"strong emotional event" indeed !
I was twentysomething, and as you both observed, I've never done it again, either.

Once, the soldering iron I was using rolled off the bench and I caught it in midair ....

The best thing about this is that it uses common substances found around the house, no need for a shopping trip or special order.

Thanks for the useful information ! 

Answers to questions above, not necessarily in order ....

 - Not a Peltier device, just a resistance heating element, 7.2 ohms, made from fiberglass and silicone rubber.  The board on the left is an old linear power supply , modified to provide 6.7 volts DC, up to 3 amps.

 - a few data points on power consumption;
   - during bench testing back in January, Arduino + heater drew 1.9 amps at 6.6 volts DC
   - Arduino + heater  + power supply measured yesterday with a P3 Kill-a-watt, 4 watts idle, 12 watts full load .  Version 2 will probably include a data logger.   

Oh, and so far no cats (or anything else) have been electrocuted !
Chewing on the line cord is about the only option for that, the cheap plastic enclosure has held up pretty well overall.

 

Thanks, everyone, for the kind words !

I thought of an automatic water-feed, but the freezing conditions would require a heated line and that just added too much complication for a project I wanted to deploy quickly.  Definitely in the works for next winter, though.  Same for the motion detector camera.

I have seen squirrels, a cat, and birds drinking in daylight hours ... heard sounds in the night, and seen lots of various tracks (raccoon and skunk for sure, maybe opossum)  when there is fresh snow in the morning.  Lots of takers too, the bowl typically needs refilling 2 or 3 times in 24 hours.

Can't resist posting this.
On successive days I saw a squirrel and a feral cat getting water by licking ice in our backyard.
To provide a source of drinking water for all the assorted wildlife living out there or passing through, I built this heated water bowl.



A thin-film heater that I got as a trade show sample years ago and a TMP 37 temperature sensor are attached to the underside of a small, cheap stainless steel pet food bowl using Kapton tape.



A Duemilanove reads the sensor on A0, controls the 6.7 volts to the heater using an FET driven by D3, really all I changed in Brett Beauregard's excellent PID library sample code was the math to change from TMP 36 to TMP 37.



Water consumption and tracks found in snow indicate that a source of liquid water in midwinter is becoming very popular.

Geiger - Muller tube will detect x-rays.  Many designs for the low-current high voltage power supply can be found online.  Old Soviet G-M tubes can  be gotten at reasonable prices.

PIN photodiodes do not require high voltages but the extremely low current signal output makes them subject to noise pickup unless well shielded.

Scintillation crystal + photomultiplier detectors are the most linear, quite sensitive, and can be very efficient, but are more costly and also require high voltage.

Overall I would recommend the G-M tube approach.

Not sure if this will be helpful in your search for code, but the vt-100 terminal was, itself, an emulation of the ASR-33 Teletype.

and the Arduino is powered by a separate supply (preferably transformer isolated),

So you would need a power supply that is rated for 4KV isolation. That is way above the normal isolation rating of power supplies, even ones using transformers.

A quick look through a DigiKey catalog that happened to be handy shows that many modern mains transformers made to VDE or other EU standards are rated between 3.75 - 5 kV isolation.

you could use the ARef input and analogReference() function to scale the Arduino input to 25 or 30 millivolts full scale

The minimum recommended reference is around 1 volt. Below that Arduino is too noisy to produce reliable results.
 
Note also that any attachment to a real thermocouple must use an "ice-point" reference compensation or else you will never know whether you are measuring the temperature at the thermocouple, or the temperature at the junction point.  That is one of the downsides of using real thermocouples.

Thanks, wasn't sure about the noise floor, thats why I inserted a caveat.

Good point about the ice point, too ... most OEM instrumentation fakes this automagically.

A standard type K thermocouple will work well above the temperature range you are talking about.
At 900F the output will only be around 20 millivolts, so you might want to use an operational amplifier to expand this to about 4 volts.  Alternatively, you could use the ARef input and analogReference() function to scale the Arduino input to 25 or 30 millivolts full scale.  Not sure offhand if one way would be better than the other.  10 bits of A/D would give you +/- about 2 degrees resolution, not counting the limits of error of type K.
A bare-bead thermocouple made of fine wire will give the quickest response time.  Might be hard to seal at high pressures.
Since you are working in a combustion environment (oxidizing? reducing?), this can affect your choice of materials .... type N might be preferable for a reactive environment.  Sheathing can improve resistance greatly but at the expense of response time.
www.omega.com has extensive reference materials on temperature measurement which may be useful.


Another thought:  what is the thermal coefficient of resistance of your ignition coil material ?
If it is high enough, you might be able to measure temperature directly by measuring voltage drop at constant current like a standard 4-wire resistance measurement (or current at constant voltage if that is more convenient).

The real issue with measurements at high voltage is one of isolation.
If your HV supply is constructed without using a chassis or other common ground, and the Arduino is powered by a separate supply (preferably transformer isolated), all that the Arduino will ever see is the voltage drop across the current shunt resistance.  This resistor should be 200 ohms to provide about 4 volts drop at 20 milliamps, so as to use most of the 0 to 5 volt measurement range for the area of interest, with an allowance for overcurrent.

Note for safety purposes that the Arduino analog input pin will be at 2 kv, but only with reference to the other side of the HV supply.  Avoid inadvertently providing a path for current flow.

Very Cool !
like the poster above, a web server may be in a future project, and i'm sure your info will be helpful then.
Thanks !

A nice piece of work !
Before the PC or Mac as we know them existed, I was writing data acquisition programs in figForth for the RCA 1802 processor and I was impressed with the compact power of the language.
Also a user of HP RPN calculators, and there are similarities.
I look forward to trying this out.

I usually stop when the rosin smoke starts making my eyes burn.
6 to 10 hours / day is reasonable.