Kevin Gilmore Class A Dynamic Headphone Amplifier

Schematic

In 2001, Kevin Gilmore designed a solid state amplifier for dynamic headphones (i.e., not electrostatic). His no-compromise approach resulted in a number of nice features, such as no capacitors in the signal path and a servo loop that operates outside the audio feedback loop. Further details on the design can be found here.

Several suggestions for improvements have been made. I opted to go with some of those made by YoungKyoo Kim. His suggestions can be found here (in Korean). The schematic of my design is shown below.

schematic

PCB artwork

I've designed a moderately compact PCB, shown below. It is half the size of a eurocard, 10 by 8 cm.

PCB

Remarks

This amplifier requires a suitable power supply. Just any won't do. It has poor common mode rejection, so you have to make sure the voltage between the positive rail and the ground is as close to the voltage between the ground and the negative rail as you can get it. Have a look at the description of my power supply to see how I solved it.

The servo does not affect the signal directly, but rather influences the current sources of the first stage. It works because the leds are not perfect components. The end result is that it can't adjust a whole lot, so the DC offset must be close to zero to start with. This supposedly requires close matching of leds, resistors and transistors. Some people have suggested not matching resistors and transistors, but just selecting leds that happen to compensate the rest. This would require testing a number of leds to find a pair, but hey, leds are cheap. I was going to do this, but the first two leds I tried gave DC offsets of less than one milivolt. Could have been luck, but when I later replaced one of the four leds, the DC offset remained below 1 mV.

The sound of this amp is absolutely stunning. It gives enormous power in the bass. The clarity is amazing.

Mechanical work

The whole thing, together with its power supply, is fitted in an aluminum enclosure made by Isel.

Update (fall 2006)

I have split the amplifier and the power supply into seperate enclosures. This minimizes the 50 Hz pickup to levels below my ability to hear even at maximum volume. The power supply has remote sense capabilities, so that the rail voltages are still on target and stable. I used DIN connectors with a litz braid.