Devo12 Touchscreen issue

I have been playing with the nightlies and it seems to me the touchscreen issue as reported in #383 is getting worse.

It sometimes happens I touch an icon, and I end up somewhere totally different and it even changed a setting. This means I can hardly trust my settings after this phenomena happened and have to go through all the settings to check if nothing has been changed.

Is there anything I can do to help to solve this problem?

As mentioned before, I have never experienced this on the standard Devention software.

That's pretty much what we did called "de-bouncing". You read multiple times until you get same or similar results. Once you get stable results, pass the value(s) on. I think that will solve the issue.

Pattaya01 wrote: That's pretty much what we did called "de-bouncing". You read multiple times until you get same or similar results. Once you get stable results, pass the value(s) on. I think that will solve the issue.

We already had the noise issue on the Devo12s ADC inputs, hopefully it is the same random noise affecting the resistive grid. A simple moving average window or low pass filter with say x10 oversampling could give a stable output.

Alternatively this Hampel filter technique we used successfully in our real time data streams to remove ad-hoc and unpredictable noise pulses/outliers.

www.edn.com/design/systems-design/434043...near-digital-filters

I do confirm there is a touchscreen issue.

PhracturedBlue wrote: note that we're actually talking about digital signals here, so you don't really want a moving average. It is likely unrelated to the ADC issue we had. It doeasn't mean we can't use some of the solutions used for noise filtering, but the problem isn't exactly the same.

Hi PB

My assumption (I don't have a 12S to test) is that the SMPS noise is affecting the resistive touch screen controller on the analogue XY input side, unless it is a plain bug. This then results in noisy (wandering) XY coordinate readings on the digital side, read by the processor. My suggestion is to low pass filter an over sampled set (or stream) of XY readings on the digital side to get the best approximation of the actual touch point.

Hope this makes sense.

I do have a 12S and I will be more than happy to help. I was the first one to report this issue and if you want, you can write a piece of testing software that I can load that creates a file or something that I can send you.

I still think your first suggestion about reading multiple times and ignoring the "way off" values will do the trick. I noticed the original Devention SW only executes a command at releasing the button, not at or during touching. So I guess they use the time while you press the button to read again and again.

Hello everyone. Did C960b8E address the Devo 12 touchscreen issue?

I am guilty of being absent and not following all comments however, I did download the nightly C960b8E a few days ago and I have noticed that the ghosting button presses are no longer prevalent, as before.

with a stylus I have noticed no ghost presses.