[Coco] The New Hires Interface

[Mark McDougall]

On 14/10/2015 4:55 PM, tim lindner wrote:

> Then I tried it with a touch pad. This reminded me how much this device
> sucks. If you press lightly it, it'll generate a position half way from
> the center to where you're touching. Press a little harder and it gets a
> little closed to the true point you pressing on. When you press hard
> enough to have the device truly report the correct position you're in
> danger of damaging the surface. The new sampling routine could not
> overcome the crappiness of this device.

Ah the joys of resistive touch pads/screens!

I've written drivers for a couple of them now. The 1st was on a windows 
platform and the filtering was quite complex; sampling quite quickly 
(sub-10ms?) and using some sort of clustering algorithm (I didn't invent 
the algorithm, just coded it). Worked very well in the end but you 
needed the CPU power to do it.

Then there's the last one I did, on a PIC in Swordfish basic. Different 
kettle of fish altogether, and unfiltered resembled the touch pad you 
describe. Unfortunately the previous contractor didn't understand how 
these work, and used the raw data as-is. Users complained about touching 
a button on the left side of the screen, and the right one registering 
as pressed! Hmm... managed to vastly improve it, by debouncing the touch 
event and using an averaging filter - no processing time to do anything 
more. Bit of lag, but at least no phantom button presses!

Regards,

-- 
|              Mark McDougall              | "Electrical Engineers do it
|  <http://members.iinet.net.au/~msmcdoug> |   with less resistance!"