Help needed

Sorry for the long post. I’m looking to design an overheat alarm. What I have is a Lycoming aircraft engine currently using thermocouples in a wheatstone bridge circuit to measure cylinder head temperature (500F max). The operating environment is forward of the firewall, air cooled, varying anywhere from 0C to 130C (engine block happy temp).
The current “factory” solution is known to be finicky and prone to erroneous readings (bad crimps, and a single gauge connected to 4-position selector switch where contacts can be bad, especially for small signals. And it only monitors one cylinder at a time.
I’d like an independent, all cylinders simultaneous monitoring system to confirm and alarm on overheat conditions (roughly 450F+). And I would prefer simplicity, and not drilling holes in the engine. I’m thinking something like a contactless thermometer. I think a simple analog to avoid the clock, data word, Arduino or similar programming makes most sense given the simplicity. Comparator trigger “Low = normal temp, High = overtemp”.
Which brings me to the crux: I’ve been looking at remote thermopile/thermistor sensors like Digikey P/N 223-1583-ND made by TE. I’m a bit confused by the tech material available on these sensors. Op temp (ambient) typically of 0C to 100C… fine. But I’d think that does NOT equate to remote temps as seen by the thermopile. I need 250-300C there.
Likewise, some come with LWIR bandpass filters (8nm and up) and some with a 4.26nm narrow bandpass. I’m familiar with Wein’s Law, but unsure of which filter to select.
And… while I understand the need to compensate for exact readings, is a compensation for ambient necessary here? It seems the thermopile voltage actually spawns off the “Delta T” rather than absolute, so I’m wondering if I cant simply do a voltage divider with the thermistor to get a V-Ambient, add the V-Thermopile, then use that voltage to drive a comparator circuit?
Or is there a better idea? Am I barking up the wrong tree entirely?
Are my voltages and currents too low after scaling for V-Thermistor per degree to equal V-Thermopile?

Thoughts welcome.


My honest thought is that the non-contact IR approach may involve more moving pieces than a person would really prefer. The app note is pretty chewy, there’s a lot of non-linearities and compensation to be done, and unless a person gets all those floppy bit nailed down, you’re likely as not to end up causing yourself more midair confusion than you’re relieving.

Other ideas: we have a stash of AD8496 available, which is something of an “easy button” for thermocouples; page 15 even describes a fixed-threshold detection function that a person could probably use to drive an indicator LED directly, or nearly so.

Thermistors are available rated up to the temperatures of interest, which by virtue of their absolute readout and stronger signal amplitude are much more convenient from a conditioning standpoint; a few well-chosen resistors and a quad comparator should about do it. They’re a lot more delicate than a thermocouple however, which would need to be taken into account.

Re. mounting of either, I’m no A&P but I’d expect that drilling holes in the engine wouldn’t do any favors for your compression ratio… If you can’t replicate the OEM technique for whatever reason, there are silicones rated for 500°F+ available that might let you tuck something into a suitable crevice.