Looking for programmer to help with possible T3 digitizer fix

[scotseaton]

Hi, I'm lookin to get in touch with anyone with PalmOS programming experience to help build a fix for the T3 digitizer problem. I think I've got a way that should work but I need some opinions/help from programmers. Please email or respond on this thread if interested.

[Tam Hanna]

hi,
feel free to email me at Tamog@gmx.at

[Cyker]

Couldn't you just post the idea here? Get more coverage then ;)

[scotseaton]

The idea is similar to other ideas discussed on other websites (I think Tam was in on one of these). The way I see it happening is to develop a program that would have you tap on a fixed set of points displayed with a crosshairs, record the x and y error that the digitizer sees on each of those points, and then develop polynomial expressions for the x and y error as a function of x and y position. This can be done utilizing algorithms for matrix multiplication and QR decomposition, which are already available in C++ on many websites. I've done this with MATLAB with other applications with a high degree of success and accuracy, but have no C++ or PalmOS experience. I don't see why this couldn't be done - just don't know how to do it myself!

Next, we'd have to create a "hack" that takes the x and y pen coordinates, inserts them into the polynomials derived in the process above, and returns the x and y error for that particular point. This error could then be applied as a correction, hopefully with the result that the digitizer corrects itself to the actual pen position, or at least close enough. I think anything would be an improvement. Not having any experience with Palm architecture, I don't know if it's possible to "hack" the pen coordinates - perhaps you guys could provide opinions?

I don't know what kind of effect this would have on processor speed - but again, it might be worth it to sacrifice speed for a digitizer that works.

Let me know what you think.