Very poor stereo separation?

[chipaudette]

In recent testing with a RevE with earpiece shield, I noticed that the isolation between the left and right audio outputs is very poor. When I play a tone in the left ear, I hear a quieter but clear version in the right ear. That shouldn't happen. It should be extremely quiet in the other ear.

[chipaudette]

Possible Work Plan (updated in response to findings on 3/28):

Initial Assessment:

• Write a dedicated test program to make it easy to test for the problem
• Test on multiple different Tympan units to see how bad our problem is
  • Different RevE units, RevD vs RevE, Tympan vs AIC shield
• Strip down the test program to be more confident that it is hardware issue (ie, the AIC configuration or the downstream electrical design) and not a software issue in the Tympan Library
  • I think that I’ve already confirmed that it’s in the AIC configuration / hardware and not in the Tympan Library software.

Second-Level Assessment:

• Test the Tympan’s line-out to see if the problem is there as well as on the headphone out
• Try different AIC settings to see if something can be done to affect the problem
  • Try muting / powering down one of the headphone amps
  • Try forcing a recalibration of the DC-offset / HP_Sense thing

Hardware-Level Hacking:

• Hack a Tympan board to use single-ended outputs (instead of the ground-centered that we’re currently using) to see if that configuration makes the problem go away
• See if BatTact has this problem

[chipaudette]

I just wrote a dedicated test program for testing stereo output (and 4-channel output, if using an AIC Shield or Earpiece Shield). It is here:

https://github.com/Tympan/Tympan_Sandbox/tree/master/OpenHearing/StereoOutput_IsolationTest

[chipaudette]

I tested this sketch on three Tympans: my blue-cased Rev E, my pure white Rev D, and the yellowed Rev D here at GH. I listened via my Sennheiser HD280 headphones (not my earbuds). In these tests, all three Tympans showed excellent stereo separation. There was no hint of whatever problem I was having before.

I tested the previous firmware (via the precompiled HEX) on the blue-cased Rev E. It had excellent stereo separation. It did not recreate whatever problem it was that I was hearing before.

I did not test on my un-cased Rev E with this latest firmware. Maybe that one has a problem? Or maybe it was the earbuds?

Next step: when I get the un-cased Rev E back together, I'll test it again to see if the problem re-appears.