If there is a bottleneck, what if it the AD part of the system used for loopback testing? Maybe the dac performance should be measured with a different ADC box? If that did not show a difference in dac performance then the answer would probably have to be in something that appears as noise, or at least looks like noise in the FFT analysis.
The OPA16XX are very good opamps but you have to look about for 1/f noise. There’s a reason why many opamp vendors quote noise at 1 kHz 
You won’t beat the AD797 for noise with low source Z’s, but as Mark says, the NE5534 workhorse is still the best bang for your buck in the audio industry.
I’ve personally not had any problems with the LM4562, but if you root around on the web, there are a few folks ranting about high 1/f on some parts. I guess if 1/f is an issue in your application (hydrophones, seismic preamps and other scientific instruments) you will have to go for the AD797. I use this in my commercial preamp and can vouch for its outstanding performance, though I am using the OPA164X on some of my diy stuff - great opamps IMV.
You won’t beat the AD797 for noise with low source Z’s, but as Mark says, the NE5534 workhorse is still the best bang for your buck in the audio industry.
I’ve personally not had any problems with the LM4562, but if you root around on the web, there are a few folks ranting about high 1/f on some parts. I guess if 1/f is an issue in your application (hydrophones, seismic preamps and other scientific instruments) you will have to go for the AD797. I use this in my commercial preamp and can vouch for its outstanding performance, though I am using the OPA164X on some of my diy stuff - great opamps IMV.
I mean difference between channel is 0.0 dB at 1K. Frequency response measured is +0.04 -0.02dB, from 20Hz to 40K.
I actually tried 2 different ADC, Apogee Symphony and Apogee ensemble. Symphony is a better product with much better spec but the only meaningful difference between them is stereo crosstalk, 84dB vs 96dB @1K. BTW, Symphony only ADDA loop is measured much better in every aspect, so its AD would not be a bottleneck in this case.
Last edited:
How about more detailed FFT measurements, not just THD+N (which isn't a very good metric for SQ)? Its better to look the level of each HD spur, and also at the noise floor level (possibly with and without an HD test signal present). Sometimes its the fine details that are found to better correlate with listening tests.
EDIT: One can also try looking at FFT data at different test signal levels and frequencies. In other words, there are some more obvious kinds of things to look at when trying to understand SQ differences before looking for more unusual types of explanations.
EDIT: One can also try looking at FFT data at different test signal levels and frequencies. In other words, there are some more obvious kinds of things to look at when trying to understand SQ differences before looking for more unusual types of explanations.
Last edited:
You should've measured the voltage at the speaker terminals. As I suspected, level matching wasn't done. You may want to try it again with the proper level matching.
I don't understand the marked term.
Is it this audio analyzer ? go to
RightMark Audio Analyzer. Products. Audio Rightmark
Latest News. Audio Rightmark
Is it possible to create the THD-character (fundamental notched out) like the diagrams in post #2 ? - go to
Choosing of best sounding OP AMPs for the lowest possible THD+N -really the best Way?
Last edited:
Sorry it took a few days to see that. Looks like the harmonic spectrums are different with the different opamps. That's despite computed THD coming out about the same. Makes it easy to see why THD and or THD+N are not considered to be very useful metrics. For one thing they don't say enough about the exact harmonic spectrum, which is something that matters for perceived SQ.
Of course what is shown in the pics is only one frequency and one signal level. Its a snapshot of differences at one operating point but that's about it.
Thank you for taking your time to look at them. I'll measure them again at different levels.
PS: One question, can you guess what is the cause of above 30K noise?
Looks like it could be sigma-delta dac noise shaping. Such dacs generate a lot of switching noise. A DSP filter type of algorithm called a 'modulator' shifts the noise up above audio frequencies so it can be more practically filtered out. Sometimes it isn't filtered enough and then can cause problems in downstream equipment.
Okay, this thread has not been updated for a long time, but I have recently recieved a package from digikey with OPA 1655/1656 and 1641/1642. Based on the positive reviews of the OPA1655/1656 i swapped the OPA2132P in my DAC against the OPA1656. The first impressions were: much more highs, but also a less impressive bass, a less fat sound. Despite the thinner sound, there was a hiss (or at least it was more present) in the voices. Something i absolutely did not expect and do not like. Same for the OPA1655, which should replace the four OPA134 in my preamp. All these OPA1655/56 were soldered on adapter by myself without any further mods/caps. I also ordered the OPA1641/42 and to my surprise this works much better for me both in the DAC and the preamp. Its again a full / rich sound, but with a more precise bass and very smooth and clean highs. The presentation is very calm and stable. Sounds very mature.
From the RMAA measurements all OPAs look very similar (at least with my simple audio interface Behringer UCA222).
Next to try is OPA1611/12. I do not expect a big difference, but some write, that the bipolar sound better.
From the RMAA measurements all OPAs look very similar (at least with my simple audio interface Behringer UCA222).
Next to try is OPA1611/12. I do not expect a big difference, but some write, that the bipolar sound better.
Interesting observations. If there would distortion diagrams with only the residual distortion (as show for various power amps in post#2) of all mentioned op amps, one would understand why one prefers the OPA1641/42 to the OPA1655/56.