer may still have routed a trace across an NC pin that's not NC on your sub, so you still need to check before you swap.

And something like the 5534 (unlike its dual-opamp brother, the 5532) has external compensation, so you DON'T want to sub around willy-nilly in this case - that and the 5534, despite its age, is still an outstanding opamp.

Change to LM4562 and you might be able to remove this cap.

With emphasis on "might", like in bold print, or a big neon sign. The LM4562 is indeed a good sub candidate for the 5532, even though it's not for the TL072. And the 5532/5534 is definately not known for its DC performance -- I agree that most newer designs are much better.

But the input bias currents of these medium-impedance bipolar opamps are far from negligible, and it is extremely rare to find balanced DC impedances in an audio opamp circuit that features an output capacitor - and this is what to look for (in addition to low DC gain) if you're planning on removing this capacitor in a circuit without a DC servo.

I take it you're fond of this part for its sound quality . . . maybe I'll have to add it to my list of things to play around when I have some time.

Typical bias current for LM4562 is 10nA. Taking into account possible unbalance of 10k we get 0.1mV unbalance.

Good, keep going. Calculate the effect of the input offset current based on the higher of the two source impedances, and add that. Then add the input offset voltage. Then multiply by the the sum by the DC gain of the circuit . . . you now have the number you're after. A mV or two of offset gets pretty scratchy-clackety on the source-selector and volume controls of a following preamp (provided they're DC coupled).

Again, it COULD work fine if you change to DC output coupling to accompany your opamp sub (10nA is pretty low for a bipolar opamp) . . . but like any modification, a full analysis of the particular circuit as it pertains to your particular mod is required if you want to be confident of a positive outcome.