The output op-amp is the chip which provides amplification of the audio signal to the headphones. With an unbuffered amplifier design such as the UHA-4 or UHA-6S.MKII, the output op-amps drive the headphones directly. With some of my amplifiers, I offer a "standard" op-amp and a few other choices. I'm often asked how the op-amps differ in audio quality or which op-amp to pair with a certain set of headphones. My short answer: choose the standard op-amp, the AD8610. For a bit more information on how I arrive at this recommendation, read on.
As an amplifier designer, there are a few key electrical traits I look for when selecting op-amps for a portable amplifier. It should have low output noise, it should be able to drive sufficient output power to a wide range of headphones with low distortion, and it should be able to do these things while drawing minimal power from the power supply. The goal is to have a "transparent" amplifier which faithfully reproduces the input signal without adding anything of its own. Because we are talking about a battery-powered amplifier, we also want to minimize the total power consumption.
The Analog Devices AD8610 was originally my standard op-amp for both the UHA-4 and UHA-6S. It excels in all the key electrical traits, and it is well-known in the hi-fi headphone amp community. A few years ago Texas Instruments released the OPA209. It also performs very well, and it even beats out the AD8610 in some measurements. I began using it as my standard op-amp for the UHA-6S.MKII, and it really seems to be gaining some fans.
In addition to the AD8610 and OPA209, I also added a few more popular op-amps to the list, such as the OPA627 and AD797. Both of these parts have large numbers of fans in the hi-fi community. Using these op-amps requires accepting some trade-offs, such as higher power consumption.
While the OPA209 has some great characteristics for portable audio, it turns out it also has a weakness: susceptibility to interference from GSM phones and other mobile devices. Many op-amps have this problem. You've probably heard the result: it's the buzzing or clicking sound that occurs when a mobile device is placed near an amplifier or speakerphone or some other type of amplified audio device. The OPA209 uses a type of transistor called BJT which is particularly susceptible to this type of interference. Op-amps which use JFET transistors, such as the AD8610, are much less susceptible to the interference. The OPA627 has a proprietary input stage called "Difet" which is similar to JFET, so it also has relativity low susceptibility.
Our hearing is ultimately a subjective experience. The measurements may tell us an op-amp performs well electrically, but how can we determine the sound quality? I'm a firm believer that if you can hear it, you can measure it… the problem is we don't always know what to measure.
So what would happen if we ignored the measurements and put together a blind listening test which allowed us to switch instantaneously between all the different op-amps which made the list? You'd find something maybe a bit unexpected: the differences are not that great. In fact, you might have trouble hearing ANY differences at first. In a poorly-designed amplifier you might end up exacerbating a shortcoming of the op-amp, causing an audible degradation, but in a well-designed amplifier the differences tend to disappear. Many others have made this same observation. Even the very expensive OPA627, often described as one of the best and most "musical" op-amps for hi-fi audio - in a proper A/B comparison its sound quality is not obviously better than the others.
The AD8610 is currently the best choice for most customers. It measures very well electrically, it has low susceptibility to interference, and it has a proven record in the world of portable hi-fi.
When might you consider the other options? If you're an experienced listener and have experimented with different op-amps, you may certainly have your favorite. But if you're not sure or just starting out with hi-fi headphone amps, the AD8610 is the way to go.