Touchpads are too "noisy" / Writing input doesn't work

[mfoster]

The touchpad assembly on the B1 systems was built incorrectly by Alps, the manufacturer of the dual-mode touchpad. An internal ground layer was added in contrast to OLPC's guidance, causing insufficient capacitive sensing signal amplitude. As a result, the touchpad is far more noise sensitive than it should be.

The real fix is to remove the internal ground plane layer, which dramatically increases the available signal sensing voltage. In the interim, the decision was made to disconnect the resistive sensor and the ground plane in the B1 build, which is an effective workaround. However, the proper fix in the B2 systems will further improve touchpad accuracy and tracking, and will allow proper operation of both the capacitive and resistive sensors.

One consequence of this change is that elimination of the ground layer will have the positive outcome of reducing the force required for written input. As documented in Trac #474, the ESD ramifications of removing this ground plane will be avoided by introducing a new metal plate that underlies the touchpad sensor assembly.

[mfoster]

The second iteration of the Alps touchpad looks really, really good. Though driver development hasn't yet caught up, the hardware was tested using diagnostics code written by Quanta, and the pads seem rock-solid, with no noise and no jitter. It's a vast improvement, and I think folks are going to like it!

[jg]

additionally, a mylar sheet was required to make it work (along with a ec work around to get up behavior to work properly).

This should get release noted (the mylar improves the B1 touchpads as well), along with how to recalibrate the touchpad, and that mylar sheets can be had to improve B1 units (all B2 units have the mylar).

Richard, close this one when we're sure the EC workaround for "UP" being reported reliably is implemented.

[cjb]

Richard, can this be closed?