[trodas]

I have a dilema and I would like to hear your recommendation for best audio dual SOIC opamp, running at +5V only. Not much is there to choose from, but I'm sort of torn between these there:

AD823 Analog Devices: AD823: Dual, 16 MHz, Rail-to-Rail FET Input Amplifier :: Operational Amplifiers (Op Amps) :: Amplifiers and Comparators
LM4562 LM4562 - Dual High Performance, High Fidelity Audio Operational Amplifier
AD8397 Analog Devices: AD8397: Rail-to-Rail, High Output Current Amplifier :: Operational Amplifiers (Op Amps) :: Amplifiers and Comparators

AD823 got a bad name for being too harsh and flat sounding (and hard to get one), LM4562 is picky a like to pick up interferences, not to mention still being a bit too much harsh/sharp ... (but I have few at hand) and I did not know much about AD8397, except that it could be get as sample easily.

The opamp will be used as LineIn input opamp on my X-Fi. Replacing a MC33078 opamp.

So, guys, are any of these even better that the TI original MC33078? And witch one would you recommend to install?

 

[error401]

I quite like LME49720, which is the 'proper' name for LM4562 (they're the same chip). AD8397 is also quite nice, but it's notorious for being difficult to keep stable.

I think I would go with the National chip, AD8397 is not really designed for an input stage, it's designed to drive heavy loads, so I think it's less likely to work well.

You might also check out AD826. I hear it sounds good, but I haven't tried it for audio.

 

[amb]

Check out AD8656. It's a nifty low-voltage rail-to-rail dual opamp, very low noise (2.7nV/√Hz), high output current (220mA), CMOS (low input bias current of 1pA), and even sounds quite nice driving headphones directly. Max supply voltage is 5.5V, so output voltage swing is limited (but the fact that it's rail-to-rail helps a lot).

 

[grenert]

I was clued in by one much wiser than I in these matters that the OPA2365 is an awesome low-power opamp. I've tried it in several applications (headphone amp driver (with buffer), PCDP line out buffer, PCDP headphone driver) and I completely agree.

 

[trodas]

Guys, thanks all for the suggestions!

error401 - you seems to favor LM4562, witch require cooper heatsink as form of EMI protection (it is a picky opamp), but then it should be good...

AD826 - slew rate - 350V/us, noise 15nV/Hz, supply 5-36V (CMRR 80dB)
AD823 - slew rate - 22Vus, noise 16nV/Hz, supply 3-36V (CMRR 75dB)
LM4562 - slew rate - 20V/us, noise 2.7nV/Hz, supply 5-34V (CMRR 120dB)
AD8397 - slew rate - 53V/us, noise 4.7nV/Hz, supply 3-24V (CMRR ~90dB)
AD8656 - slew rate - 11V/us, noise 2.7nV/Hz, supply 2.7-5.5V (CMRR 100dB)
OPA2365 - slew rate - 25V/us, noise 4.5nV/Hz, supply 2.2 - 5.5V (CMRR 100dB)

I must say I'm a bit affraid to use opamp so close to their minimal voltages (LM4562, AD826) - the AD8656 is clearly designed for low voltage that I have there and it has a very low noise too... But will the slower slew rate 11V/us matter?
Or will LM4562 keep the 120dB even at 5V?

AD8656 has some advantages, mainly it is for the low voltages designed and since +5V is near his max, it will perform maybe better that LM4562 for witch is the +5V the total low end of usable voltages...
OPA2365 is kinda interesting too, but a bit higher noise is there and also "only" 100dB... But it is clearly optimized for low voltages, so, good choice, likely.

I just wondering witch give me better results when using them as input opamp before the ADC in my X-Fi on line in input. With the MC33078 I got these results:
RightMark Audio Analyzer test: X-Fi Fatal1ty MKT input caps, AD8599/MC33078 opamps
(maybe it is a time to better shield the caps in the LineIn input... the stereo srosstalk results are bad, IMHO)

 

[majkel]

The best if works is AD8022, if not, add AD8616 to your list. Save your time and money for AD826, it's crap for audio.

 

[error401]

I have used LME49720 without any kind of EMI protection. My phono pre is using two of them, and it's not even in a case at the moment. I've had no problems with it. I don't think it's as hard to work with as you think, but I haven't tried it inside a noisy PC case either

.

I think your analysis is spot on, and I would take amb's advice in high regard, he certainly has more experience with this than I do. Specs are not so important, all of these opamps have specs that far exceed what's needed in audio. I'd be inclined to go with the AD8656 given the choice you're looking at. You know how it's going to perform since it's specified at 5V. It's cheaper than LM4562 too.

Though I doubt it will have a major impact on your RMAA results. It's still an X-Fi.

 

[trodas]

Quote:

I have used LME49720 without any kind of EMI protection. My phono pre is using two of them, and it's not even in a case at the moment. I've had no problems with it.

Yep I trust you, but my friend did use LM4562 in his highly EMI - polluted case and after he by my suggestion added cooper heatsinks on them, he get even RMAA results +2dB improved, so... LM4562 is likely good choice, but known to be a picky suxxkas, love to pick uz much EMI w/o shield...

majkel - thanks, AD8022 is certainly very intersting opamp to consider.

AD826 - slew rate - 350V/us, noise 15nV/Hz, supply 5-36V (CMRR 80dB)
AD823 - slew rate - 22Vus, noise 16nV/Hz, supply 3-36V (CMRR 75dB)
LM4562 - slew rate - 20V/us, noise 2.7nV/Hz, supply 5-34V (CMRR 120dB)
AD8397 - slew rate - 53V/us, noise 4.7nV/Hz, supply 3-24V (CMRR ~90dB)
AD8656 - slew rate - 11V/us, noise 2.7nV/Hz, supply 2.7-5.5V (CMRR 100dB)
OPA2365 - slew rate - 25V/us, noise 4.5nV/Hz, supply 2.2 - 5.5V (CMRR 100dB)
AD8022 - slew rate - 50V/us, noise 2.5nV/Hz, supply 5 - 24V (CMRR ~95dB)
AD8616 - slow rate - 12V/us, noise 8nV/Hz, supply 2.7 - 6V (CMRR 100dB)

AD8022 has very fast slew rate, lowest noise from all and it is high on the dB too... Now given the application as input opamp for LineIn/Mic - it is more important to grab on the LM4562 with lower slew rate but higher dB, or vice versa - took as favorite the high slew rate and lowest noise and very good dB ratings of AD8022?

 

[majkel]

For Mic input use FET input or MOSFET op-amp, and look at the current noise rather than voltage noise. The AD8022 has got significant input bias current which makes it unsuitable for input stages, especially when you use high impedance sources like microphones. For line-in, it doesn't matter that much. LM4562 has got pretty good specs as for a bipolar input op-amp but it sounds mediocre to my ears. AD823 is tolerable but it sounds better when supplied from higher voltages. Finally, don't look at the slew rate, all of them have got it sufficient for audio. The ones that remain interesting for you are AD8656, AD8616 and AD8066 IMHO, the first preferably but I don't know its sound signature.

 

[alphaman]

The AD8656 is a relatively modern (= mostly unknown) device from Analog Devices and is very nice! It's used in the line-out section (after WM8740 DAC) of QLS QA-350 DAP. The newer QA350 v2 uses AD8397 in the headphone-out (HO), replacing v1's MAX9722A. I never use the HO so I can't comment on that. I did experiment with rolling various OPAs, including LT1364, but still like the stock AD8656 best in the QA-350. QLS's Clark Chen has noted that AD8397 can sound better (than AD8656) for LO as well. I have samples on order and will comment accordingly ... stay tuned ...
Note: bypass it with a 47-100uF low-ESR high-quality electro cap (Panasonic FM, Sanyo OS-CON) and SMD PPS 0.1uF, as close to the OPA supply pins as possible (I soldered these right across the top of the SOIC OPA itself!).