Saturday, May 21, 2016

Minimum Viable Prod

This weekend I helped my pals Arlen, Sarah, and Sam at the Stupid Shit No One Needs & Terrible Ideas Hackathon. This one I can blame mostly on Arlen: it's a cattle-prod-selfie-stick "hybrid" (the polite term). It takes a selfie of you shocking yourself with the cattle prod.

Disclaimer: this is not even a good hack, it is a bad TERRIBLE AWFUL NO-GOOD HACK. NOT WITH A BARGEPOLE; SEIZURES, DEATH & DISMEMBERMENT; NICHT FÜR GEFINGERPOKEN; HURT THE WHOLE TIME YOU ARE DYING, ETC., USW. It's described here for entertainment purposes only as they say.

We started with a livestock prod Arlen found online. It was powered with 2 C batteries. The case was solvent-welded shut which made it reasonably waterproof but precluded an easy teardown. So we left it intact. The HV circuit inside is almost certainly a typical transformer inverter type: you could hear the oscillator frequency (about 2kHz) when the trigger button is pushed. It may have had voltage doubling stages: in any case it put out higher voltage than we could measure with any equipment we despised enough to test it with.

One design criterion is that we wanted a no-free-rider operation, that is: no shocky, no selfie. It wasn't enough to trigger the selfie after pushing the prod button, we needed a circuit that would detect the actual shock through the body before triggering the shutter. This turned out to be a little tricky. After some experimentation involving a scope, a six-pack, a sausage test load, we came up with the hack described herein. (We used a sausage as a dummy load so we wouldn't have to shock ourselves quite so often when testing -- we're not that idiotic...)

Here's the schematic, scribbled, appropriately, on the back of the shopping bag in which someone brought hackathon snacks.

The principle of operation is kind of as follows: Input from one probe of the prod is capacitively coupled to the gate of MOSFET through a .1uf capacitor. We used some high-voltage ones we found at Noisebridge. The 10K input resistor (later changed to 1K) reduces the input impedance so the circuit does not trigger unless ground is referenced through the skin via the ground electrode. It also bleeds the gate capacitance of the FET. The Zener diode is ostensibly there to protect the gate from high voltages: it may or may not have been effective. We used an IRF521 FET mostly because it was the first thing I could find in my junkpile; I had a few left over from some motor controllers I had built so I knew they would stand up to abuse! It also had a low Vth of 2V or so which may have helped.

On the output side we need a short negative pulse to trigger the selfie. We used a selfie stick with a trigger circuit that used a momentary switch to switch on a (PNP?) transistor. (Curiously just shorting the transistor did not work, but shorting the base like the switch did work. Didn't have the time to investigate...) Looking at the output waveform on a scope, the FET switched on and off at every HV oscillation. This confused the shutter trigger, so we put in an RC lowpass to make a longer trigger pulse. An obvious improvement would be a 555 or something here, possibly with a trigger delay, though there was enough latency through the cellphone that it wasn't necessary. In the end it turned out to work only with an iPhone, though early Android experiments had too much trigger latency to be completely useful.

Here's a close up of the finished hack, with a circuit board neatly soldered by Sarah. The selfie stick and prod are ziptied roughly coaxially. Copper tape around the handle serves as a skin reference for the trigger circuit.

In practice the victim participant puts the prod to some portion of their anatomy, presses the trigger to administer the shock, which automatically triggers the selfie.



RSSicon.png  RSS feed