Too hot to comfortably put your finger on means very little in the context of a DAC circuit that is known to be power hungry, and thus, dissipate quite a lot of heat. The opamps themselves might not have crazy high dissipation but if all the surrounding circuitry does then the overall board temperatures gets hot quickly. Small differences in package dissipation, and internal construction, can quickly result in reasonably big temperature differences, as felt, from the top of the package. Remember you only need like a 5 degree difference between one package and another for one to feel comfortable and the other to feel ouch.
Some parts are very good at dissipating their heat out of their pins/legs and into the PCB, others are much worse, and then some have dedicated thermal connections.
The OPA1622 is a slow device compared to the OPA1612/OPA1656. It might require more capacitance in the LPF of the I/V stage to keep it happy Vs the others. Also the OPA1612 does have the highest quiescent over the other opamps we're talking about it will be the hottest. Without scoping the outputs or measuring the DAC there's no telling if things are happy or stable.
Some parts are very good at dissipating their heat out of their pins/legs and into the PCB, others are much worse, and then some have dedicated thermal connections.
The OPA1622 is a slow device compared to the OPA1612/OPA1656. It might require more capacitance in the LPF of the I/V stage to keep it happy Vs the others. Also the OPA1612 does have the highest quiescent over the other opamps we're talking about it will be the hottest. Without scoping the outputs or measuring the DAC there's no telling if things are happy or stable.
Who is 'ALL?' Aliexpress? If so, the opamps they recommended are fakes. They cost around $12/each here. They will happily sell you fakes and make it look like a discount.
Besides, AD797 is a single, not a dual opamp. Putting two of them on a dual adapter will get hot and may oscillate. Also, most opamps can't handle the RF coming out of dac outputs. Therefore distortion and or noise tend to be higher than for OPA1612. OPA1656 tends to have thin bass in comparison to OPA1612. IOW, there is no opamp swap fix for a bad layout. You can try swapping opamps to find the least bad sound, that's about it. Sorry.
AD797 is not first choice for I/V as it needs 100R+ in series to the input with unity noise gain. That's why in ESS' own old reference schematic there is 100R in series with the integrator capacitor, allowing some RF still come through.
ESS DACs should be used with glitch-killer capacitor to keep requirements for OpAmps resonable, see ES9039pro datasheet. In this datasheet there also finally is a proper diff-amp topology that has a) zero DC offset and b) can be used single-ended with almost no penalty. It uses OPA1612.
ESS DACs should be used with glitch-killer capacitor to keep requirements for OpAmps resonable, see ES9039pro datasheet. In this datasheet there also finally is a proper diff-amp topology that has a) zero DC offset and b) can be used single-ended with almost no penalty. It uses OPA1612.
I trying opa1612, opa1656 and opa2156. The last one is the best from all three. In fact opa2156 is a opa1656 with laser trimming, better precision and the input stage changes so as to increase its slew rate depending on the signal and this is felt by the ear. By the way, what is the continuous voltage between the inputs of the I/V converter?
Both opa1656 and opa2156 are CMOS. Don't let yourself be fooled, both are excellent for audio. Input differential impedance at opa2156 is 100Mohm, like opa1656 
THD+N=-132dB at opa2156 and -131dB at opa1656 Do you think it is not good for audio ? Both op amps are "brothers" with opa2156 a little more precisely and faster than opa1656. I use them in SMSL D400EX.
THD+N=-132dB at opa2156 and -131dB at opa1656 Do you think it is not good for audio ? Both op amps are "brothers" with opa2156 a little more precisely and faster than opa1656. I use them in SMSL D400EX.
Yep, OPA2156 looks quite similar but certainly not identical to OPA1656, key difference is dual (NMOS and PMOS) input stage, rail-to-rail input whereas the OPA1656 is NMOS only. The rest of the design might be exactly identical (2nd stage and output stage).
It still is an excellent OpAmp for Audio..
It still is an excellent OpAmp for Audio..
If I'm not mistaken, opa1642 is a real JFET. It has rail to rail differential inputs, it doesn't have antiparallel diodes. Svadim79, if you have some DC voltage between the inputs of the I/V converter, maybe opa1642 is better. The GBW is a little small but who know, maybe it sound very good, why not.
Correction: PMOS input stage, not NMOS, as far as one could infer from the specs.
There has been discussion of OPA2156 and OPA1656 here before:
https://www.diyaudio.com/community/...ormance-cmos-audio-op-amp.335416/post-6071979
https://e2e.ti.com/support/amplifie...-about-the-input-offset-voltage-specification
https://www.diyaudio.com/community/...ormance-cmos-audio-op-amp.335416/post-6071979
https://e2e.ti.com/support/amplifie...-about-the-input-offset-voltage-specification