Touch sensor questions

[Wilba]

Directed to TK:

As per our discussion in another thread, I'm planning to make an array of touch sensors. The DIN module design is quite clever - I wish I thought of that when I was prototyping!

What I'd really like to know is how they work, electronically and in software.

I am assuming pin 14 goes high just prior to sampling the DIN inputs, thus charging any fingers touching the contacts, then it goes low and the DIN inputs get sampled.

How does the parameter set by MIOS_SRIO_TS_SensitivitySet control sensitivity? If my assumptions above are correct, I'd also assume it controlled how long pin 14 goes high prior to sampling, or control a delay between it going low and the DIN inputs being sampled.

I'm asking all this because I plan to have a long cable between the touch sensor contact and the DIN inputs and I'm wondering if that will affect things... Also wondering how the 47k resistor value was determined - it pulls DIN inputs low, but can sensitivity be improved by increasing current to the fingertip and getting pin 14 to drive a current sourcing transitor rather than source the whole current from the PIC?

I'll have my core module/DIN module finished soon and I can do some experiments of my own... but I still need some idea what's going on first.

Regards,

Wilba

[TK.]

I made some snapshots from my scope for a better understanding:

This snapshot shows the trigger signal (CORE:J14) in relation

to the DIN input signal of the 74HC165 when the touch sensor

is not pressed. Due to the low capacitance the voltage level raises

fast.

This shows the DIN input signal when the touch sensor is pressed.

Your finger leads to a higher capacitance, therefore the voltage

level doesn't increase so fast anymore.

The MIOS_SRIO_TS_Sensitivity value variys the pulse width in 200 nS

steps.   A higher value will lead to a a higher sensitivity, since it increases

the charging time.

Now some words to the sample point. The RC signal of the

shift register in relation to the touch sensor trigger: the

DIN value will be loaded with the rising edge of RC

RC in relation to the DIN signal --- MIOS will sample a logic-1

when the voltage level of the DIN signal was higher than

ca. 3.5V at the rising edge of RC

The same when the touch sensor is pressed. The voltage

is lower than 3.5V, therefore MIOS will scan a logic-0

To your questions:

I'm asking all this because I plan to have a long cable between the touch sensor contact and the DIN inputs and I'm wondering if that will affect things...

Yes, it will increase the danger of crosstalks. The DIN registers should be as near by the fingertips as possible (this means: shorter than 1m (meter))

Also wondering how the 47k resistor value was determined

Keyword: heuristic analysis ;-)

I used a trimmer to determine the best value. Higher values will lead to (the damned) crosstalk effects, lower values lead to a poor sensitivity.

It's the same for the sensitivity value (-> pulse width) offered by MIOS.

but can sensitivity be improved by increasing current to the fingertip and getting pin 14 to drive a current sourcing transitor rather than source the whole current from the PIC?

I've no experiences with such modifications (I'm no analog expert). I guess that it won't make a difference, because the PIC already contains strong output drivers (up to 20 mA).

If you run into problems with your very special touch sensors (I'm using this simple circuit for motorfaders), you could add Schmitt Triggers to improve the logic level "detection". They could possibly also help to eliminate the crosstalk problem (if it occurs...)

Best Regards, Thorsten.

P.S.: these pictures have been made with the touch sensors of my MIDIbox LC - 16 shift registers (=128 inputs) are connected in a chain, the touch sensors are the last in the chain, this explains the transients on the high-frequency RC signal

[Wilba]

TK, you continue to blow me away with detailed responses... thanks a lot!

Best regards,

Wilba

[Trapstate]

Wilba, maybe this is interesting for you:

http://www.qprox.com/downloads/datasheets/qt113_102.pdf

These small IC's measure the capacitance between a small conductive plate and your hand, and thus also the capacitance to ground. Sensitivity is depends on configurantion, size of the metal plate and the overlap of your hand. See this picture to get an idea, the small boards on the right are the sensors:

www.putzi.nl/pics/0070.jpg

Output is TTL (well, more or less), so directy usable in digital logic. They also have a matrix module  (only smd afaik), maybe also worth to check out.  

Good luck,

trapstate

[Steven_C]

TK, you continue to blow me away with detailed responses...

me too! I think he does it while he's sleeping! ;) Thats the only way he could fit it into his life! ;) I wonder what he dreams of while he's doing all this!!

[Wilba]

Trapstate: I've checked out that datasheet before... but initially thought it would blow out my budget... and also got the impression they would trigger even if your finger was near the contact (not what I want). Maybe I'll check it out in more detail...

Can you explain that picture? What is it?

[Trapstate]

If I remember correctly, they are not *that* expensive, and the matrix version is surely not.  How many sensors do you want to make?

As for the sensitivity: yes you can trigger them without contact (that's what they're designed for). With a sheet of copper foil I managed to trigger them from 15-20 cm. But if you reduce the surface and configure the ic's differently, the distance gets smaller.

The picture is the construction of a lamp, only 3 high power leds burn at once, and the height can be adjusted by moving the hand over the back. In this case the tiny metal plates were covered with non-conductive paint, but still the sensors trigger when moving the hand over the back of the lamp. See this picture to get an idea of the lamp:

http://www.putzi.nl/pics/3leds_active.jpg