Operational amplifiers selection for audio

I think this is the most difficult and controversial part in nowadays audio.

Still is a big question which one are sounding better: made with tubes, from discrete elements (transistors) or IC.

First section (tubes), will be excluded from this presentation, because is an endless story - and always tube funs comes with warm sound of tubes - but I think this warmness rarely can be found in real music, but certainly is very pleasant.

Pro and contra of discrete operational amplifiers: you could not made them perfectly similar with each other's regarding tolerances of components and also different time made batches or different factories made semiconductors parameters never will be same. If you consider also the size of those discrete opamps, certainly the IC will be the winner.

Now we arrive back to the basic question: which is the best low signal operational (integrated) amplifier for audio application?

Surfing on the net, you can meet lots of forums, opinions, and is difficult to find a clear answer - but this just highlight how difficult and complex is this question.

Some common mistakes:

- Operational amplifiers not always are interchangeable with each other. Some has fet inputs, some bipolar. If you put an opamp with bipolar input instead of opamp with fet input, the sound probably will be worth.

- Technical specifications are very useful, but please consider: the sound of any amplifier cannot be described with technical parameters! This is like, when you go to market, and would like to buy a CD player or power amplifier - in their technical specs all data are stunningly perfect (also for low grades models) THD is incredibly low, and S/N are over 100dB, but if you listen them, often equipment’s with best specs sound worst!

- The price of an opamp are not related with its sound's quality, an expensive one does not mean its sound will be better,

- If you change a cheap opamp in a cheap equipment with an expensive, so called "best on market" one, do not be surprised about the result -> will be under your expectation. Why? Because the whole environment does not match with new opamp requirements – here I mean the power supply cleanness, noisy and low grade components/cables, not well designed PCB pattern, etc. In equipment design, especially for consumer electronics, one of the most important factor of product are materials/components cost. Be sure, they did not save money using just a cheap opamp, and the rest of components are audiophile grade! Forget this! The best makers are using normal components (means cheap or midrange category). Their knowhow are to make a perfect synergy of whole system and hundred hours design and listening tests, which cannot be compared with any homemade trials. That does not means to stop your "improving" activity! The fortune can appear on your side, you can be the lucky guy who find a better solution – but this never will be applied by manufacturers, because the material cost will be too high for them.

Theory: if you are not familiar with technical terminologies used in catalogs/datasheets, please read this short summary about them (How to Chose An Op Amp by Qool Jaq on qooljaq.com).

More details about Myths and facts of operational amplifiers can be found on this link posted by Northwest Audio & Video Guy on his blog page.

Summary of best operational amplifier ICs:

OPA627. Almost on any forums are marked as best opamp for audio application. Very expensive.

OPA637. Uncompensated versions of OPA627, sometime marked as "has more open sound picture then OPA627", because has no internal compensation capacitor.

OPA827. Are recommended as replacements for OPA627 by TI, but audioholics does not believe it.

LM4562 has a natural, dynamic sound in whole sound spectrum. It's sound is not as warm as OPA627, but is a good choise.

LME49990 I never try it

OPA1611 is good, but less dynamic then LM4562

NE5532. Best are the original Philips ones. This is widely used by semi- or professional manufacturers. This IC can be a good basis for improvements of old equipments. Some DIYers marking it with warm but boring sound. I think, for some circuits are an ideal solution, especially for pure digital circuits, like DACs.

LM6172 I never try it.

OPA2134 often are mentioned as replacement of cheap opamps, but has a bit harshi sound

AD797 precize, cold sound, very low noise. Are mostly used in RIAA for MC or microphone preamplifiers

AD825 is also a good choise for replacements, including the very low noise power supply circuits

AD8066 or AD8065 has good, pleasant sound, is very fast with wide bandwidth

AD8397 high output current, good sound ->headphone and line driver

Please consider above list just as it is, is not an order, and intentionally I did not recommend, which is the best from those. In some application can be the OPA2134, in other OPA827 or LM4562. Try them yourself, and which sounds better for you, use it!

Sometime upgrading funs forget to listen the music, their attention are focused on exceptionally small differences regarding changed opamps, capacitors, cables, resistors etc. Till some level I also accept it, but I think we should reserve time for listening, not just critically check the sound of components.

Some manufacturers, like TI gives a dedicated page, where they recommend which opamp optimized for audio applications: Upgrading Op Amps in Audio Equipment. Is a forum discussion.

Interesting pages or forums about operational amplifiers used in audio applications:

Op Amp Shootout at Cycfi Research

Chiarra Grand Op Amp test from Rock Grotto blog

Flavors of Audio Op-Amps on Audio Reviews by Bram Probadi

Which opamps are best at pinkfishmedia forum

What's the absolute best opamp for audio purposes? at All About Circuits forum

best op amps at Head-fi forum

Best sounding cheep op-amp for CMoy's? at Head-Fi forum

Best sounding op-amp at AudiogoN forum

Notes on Audio Op-Amps at TangentSoft

Operational amplifiers noise figure

Operational amplifiers input noise specifications are usually given in terms of nV/vHz for noise voltage and pA/vHz or fA/vHz for noise current and are therefore directly comparable with resistor thermal noise. Due to the fact that noise density varies at low frequencies, most op amps also specify a typical peak-to-peak noise within a "0.1Hz to 10Hz" or "0.01Hz to 1Hz" bandwidth.
These equations calculate the noise figure of an operational amplifier using parameters provided by vendors at the operating frequency and gain configurations. Be sure to note units both for the input parameters and for the equations, or you will end up with really bad results.

Approximation Formula

V_NOISE = Noise voltage density, i.e.  nV/sqrt(Hz)
I_NOISE = Noise current density, i.e.  pA/sqrt(Hz)
R_S = Source resistance, i.e. 50 W
k = Boltzmann's Constant = 1.380 662 x 10^-23 J/K
T = Ambient temperature in Kelvin

 

Complete Non-Inverting Opamp Noise Figure Formula (for R_S = R_T)

 

Complete Inverting Opamp Noise Figure Formula (for R_S = R_T)

E_NI = Input noise voltage

I_BN = Non-inverting input noise current

I_BI = Inverting noise current

R_S = Source resistor

R_T = Termination resistor

R_F = Feedback resistor

R_G = Gain setting resistor

The greatest influence on overall noise perfomance is the source impedance associated with the signal. For best noise performance, limit the bandwidth with passive or low noise active filters.