As I described there, the resistors on the Thing-O-Matic MK5 Extruder Thermal Core operate in impossible conditions. To summarize, each resistor is rated to dissipate 10 W at 25 °C, but is actually dissipating nearly 30 W at well over 225 °C. Ouch!
I wanted to figure out just what was going on inside the Extruder Head, which means some instrumentation was in order, which meant I had to figure out how to attach a set of thermocouples to the Core. This picture shows one approach: epoxy a set of small brass tubes to various parts of the MK5 Extruder Head.
JB Industro-Weld epoxy is rated to 500 °F = 260 °C, which is barely adequate for the job at hand.
The general idea is that each tube provides an isothermal mount for a thermocouple bead, without the inconvenience of drilling holes in various metal bits and messing with high-temperature thermal compound. I am assuming that putting the beads inside the tubes, heating the Core, then waiting for the temperature to stabilize will produce meaningful results.
I have a motley assortment of meters that allegedly read temperature from Type K thermocouple beads. The business end of those thermocouples looks like this:
The twisted one in the middle has a completely non-standard red-black insulation color code, but as long as the meter it came with is happy, I’m happy. The two on the right have industrial-strength wires, as befits the fact that they plug into a Fluke 52 dual-thermocouple meter; them, I trust.
I skinned down the insulation a bit so they’d all reach into the middle of the tubes and filed down the bead on the right just a smidge.
It shouldn’t come as any surprise that each of the five thermocouples reported a different number for the ambient temperature, which meant a calibration run was in order.
Up next: an isothermal block.
The Smell of Molten Projects in the Morning 
It looks like you’re soldering your thermocouples. What do you use? I use the very primitive spotwelder to weld mine (and end up blowing off a lot of tips.)
A project you might enjoy looking at, although it’s probably more work than the reward, is the opensource one wire file system at http://owfs.org/: they implement a dallas one-wire network of sensors on top of FUSE, so each sensor acts as a standard linux hierarchy of directories. One of the dallas chips intended for controlling charging of lithium batteries has an A/D sensitive enough to read directly from a thermocouple, and the OWFS framework has a calibration routine so you can read from the chip/typek (or j or whatever) directory and get an already-linearized temp reading. The chips only cost like $6, and you can wire up 200-some using phone cord, so if you ever need an inexpensive way to thermally characterize a complicated thing it’s a pretty good system.
soldering your thermocouples
Oddly, those are all straight from the source: packed with the multimeters! I have no idea if they’re soldered or welded, but I betcha the Flukes are welded the right way.
Now that I have a lifetime supply of tungsten electrodes, I want to try welding little bitty stuff like that. Won’t get around to it for a while, but … revamping the control code for that hulking resistance welder ought to get me most of the way to the goal.
one-wire network
Way back in the day, I was crushed to discover they really needed two wires…
Way back in the day, I was crushed to discover they really needed two wires…
Well, ya could use earth return… When we went to Iceland, they still had outlying areas where they ran a single power wire and relied on earth ground. They also had enormous autotransformers in their houses, so they could adjust the in-house voltage manually to account for voltage drop across the entire supply circuit. I think if I lived there I’d invest in UPS equipment for every single thing I owned.
I’d invest in UPS equipment for every single thing I owned
And rubber floor mats…