Hey guys had great success in the past using AD844 (with no nfb), just want simple this time, got a PCM1796 dac, looking for what's the best dual I/V opamp today. Even singles at a crunch, if I get a Brown Dog converter boards.
Just looked at the old AD812 seems to have some good specs for I/V duties.
What do you guys think
145mHZ
1600 V/us slew rate
50ns settling time
-90db distortion
can't find out if it's unity gain stable or not.
Cheers George
Just looked at the old AD812 seems to have some good specs for I/V duties.
What do you guys think
145mHZ
1600 V/us slew rate
50ns settling time
-90db distortion
can't find out if it's unity gain stable or not.
Cheers George
This looks good, what about the distortion though, good enough???
Second Harmonic f = 10kHz −110 dB
Distortion(3) f = 5MHz −50 dB
Third Harmonic f = 10kHz −105 dB
Distortion(3) f = 5MHz −50 dB
Cheers George
Last edited:
Hi Abraxalito, yes it looks good also, but not easy just to play with.
I'm cradled in this dac and have tried many different ones, and so far the old OPA2132 has come up trumps, with clean sounding treble, great mid body, a nice strong bass if just a touch thick.
What do you think about the LM6172 for PCM1796? I've always valued your opinion, got some samples coming.
Cheres George
I'm cradled in this dac and have tried many different ones, and so far the old OPA2132 has come up trumps, with clean sounding treble, great mid body, a nice strong bass if just a touch thick.
What do you think about the LM6172 for PCM1796? I've always valued your opinion, got some samples coming.
Cheres George
LM6172 has long been one of my recommendations for I/V ever since I tried it in an Asus soundcard many years ago. I think that card was using PCM1792 so probably its well suited to 1796 too.
I'm working on a DAC using AD815, first gen PCBs have some errors, hopefully after a re-spin I'll share gerbers.
I'm working on a DAC using AD815, first gen PCBs have some errors, hopefully after a re-spin I'll share gerbers.
Unpopular opinion-
If you are talking about measured performance in the audio band, it probably is OPA1612 or LME49720. Yes, I know everyone likes to think you need GHz op-amps for I/V.
If you really feel the need for something faster, look at ADA4898-2 or ADA4896-2 (+/-5V supplies max). Their distortion is not specified in the audio band.
OPA1656 is rather interesting for this, and if you like OPA2134 then you probably won't do any worse with OPA1642.
I would not use any fast op-amp with an adapter. I'd also want to verify anything faster than what the board was designed with is stable.
Last edited:
I tried ada4898-1 on a pcm1701 output(less than 1ma max iout) at +9v/-12v suply and in one minute was melting ...dead.
I tried lm6172 on tda1541 in Sony cdp-750(similar to nakamichi cdp 2e) who had only one ic for both channels so ut made a lot of sense using lm6172 and the rsult was magnificent.I had to admit that i fully appreciated a very old post made by abraxalito where he said that lm6172(at that time) was the best option.
Now i think that for a classical i/v based on virtual ground in closed loop op-amp we need to assess the noise of the circuit and its needed speed.
For modern dac's you don't need so much higher speed, opa1611 oe opa1612 is probably the best option, for old dac in NOS or even classic oversampling but still noisy output i have to agree that faster op-amps are better.Here's a catch though:
lm6172 works best on tda1541 because of its higher iout .
Some other old dacs of the past are much lower iout , 4 ...40 times lower iout than tda1541 and you need lower noise op-amps.I found THS4631(fet op-amp) better for its very low current noise while its voltage noise is only 2nv which is again better than many op-amps outhere.I didn't try it though...
I tried lm6172 on tda1541 in Sony cdp-750(similar to nakamichi cdp 2e) who had only one ic for both channels so ut made a lot of sense using lm6172 and the rsult was magnificent.I had to admit that i fully appreciated a very old post made by abraxalito where he said that lm6172(at that time) was the best option.
Now i think that for a classical i/v based on virtual ground in closed loop op-amp we need to assess the noise of the circuit and its needed speed.
For modern dac's you don't need so much higher speed, opa1611 oe opa1612 is probably the best option, for old dac in NOS or even classic oversampling but still noisy output i have to agree that faster op-amps are better.Here's a catch though:
lm6172 works best on tda1541 because of its higher iout .
Some other old dacs of the past are much lower iout , 4 ...40 times lower iout than tda1541 and you need lower noise op-amps.I found THS4631(fet op-amp) better for its very low current noise while its voltage noise is only 2nv which is again better than many op-amps outhere.I didn't try it though...
I've used ADA4898 before at unity gain and it didn't melt, so something was definitely wrong. The decoupling and layout have to be decent for high speed amplifiers.
Use LM6172, add BUF634 or any preferred buffer. It needs because 1796 output current are a bit on the high side and you better to decouple opamp crystal from relatively high I/V resistance.
I never forget the 100 ohms series resistor with the feedback integrating capacitor...50 ohms might work too...AD797 datasheet is often ignored...lm6172 is way too fast to be certain that your buffer will not loose the phase response. That huge iout current is not needed.
This is , i think, the least understood wrong practice of buffering dacs to get the most out of them ignoring that you need to buffer that voltage further and even decrease the output by 4...8x to get a clean line out.
Higher i-out dacs are used to improve the SNR ratio , but then you loose that original SNR to poor CMRR and PSRR in the filters and buffer stage because of unity gain stages and resistor dividers or rlc filtering devices used to further filter the i/v stage output.
The most common effect of these stages is adding noise because most of the unity gain high bandwidth op-amps aren't great at current noise specs and some not good enough at voltage noise either.
Making the i/v stage output smaller, the high frequency noise is decreased, the clipping is softer at lower rail supply voltages, the thermal behavior and added nonlinear distortions are improved and i can use old days cheap ultra low noise audio use(lower bandwidth) op-amps with a little bit of gain after the i/v stage.
I started my own little research a few months ago and i found that even the standard 2vrms is too much output and i don't get more than 0.5...1vrms out of any dac just to give a gain to the output op-amps buffers.This way , any i-out dac will sound better, even those well known for bad sounding...
You might find out that you don't even need high Q ultrasonic filters if you're using high bandwidth op-amps all the way down to your speakers because those op-amps are not producing their own artifacts which seem to be the real culprit in high frequency noise endangering the audio equipment .The speakers themselves will kill that HF noise and your ears will never hear anything, but the music.
Last edited:
The ADA4898-1 uses a die attach pad, which must be soldered to a pad on the PCB that is attached to V- (the datasheet shows the details). If you use a proper footprint, they run quite cool and are lovely to use. Too bad for me, their distortion is a little on the high side, with a surprising amount of 2nd harmonic. But, you absolutely need to deal with their die attach pad - they idle at 8mA each, and that's 0.24W at ±15V.
I know...i had quite a big radiator attached...there must have been some sort of oscillation unless its sensitive to unbalanced rails...it's weird, but i think that these op-amps should be used at +-6v max...I'm thinking of a different strategy , to use bootstraped rail supply with ada 4899 which has its pad at ground.
Last edited:
lol 
In this case of vintage choice I would more try NJM 4580D : bipolar, very neutral, simple scheme... smal slew rate though. Certainly worth to try instead the NE5532 which is a little dark and veiled... But I disgress..."These days" asked the op.
So if no AD844 or OPA861 as per Georgehifi older post choices, perhaps todays
opa1642, opa1656, opa1671, opa1678, opa826, opa827, opa2810...are more on topics ?
OPA1656: High-Performance CMOS Audio Op Amp
In this case of vintage choice I would more try NJM 4580D : bipolar, very neutral, simple scheme... smal slew rate though. Certainly worth to try instead the NE5532 which is a little dark and veiled... But I disgress..."These days" asked the op.
So if no AD844 or OPA861 as per Georgehifi older post choices, perhaps todays
opa1642, opa1656, opa1671, opa1678, opa826, opa827, opa2810...are more on topics ?
OPA1656: High-Performance CMOS Audio Op Amp
There are three old chips that are better than ne5532 as i/v stage, but there's no discussion whatsoever on DIYaudio about them and i won't ruin the secrecy tradition mainly because one of them was reactivated after 15 years for a short period of time at Mouser and is programmed for EOL, the stock being limited.I will only tell the fourth , which is njm2114.It was highly disregarded as it wasn't as good as ne5534 in phono preamps because of its lower input impedance, but for i/v circuits is quite good.
It looks like there are small conspirations here and there cause expensive chips need to be sold...
As a separate issue, no matter how stupid it sounds, i think that there's a real conspiration in the whole electronics industry about the use of op-amps in inverting topology as attenuator as there's not a single book or application note on talking about its performance.You can only find some small talks here and there about that finding that you can easily make a NON-stable op-amp working in a negative gain stage with better performances than in a 10x gain as noninverting op-amp...
For example this all that Analog Devices have to say about it:
https://www.analog.com/media/en/analog-dialogue/raqs/raq-issue-56.pdf
When we actually had Soundcraft mixers...and many ultra professional products using it!
First they'd tell us that it's too noisy, then they show us circuits that never meet the noise criteria we really need in exchange for ultra low distortions that we never needed!
It looks like there are small conspirations here and there cause expensive chips need to be sold...
As a separate issue, no matter how stupid it sounds, i think that there's a real conspiration in the whole electronics industry about the use of op-amps in inverting topology as attenuator as there's not a single book or application note on talking about its performance.You can only find some small talks here and there about that finding that you can easily make a NON-stable op-amp working in a negative gain stage with better performances than in a 10x gain as noninverting op-amp...
For example this all that Analog Devices have to say about it:
https://www.analog.com/media/en/analog-dialogue/raqs/raq-issue-56.pdf
When we actually had Soundcraft mixers...and many ultra professional products using it!
First they'd tell us that it's too noisy, then they show us circuits that never meet the noise criteria we really need in exchange for ultra low distortions that we never needed!