Question: What’s a good Op Amp for audio? The answer: depends. There are many factors to consider and there’s no single best answer. There was a time when I was quite happy with the cheap and simple TL072 for most tasks. If I needed something better, then there’s the more expensive (at the time) 5532. It’s notable that after all these years, these critters are still quite capable…
Note: this blog entry has been superseded by the Op Amp Shootout page. The page is continually being updated.
I was thinking of suggesting the excellent TI OPA2277 and OPA2107 as these are ridiculously good in the various pre-amps I make. They are however crucifyingly expensive….however TI will happily send you three of each as samples….
OPA2277
Slew Rate – 0,8v/µS
GBP – 1MHz
THD+N – 0,002%
Supply – 4v .. 36v, 800µA/amp
OPA2107
Slew Rate – 13-18v/µS (!!!)
GBP – 4.5Mhz@G=100
THD+N – 0,001%
Supply – 9v .. 36v, 4.5mA/amp
Thanks, Carl! OPA2277 is not too expensive at $2, but the OPA2107 is rediculosuly expensive at $10!
They don't come quite close to the specs of the more modern Op Amps listed in the table though, do they? I say modern, because the specs for OPA2107 is dated January 1989 while OPA4277 is dated March 1999.
Take a look at the specs for LME49990 in comparison, which is a very recent design dated December 2009 (http://www.ti.com/lit/ds/symlink/lme49990.pdf). Check out the specs of that critter and compare.
Great point! I've never dated these op-amps and I'm surprised at how "old" they are. Still, far superior to the 4558 which pedal fanboys clamour over.
In my experience, the first two of the points in your checklist are not that important in non-hifi unity gain applications such as guitars. Micro power op-amp's seem to suffer in the slew rate area at ultra-low voltages which directly impacts high end. The pre-amps used in EMG pickups for example, are run very close to the bone…. any lower and the treble would roll off.
I fear that using a low supply voltage would be a large problem since pickups can create quite high transients requiring significant headroom. My "darling" pre-amp buffer circuit uses dual op-amps such as the 2277/2107 in a differential configuration to totally kill noise but uses a relatively high supply of ±9v. I'll email you the design when I get to my desk….
Here's my take on this:
I agree about 2 (Low THD+N) not being that important in a non-hifi application. To some extent, I would even embrace the harmonic distortion as long as it is harmonically pleasing. The coil in itself would be the prime source of coloration.
As for 1, however (Low noise), keep in mind that later stages (e.g. compressors, overdrives and distortions) along the signal path will amplify any noise present in this stage. This noise ends up as hiss.
As for the headroom, IMO, it's just a matter of scale. The guitar for example can go up to 4v peak to peak. Scaling that down to 1v peak to peak (6db attenuation) at the source should give ample headroom to capture all the transients. That attenuation can be reclaimed at the other end before going through the Amp, if you want to go direct. For this project however, I think going straight to digital is best and for that matter, any signal normalization level would be just fine as long as there's sufficient resolution. Then again, that's where the low noise factor comes in. Noise will drown the low bits hence a sufficiently low noise floor is crucial.
Carl, would you mind if I take this discussion to the cycfi forum (https://plus.google.com/communities/101615575500130501032)? The word-press comment system is not that good. Or, we can also move the discussion over to the Project Guitar forum, for a livelier discussion. I'm loving that forum. I wish though that it had sub-threads. An overly long single thread is just crazy (e.g. the sustain thread that kept on sustaining forever 🙂 ).