Noise handling algorithm in 8 wire touch screen

[hanhao]

Noise handling algorithm in 8 wire touch screen
here's how a touch screen work http://focus.ti.com/lit/an/slaa298/slaa298.pdfProblem:
when the touch pen is stationary on a point and reading ADC values from a touch screen using the micro controller, the ADC values are never the same
eg, 1st reading might be 500, 2nd reading is 510, 3rd is 505. This causes the the cursor to drift about the touch point, not something nice for the user . caps are already in place. it's a standard schematic so hardware shouldn't be a problem.

Proposed solution:
Now what i did was to create a boundary circle around the point, effectively locking all other bad readings into the circle.

http://img263.imageshack.us/img263/8210/circlejy5.jpg

http://img266.imageshack.us/img266/5989/circlehf0.jpg
but the problem is that
- for slow moving touch pen speed movements, due to this boundary circle, the cursor jumps from 1 circle to another. very ugly and it disallows slow movement.
- the size of the boundary circle has to be expanded when the touch screen wears out because of more noise

does anyone have a better solution?
thanks in advance
- hanhao

 

[NoTime]

Have you considered averaging a FIFO buffer?

 

[hanhao]

Have you considered averaging a FIFO buffer?

yes i did average
averaging does help reduce "jumping mouse", but it still causes the problem

 

[mezarashi]

Noise filtering is a relatively complex subject in my opinion. If I recall, most of the better algorithms are patented or kept secret by the companies and organizations that created them.

My suggestion here is to just try all the basic filtering techniques you have as for any digital signal. I assume you have no chance of improving the performance of the ADC at the circuit level. Is it possible to convert a reasonable sample of your data to the Fourier domain using the DFT and do some analysis of the signal there?

 

[hanhao]

Noise filtering is a relatively complex subject in my opinion. If I recall, most of the better algorithms are patented or kept secret by the companies and organizations that created them.

My suggestion here is to just try all the basic filtering techniques you have as for any digital signal. I assume you have no chance of improving the performance of the ADC at the circuit level. Is it possible to convert a reasonable sample of your data to the Fourier domain using the DFT and do some analysis of the signal there?

the micro controller takes ADC samples from a waveform like this, very highly zoomed in
http://img133.imageshack.us/img133/9835/sampleir1.jpg
what analysis should i do?

 

[mezarashi]

It is difficult to do any sort of filtering other than some sort of averaging filter in the time domain. You should do a frequency domain analysis. Do any of your DSP programs support a Discrete Fourier Transform? A low-pass filter might help.

 

[hanhao]

It is difficult to do any sort of filtering other than some sort of averaging filter in the time domain. You should do a frequency domain analysis. Do any of your DSP programs support a Discrete Fourier Transform? A low-pass filter might help.

although averaging does reduce the problem, it did not give much of an effective result. mouse cursor drifts lesser, but the drifting is still there.

on the micro controller there is no frequency analysis feature, unless i program it myself. if i were to program a frequency analysis algorithm, what would be the algorithm for it? what analysis is needed to be done? how do i obtain a usable result from this analysis?

 

[light_bulb]

make a dithering algorithim?

 

[mezarashi]

Well first of all, scour the internet for a frequency analysis program. There's bound to be something free out there. You'll need the graph to get a good look at where the noise is and what you can filter.

Then you can probably implement your own filters, but again, the internet is available. Search for an implementation of the type of filter you would need. For example if you see from the frequency domain that most of your signal is of low frequency and there's a noise spike in another band, a low-pass filter might be suitable. Googling for: c implementation low pass filter, I get:

http://ccrma.stanford.edu/~jos/filters/Definition_Simplest_Low_Pass.html

P.S. Oh, I'm talking from a results point of view here. If the purpose is to learn, then yeah, it'd be better you do all of this yourself. It will really reinforce your Digital Signals Processing knowledge.

 

[hanhao]

thanks for the ideas, ok i will check these out