Darlington Touch circuit - Is this a simple circuit or dangerous circuit?

Hi everyone,

I'm not very at home with circuit design. Currently I'm trying to build some fast touch sensors (capactives are to slow). I came across the following circuit:

I replaced the PIC with an Arduino (Nano, pin 12) and used a 14kOhm for R1, a 5MOhm for R2 and 2x a D331 NPN's for the Darlington. However I found that I really needed to press the touch-pads (aka jumper cables) into my fingers to make the arduino sense a current on pin 12. This wasn't very usefull. I couldn't find the problem (also tried different resistors).

So I 'redesigned' (stumbled upon it when omitting R2 and simplified) the circuit like in the attachement. Now it's very responsive and I only need to touch it lightly. Now pin12 is high, unless I touch the base of the Darlington. (So it behaves in the opposite way)

However it worries me nobody else uses this circuit. Is it safe?

thx for your replies, Nard

It is not very safe and not reliable.
When you have static electricity, it could harm the Arduino.

The Arduino capacitive sensor library is fun, and it is nice that it is possible. However, for a fast response you need special chips for capacitive sensing.

Add 10k between the Arduino pin and the transistor/darlington collector and its pretty
robust - you might damage the darlington, but not the Arduino.

If you have a non-isolated PSU without an earth (unlikely), then you have to
worry about protecting the person from the circuit - add 100k inline with each
probe wire is going to help there (and protect the darlington from static).

Thx! (and added resistor)

Is a 9V battery a non-isolated PSU?

Still wondering why the simple 1-pad circuit is much more sensitive than the 2-pad circuit. Any ideas?
Are the risks of damaging the Darlington a concern for both circuits, or only the 1-pad circuit?

NardJ:
Is a 9V battery a non-isolated PSU?

A battery is very much isolated, in general.

NardJ:
Still wondering why the simple 1-pad circuit is much more sensitive than the 2-pad circuit. Any ideas?

Because you have used a Darlington pair - you have massively increased the gain. It is not picking up your conductance between the pads (and having an external pad connected to 5V is a very bad idea - the circuit should have been inverted with a PNP transistor and the pad to ground), you are actually picking up the power mains radiation within your home by your body capacitance when you touch it - you will find that it is not a constant "touch" but an AC signal at the mains frequency - and it may not work anywhere near as well outside.

NardJ:
Are the risks of damaging the Darlington a concern for both circuits, or only the 1-pad circuit?

Bipolar transistors are pretty robust, and a Darlington is merely two connected together, no more or less robust than one. A FET (or a 74HC series gate such as a 74HC04) would be equally (or even more) sensitive but would need protective diodes.

Their achilles heel is the base-emitter reverse breakdown, usually about -5V, but
it takes a certain minimum current to do damage even so. Its probably wise to limit
the base current with a resistor, so long as that resistor is small compared to the
resistance you are trying to sense.

If you build an external capacitance sensor, you could make it respond more quickly. Just how quickly do you need it to react?

You may simply need a sensitive switch like a microswitch.

The problem with using a transistor as you are doing relies either on:

  1. Skin conductivity which will vary widely and can be affected by contaminants causing false positives.
  2. Electrical noise pickup that relies on using your body as an antenna, the level is lower with an isolated circuit and nonexistent if you are far from sources of electrical noise.

Define fast enough, and what kind of capacitive sensor have you used?

Thx all for your advice!

@polymorph: I'm building a midi controller (think keyboard) for which the keypresses have to register fast enough (1/16 notes, so I guess the keypress has to be registered within 1/64 second or 16 msecs ) My current experiments with capacitive sensors show me they are slow (Arduino Playground - CapacitiveSensor). I assumed this is not due to the circuit, is this incorrect?

Most capacitive sensors rely on measuring the increased capacitance caused by your finger being near or in contact with a touch surface.

There is another kind where your finger works to attenuate an AC signal in order to detect a touch. In that case, having it followed by a missing pulse detector makes it as fast as the oscillator driving the touch sensor.

For instance, a 100kHz square wave goes into a 47pF capacitor, then to another 47pF in series, then to a missing pulse detector.

The junction between the 47pF capacitors is the touch point. When you touch it, your body capacitance becomes a voltage divider with the 47pF capacitor, dropping the voltage below the detection threshold of the missing pulse detector.

This can also be done with resistors instead of capacitors. In that case, the resistor and your body capacitance act as a lowpass RC filter that drops the 100kHz signal below the detection threshold of the missing pulse detector.

When relying on a capacitive sensor, it is a very good idea to use a metal case, or at least have a large metal shield inside the box. Even better if it uses a 3 prong plug and connects the circuit ground to AC ground through a 0.1uF capacitor in parallel with a 1M resistor.

Here is another type that relies on the difference in delay caused by the extra capacitance caused by a body touching the sensing contact:

http://electronicdesign.com/analog/capacitive-proximity-sensor-provides-accuracy-and-speed