Dual Class-A Line and Headphone Output Board Construction Information

[bobkatz]

I forgot to post the noise measurements:

Noise wideband, flat, unweighted, 1 Hz to 96 kHz: -60 dBu.
Noise 20-20 kHz, A weighted: -95 dBu

[mediatechnology]

Crosstalk at 1 kHz with a 20 ohm load is not fantastic. I measured about 59 dB. Something about the current flow through a Left/Right channel headphone jack that has a common ground. It's close to 90 dB with no load, so the crosstalk is not due to any design error.

Using 60 dB to round out the numbers suggests a common impedance in the ground line to be about 20 mΩ (0.02Ω) so about as good as you can do with wire.

An interesting factoid about the crosstalk is that due to the drivers sharing a ground wire (or pair) to the 0V is that the potential developed across the common resistance (due to IxR losses) presents itself to the undriven driver out of polarity. Thus, what's presented to the listener is slightly wider as a result of it.

Mono elements produce a common mode voltage in the IxR losses of the ground return which doesn't affect imaging.
It's the difference signal that becomes slightly wider due to the common impedance.
When the ground wire is lifted completely all you'll hear is L-R.

I experimented driving 'phones using MS techniques for low voltage battery-powered gear where the MS to LR transformation was in the drivers.
The objective was greater headroom driving the tip, ring and sleeve with a 3 channel amplifier.

[AnalogJoe]

The MS thing sounds very interesting, did it make significant differences?

[mediatechnology]

It did but I decided to change the weightings of side that were presented to the sleeve versus the tip and ring to further maximize headroom.

The thread is here - there's a lot I've forgotten: viewtopic.php?p=5607