According to the TI datasheets OPA1612 has less noise than OPA1656. But that does not say anything about the SQ.
OPA1612 sounds better than any transistor preamp I've ever heard and better then my own 6N6P 2 case -no barrels hold- 36kg monster preamp with a 100% double blind test win rate. Yes it destroyed many many extreme expensive commercial preamps in the last 25 years. The OPA1656 is maybe even better, ahhh I like this hobby. The sound is getting better and better and the beloved music nearer and nearer... And even much much much cheaper...
OPA1612 sounds better than any transistor preamp I've ever heard and better then my own 6N6P 2 case -no barrels hold- 36kg monster preamp with a 100% double blind test win rate. Yes it destroyed many many extreme expensive commercial preamps in the last 25 years. The OPA1656 is maybe even better, ahhh I like this hobby. The sound is getting better and better and the beloved music nearer and nearer... And even much much much cheaper...
Last edited:
Yes, that is what it means. "How much more?" is not a trivial question to answer. It depends on how you define noise (unweighted, A-, C- or Z-weighted), and the source and feedback shunt resistances.
Let's compare the noise density at 100 Hz for each of these amps.
--------------Noise in nV/rtHz
OPA1611 --- 1.5
OPA1642 --- 5.8
OPA1656 --- 11.8
OPA2156 --- 12.0
Now for 10 kHz (the front page data sheet number)
--------------Noise in nV/rtHz
OPA1611 --- 1.1
OPA1642 --- 5.1
OPA1656 --- 2.9
OPA2156 --- 4.0
For the OPA1611, you also have to consider current noise, because it's not a FET input like the others. If your Rin is 1Kohm, input noise rises to 4.5 nV/rtHz. It's now similar to the others at 10 kHz, but still wins at 100 Hz.
Last edited:
Used OPA1656 in I/V of PCM1792 in dual mono configuration. Tried in this place OP275, OPA1612, OPA1642, OPA1656 and OPA828. Best SQ have with 1656. May be 828 will be better, but i have to do some thing in the board to extract best of it. In final have to say, that OPA1656 is best of the best in I/V for me to now. 
The main noise contributor to your signal path is (or should be) the first amplifier where the majority of your gain is. Any noise before the gain (including input noise of the gain amp itself) is amplified by the gain.
For old-time audio, that would be phono input or the tape deck head amps. For modern audio, that is the DAC I-V converter. Once that amp is out of the way, you don't really have a need for op amps with input noise below 5 nV/rtHz. You care more about how they interact with filter RC networks, and their ability to drive audio cables and 600 ohm line impedances.
For old-time audio, that would be phono input or the tape deck head amps. For modern audio, that is the DAC I-V converter. Once that amp is out of the way, you don't really have a need for op amps with input noise below 5 nV/rtHz. You care more about how they interact with filter RC networks, and their ability to drive audio cables and 600 ohm line impedances.
When I worked at TI, the Burr Brown guys out in Tucson were considered the best. They worked hard to maintain their "Burr Brown" culture, and I believe they have managed to do so. Someone came along about 10 years ago and lit a fire under them to define the state-of-the-art in audio op amp performance, and not take a back seat to AD/LT. I don't have any insider knowledge, but that's the same time frame that TI bought National. There may have been some "synergy" there, or something.
@Russell
If you want to know more about the OPA1656, John Caldwell from TI explained some of the design choices in the OPA1656 thread.
OPA2156: Rail-to-rail inputs, uses slew-boosting, trimmed for lower input offset voltage
OPA1656: Pole-zero compensation for higher open loop gain in the audio band
Those are the main differences I remember (and that are also reflected in the datasheets), maybe there are more that I forgot about. Not going to read the whole thread again though
If you want to know more about the OPA1656, John Caldwell from TI explained some of the design choices in the OPA1656 thread.
OPA2156: Rail-to-rail inputs, uses slew-boosting, trimmed for lower input offset voltage
OPA1656: Pole-zero compensation for higher open loop gain in the audio band
Those are the main differences I remember (and that are also reflected in the datasheets), maybe there are more that I forgot about. Not going to read the whole thread again though
Very few high-end audio applications in the signal path need low offset or RRIO. Low voltage ADCs are about all I can think of. That's why I prefer the OPA1656 and don't really think of the OPA2156 as an audio amplifier.
I do some current sense functions in power amplifiers using RRIO op amps with low offset. I don't really care about the THD or noise of those functions. I just need the input range to include the supplies and have enough bandwidth to accurately measure the current of a 20 kHz signal.
Hey, look, I've brought the thread back on topic of "Solid State Power Amps" by mentioning power amps.
I do some current sense functions in power amplifiers using RRIO op amps with low offset. I don't really care about the THD or noise of those functions. I just need the input range to include the supplies and have enough bandwidth to accurately measure the current of a 20 kHz signal.
Hey, look, I've brought the thread back on topic of "Solid State Power Amps" by mentioning power amps.
I think you have a VERY good point here---a view of opamp noise that I had not previously noticed. BUT, just for the record, let's correct some things in your post;
The "front page data sheet number" is generally for 1 KHz, not 10K. Also, the numbers you have quoted for the opamps ARE @ 1KHz; except for the OPA 1656---which you have as 2.9, it should be 4.3. The 2.9 figure is the voltage noise density @ 10 KHz.
I'm wondering why the OPA1656/2156 are so much worse at the low frequencies?? Is it something to do with the FET front end or ???????
You are really picking at nits here. I simply wanted a frequency well above the 1/f knee. I guess they don't always use that number on the front page. Sue me. I told you these are the 10 kHz noise numbers, and they are.
Yes. It's the flicker noise of the input pair. In order, BJTs typically have the lowest flicker noise, followed by JFETs, and then CMOS. This paper talks about it. At least 95% of the noise of a well-designed op amp is from the input pair. I could say more, but I'd be repeating information in the link.
https://www.ti.com/lit/an/sboa355/sboa355.pdf?ts=1614581054089
Thank you Russell for providing this info, it sheds a good light on the company that I like so much.
Yes. OPA1656 is a bit noisy at low frequencies.
Below is a comparison between OPA1656 and OPA1642. Two identical gain 11 amplifiers were tested. Every amp. consists of 2 x 10 op-amps in parallel and the two 10s are balance connected. The input is essentially shorted with 10mohm resistor in order to have a -140db control signal. Idea taken from Dicks site. The output of the 1656/1642 amplifiers is additionally amplified 90db. The measuring setup is calibrated in dbV and Virtins shows the noise spectrum - so no additional spectrum corrections are needed.
In this particular implementation 200nV/sqHz for OPA1656 and 40nV/rtHz for OPA1642
Below is a comparison between OPA1656 and OPA1642. Two identical gain 11 amplifiers were tested. Every amp. consists of 2 x 10 op-amps in parallel and the two 10s are balance connected. The input is essentially shorted with 10mohm resistor in order to have a -140db control signal. Idea taken from Dicks site. The output of the 1656/1642 amplifiers is additionally amplified 90db. The measuring setup is calibrated in dbV and Virtins shows the noise spectrum - so no additional spectrum corrections are needed.
In this particular implementation 200nV/sqHz for OPA1656 and 40nV/rtHz for OPA1642
Last edited: