# The accuracy of Force Sensitive Resistors (FSRs)

I am trying to make a large scale resistive touch sensor with 4 FSRs, one for each corner of the rectangular sensor. The pressure that I'm going to be applying will be a moderately pressured finger press. I have heard that FSRs are mostly inaccurate as far as precise pressure goes.

For example, the Interlink 402 says in its datasheet that its accuracy is within 2% of its resistive value, and changes with time at a rate of 5% * log10( days ).

As far as tap sensitivity goes, the FSR seems decent, at 10% change after 10 million taps.

I'm wondering if with a passive load of 2kg, and a dynamic load of 1kg, would the 4 FSRs be sensitive enough to make a decent (512x512 (9-bit accuracy)), reliable resistive touch screen, or are other materials better for the cost? If I were to make this, what would my theoretical accuracy be?

• Link to the data sheet for the FSR would help. – Andy aka Jul 19 '17 at 17:34
• @Andyaka boom. I was editing the question as you posted that comment. – tuskiomi Jul 19 '17 at 17:35
• I think the data sheet provides you all the information you need. – Andy aka Jul 19 '17 at 17:53
• 1kg load sounds like a hammer more than a finger. Note the sensitivity is given for one batch only and tolerance between batches must be compensated or calibrated and a matched lead-lag filter is my suggestion to compensate for these tolerances made to conjugate the fundamental spectrum of the dynamic load. ( fo=Hz toggle rate) ie. gain above DC null > fo = 10dB – Sunnyskyguy EE75 Jul 19 '17 at 19:55
• It looks more like table banging sensor – Sunnyskyguy EE75 Jul 19 '17 at 21:28

simulate this circuit – Schematic created using CircuitLab

300mV deflection at 2/3 FSR rating for load from 2/3 to 1 x FSR rating ?

Static load is probably quasi static, so dynamic touch will be problematic so DC tracking by AC amplification

Some problems with OEM specs at 10k expecting 2.5V at 50% FSR but only seeing 20% on left.

• batch to batch variations may need a pot anyways.

## Conclusion

• possible if static load is true static.

• undefined problems with this approach.

• Interesting. There is likely a better way then, huh? – tuskiomi Jul 19 '17 at 22:05
• First specify all user interface dynamics in 4D, XYZ,t including resolution delta {XYZ,t} – Sunnyskyguy EE75 Jul 19 '17 at 22:27
• Nono, i can't expect you to do any more for me; you've done much more than enough. nor anyone else. I'll go on to do as much as i can until i hit a block – tuskiomi Jul 19 '17 at 22:52
• start with good interface specs not design. – Sunnyskyguy EE75 Jul 20 '17 at 0:27