Don't look too much at brand or supposedly better topologies but look at the final results instead. No one would have thought CMOS opamps would come so far and suggesting these for audio would cause people to make fun of you.
Now they are in the top 5. For audio. IMO they are even number 1.
Now they are in the top 5. For audio. IMO they are even number 1.
Last edited:
Indeed, I probably will be exhausting my current LM4562 for some 'fun' build, and slowly moving on to those new parts.
BTW, I believe the improvement is attributed to better circuit design and not some fancy semiconductor technologies according to TI representative.
BTW, I believe the improvement is attributed to better circuit design and not some fancy semiconductor technologies according to TI representative.
Definitely got noise, like any amp, 5.1nV/√Hz and 0.8fA/√Hz according to datasheet. That voltage noise is equivalent to 1k6 of series resistance on the input or in the feedback network.
Well < 1k would be more accurate. But for line level you may be happy with the 13nV/√Hz of a 10k noise source anyway, which allows a wider choice of opamps.
I noted that the OPA228 and OPA2228 have low current noise (as do the NE5532 and NE5534).
It depends on the application but I am impressed with the OPA228/2228.
However I found it hard to beat the humble RC5534 in my QUAD405-2. (RC5534, OPA1611 and OPA228 all performed very well. I did not have OPA134 to try.)
The RC5534 was almost free compared with the OPA1611 and the OPA228.
I recommend not laughing at the NE5534 and the performance it achieves in the Ti PCM1794 reference circuits.
It depends on the application but I am impressed with the OPA228/2228.
However I found it hard to beat the humble RC5534 in my QUAD405-2. (RC5534, OPA1611 and OPA228 all performed very well. I did not have OPA134 to try.)
The RC5534 was almost free compared with the OPA1611 and the OPA228.
I recommend not laughing at the NE5534 and the performance it achieves in the Ti PCM1794 reference circuits.
Last edited:
you're right but noise is not only a number in a datasheet. I used this AOP in military system where there are a lot of noise sources around and this is the best AOP I test for this application. Of course power supply of AOP is very important ("radio" LDO is the best choice because they have the minimum noise on the output line).
Compare noise of R to AOP is incorrect if you're not taking account the gain gain of your cicruit
If gain is 100 then R noise have less contribution to the total noise
yes, if you have a high input Z (like for guitar ampifier), then the parameter to take into account is the current noise (more important than voltage noise to compare between AOP) so a JFET AOP is a better choice.
The OPA228/2228 seemed interesting because they are bipolar but with quite low current noise for a bipolar input. The NE5534/5532 are also bipolar but with quite low current noise.
I mentioned them since many FET op-amps have higher voltage noise and also distortion. (But newer designs are getting better and better.)
The RC5534 surprised me in the QUAD405-2 giving almost identical results as the OPA1611A that I bought just for the QUAD405-2. The RC5534 and OPA228 gave quite similar results. Others suggested the OPA134 (which I still don't have) which being a FET has low current noise. So yes, sometimes the lower current noise design wins.
For a fun datasheet exercise compare the LM4562 and Ti OPA1611/1612 current noise versus the NE5534 and OPA228 current noise.
I mentioned them since many FET op-amps have higher voltage noise and also distortion. (But newer designs are getting better and better.)
The RC5534 surprised me in the QUAD405-2 giving almost identical results as the OPA1611A that I bought just for the QUAD405-2. The RC5534 and OPA228 gave quite similar results. Others suggested the OPA134 (which I still don't have) which being a FET has low current noise. So yes, sometimes the lower current noise design wins.
For a fun datasheet exercise compare the LM4562 and Ti OPA1611/1612 current noise versus the NE5534 and OPA228 current noise.
Last edited:
OpAmp such OPA1641 has been designed for a such environment by adding a RFI filter inside the amplifier.
OPA1656 has same feature and a big output current if necessary. Some better performances about noise too. So it depends on price but for 2 or 3 AOP, it's not a big deal
OPA1656 has same feature and a big output current if necessary. Some better performances about noise too. So it depends on price but for 2 or 3 AOP, it's not a big deal
Today I swapped TLE2074 in older Apogee D/A converter to OPA1644 (quad OPA1642). Different package is so workaround was needed.
I'm wondering why OPA1644's audible sound quality is so much better while RMAA measurement result is about the same as 2074...
Cyrrus CS4272 DAC chip spec is not really good, so is it the bottle neck?
I'm wondering why OPA1644's audible sound quality is so much better while RMAA measurement result is about the same as 2074...
Cyrrus CS4272 DAC chip spec is not really good, so is it the bottle neck?
Just goes to show that there is more to SQ than RMAA type analysis measures.
Regarding a bottleneck, probably there is not one but several. Usually a well designed dac is designed to balance out the engineering trade offs. For example, don't spend more money on a PCB than the rest of the design can make the good use of, etc. Don't spend more money on a clock than the rest of the design can make good use of. And so it goes.
Beyond that, usually the retail cost of a consumer product is about 3 to 5 times the incremental cost of manufacturing, packaging, and shipping one unit from the factory. For high end audio devices the retail markup is often more like 6 times, since sales volumes in that market tend to be small.
Regarding a bottleneck, probably there is not one but several. Usually a well designed dac is designed to balance out the engineering trade offs. For example, don't spend more money on a PCB than the rest of the design can make the good use of, etc. Don't spend more money on a clock than the rest of the design can make good use of. And so it goes.
Beyond that, usually the retail cost of a consumer product is about 3 to 5 times the incremental cost of manufacturing, packaging, and shipping one unit from the factory. For high end audio devices the retail markup is often more like 6 times, since sales volumes in that market tend to be small.
Likely it's just a perception stemming from subjective listening method. If it was actual audibly different sound quality, it would have revealed it in objective listening comparison.
It is not. I fed signals from output of 8ch DAC (2ch modified) to 8ch ADC, matched level 0.0db difference and ABX tested with custom macro. Actually I did not have to do it because the difference is way too obvious, especially mid low frequency congestion. They are measured with RMAA DAC ADC loop test, the number is almost the same. I expected to see some difference, so I was surprised to see the test result...
Last edited:
Th RMAA measured result is about the same as DAC chip's catalog spec, except THD+N is 8dB worse, about 82dB. So I wanted to check if a better op amp can improve the spec, but it did not. I can only hear some difference in listening test. I guess it could be a power supply noise, but I don't know. The converter is a 10 years old $2500 mid class professional ADDA, Apogee Ensemble. Apogee is famous for its distinct coloration, so I was not surprised to see TLE2074 are there...
Last edited:
I started thinking that measurement and listening test mismatch is caused by the bottleneck of the signal chain which hides the smaller difference in different parts.
We can't see any meaningful difference between 2 DACs from the speaker measurement, but we can still clearly hear it, because speaker is the bottleneck. It's most likely the same thing when we measure DAC itself, if some part of the DAC is really a bottleneck, we can't measure the difference of the other parts of the DAC while it is audible.
We can't see any meaningful difference between 2 DACs from the speaker measurement, but we can still clearly hear it, because speaker is the bottleneck. It's most likely the same thing when we measure DAC itself, if some part of the DAC is really a bottleneck, we can't measure the difference of the other parts of the DAC while it is audible.