Read the warnings at the bottom!
Bolting two mounting blocks to the MK5 Thermal Core produces a rather chubby-looking hot end, but it’s actually not much bigger than the original Core+Resistor version.
All that steel makes for a longer thermal time constant, which (as it turns out) may not be such a Bad Thing in an extruder.
I applied some of the same ceramic-wool oil burner combustion chamber lining insulation that I used before. The stuff is hygroscopic and goes on as a moist sheet, then bakes to a solid shell after a few hours at high temperature. I used a half-thickness layer all around and snugged it in place with Kapton tape, which gives enough clearance on the bottom to avoid snagging the print or the nozzle wiping brush/pad.
In principle, the cartridge heater elements are embedded in solid ceramic insulation and cannot short against the shell, but you still need a static drain line on the Extruder head to prevent charge buildup from the filament. That’s the heavy red wire heading off to the upper left.
The coil of blue wire in the middle left comes from the cartridge heaters: it’s actually long enough to snake up and down and all around to the Extruder Controller, but I already had a wire to the Z stage and an LED that monitors power to the heaters.
With everything in place, I fired it up and recorded the temperature rise…
In round numbers:
A pair of 25 W elements heats the Core from 14 °C to 225 °C in 15 minutes, then cycles off-and-on with < 2 minute period. I don’t have a good number for the duty cycle yet.
The Kapton tape around the insulation seems to run at 150+ °C, but that’s not a good number. I must add some probes around the insulation after it hardens.
With P=100, I=0, D=0 to get bang-bang control (more on this later), the temperature stabilizes just fine. The heater turns on at -1 °C from the setpoint and turns off exactly at the setpoint, with the temperature varying ±2 °C around the setpoint.
The insulated Core heats at an average 20 °C/min (80 °C from 1 to 5 minutes), about 4.5 °C/minute around 200 °C and cools at 5.6 °C/minute from 150 °C. Those numbers can go into the appropriate Skeinforge slots, with the usual caveats on reliability.
All the numbers have rubbery tolerances, because the ceramic insulation sweats water as it heats and that certainly affects the temperature rise. The stuff goes on flexy and hardens like a rock after the water departs; I left it steaming at 120 °C for a few hours after making those measurements.
In comparison with 36 W from a pair of 2 Ω resistors in series: those heated more slowly and ran at 50-75% duty cycle. The new setup has more thermal mass, 40% more power, and thinner insulation, so it’s something of a wash. I expect the duty cycle to settle around 50% when all is said and done.
Important Warning!
Before you deploy cartridge heaters “for real”, remember that this is a test lashup, not a production system.
With the stock MK5 aluminum-case power resistors, you could be fairly certain they would burn out before melting the extruder support arches into slag or igniting a fire. Verily: resistor failure is why we’re here, eh?
In contrast, cartridge heaters will happily run at white heat, a lethal situation inside a plywood & plastic box. They will not burn out before causing further damage.
Guesstimating that the mounting blocks triple the 11-minute time constant for the resistor-heated Thermal Core, figure a 30-minute time constant. The temperature rises 58 °C in the first 3 minutes, so the steady-state temperature would be around 600 °C if nothing changed. I expect the actual temperature to be somewhat lower, but even 500 °C = 930 °F seems risky to me: it’s up in the red heat range..
A firmware error, a random glitch, a failed-short MOSFET switch, a stuck relay, or any random problem with a TOM that results in a stuck-on cartridge heater will cause a fire.
You must install a thermal cutoff that:
- Does not depend on firmware or the existing thermocouple
- Positively disables both the heater and the Extruder motor
- Requires a manual reset after a fault
- Indicates the fault condition
A simple thermal fuse gets you the first three points, although you need one that can handle 5 amperes and is mounted in a known-good spot so it will cut out before the acrylic slumps. Adding an LED indicator across the fuse gets you point 4.
You must also turn off the Extruder Motor, because trying to extrude solid plastic won’t end well. Some of the hyperthyroid extruder designs will likely break something before they rip a slot in the filament and a simple thermal fuse won’t prevent that. It’s a step in the right direction, though.
To repeat: the thermal cutoff must not depend on software. All of your instincts to piggyback this on the existing firmware, add a PIC to measure the temperature, or trip a solid-state relay from the PC are wrong. You must assume that any event capable of glitching the TOM will also glitch your code.
The only absolutely certain way to shut off the Extruder motor is to kill the power. Yanking the Power Enable line (from the ATX supply) high should do that; this will require a mod to the ATX connector at the Motherboard to insert a mechanical relay. Killing the power also shuts down the Extruder motor, which may justify doing it that way.
Musings:
I think a thermal switch and DPDT relay can separate the sensing and current problems: relay held on until the thermal switch opens, then it’s locked out. That will require a push-to-heat button, which isn’t terribly bad in the overall scheme of things. The TOM desperately needs more indicators anyway.
Putting a thermal cutoff above the extruder, against the inside of the acrylic base under the filament frame, seems reasonable, but really, really awkward.
Put it against the insulation outside the Thermal Core? I’m using much thicker insulation than the stock ceramic tape, so my measurements aren’t relevant for stock MK5 heads, but it’s certainly a promising location.
Monitor the Thermal Riser tube temperature at the heatsink? The numbers suggest there’s a 5 °C/W thermal coefficient between the heatsink and the Core, but better measurements are certainly in order. A quick-and-dirty test says the heatsink exceeds 90 °C with the Core at 230 °C; maybe that’s too hot for acrylic in the first place.
Tucking a switch inside the Core insulation would be much better, but you need one that operates reliably at 250 °C and trips at, say, 300 °C.
Bottom line:
Don’t install cartridge heaters without a thermal cutout: your insurance agent should not be given an opportunity to die laughing.
The Smell of Molten Projects in the Morning 
I don’t know much about thermal fuses, but it seems to me that solders melt pretty consistently around a useful temperature for this purpose.
BTW, when you get this worked out do you think you might make this available as an upgrade kit? I think I may get a ToM to mess about with, soon, and would like to get this particular bugaboo fixed from the outset (since it seems all but certain to occur with the existing design).
On a closing note, I have been reading your words for probably 20 years and have always enjoyed doing so. Thank you so very much for producing this blog.
Indeed! As long as there’s no appreciable current, a length of solder would slump at pretty nearly an exact temperature every time. The trick would be finding the proper alloy, then conjuring up electrical connections: the alloy formed at the terminals would (by the conservation of perversity) certainly melt at a lower temperature!
The table there says 60/40 solder melts at 186 C, 4% silver solder at 221, and pure tin at 232 C. You’d need something around 250 C, I think, to work while snuggled up against the Thermal Core under the insulation blanket, which is where it really should be for best results.
The thought’s crossed my mind, but I think it’d be better all around if somebody with an existing pick-and-ship retail operation did the deed: Makerbot, for example!
A long time ago I was a tiny electronics parts supplier and there’s a surprising amount of work behind each package…
Zowie! Time really does pass quickly when I’m having fun… thanks!
I am amazed, your mod’s to the thing o matic are great, but i do not have the machinery to do it myself, also not skills , or the time ;-). Small question would a 25W 4,7 ohm (these are a bit bigger but i could work around that) instead of the 10W 5 ohm resistor ease the stress a bit (aka live a lite longer than the 10W version). Anyway keep up the good work !!!!
Unfortunately, the MK5’s required power dissipation far exceeds the ratings for any size of the aluminum power resistors, which must dissipate no more than 10% of their nominal wattage at 250 C. The 25 W resistors top out at 2.5 W: too small by almost an order of magnitude.
While reducing the power does reduce the stress, yesterday’s autopsy on those two resistors shows one was deteriorating rapidly with only a factor of 18 overload. The other seemed to be OK, but … it’s only a matter of time.
As nearly as I can tell from the forums, a majority of MK5 heads continue to operate normally for months, even as the resistors deteriorate. If you insulate the thermocouple bead from the Core (and ground the Core), a failing resistor won’t kill the Extruder Controller: you just replace the failed resistor when you notice the extruder takes an unusually long time to heat up. If you’re willing to keep a handful of spare resistors on the shelf, the head can work fine with periodic maintenance.
Believe it or not, I didn’t buy a Thing-O-Matic just to spend the next two months modding the daylights out of it.
Thanks for the good words; it’s been an interesting experience and I hope other folks will benefit from my tinkerings…
How complicated is the center block in your three block sandwich? The thermal TC should be much shorter if you reduce the physical mass. How about making a new center block that has two holes for the heater cores almost but not quite intersecting the center bore?
Can you post the source of the heater cores and their datasheet?
Maybe the heater cores could just be inserted directly into a modified MK5 body:
I’ll try to inline the image:
[Edit: Here’s the image…]
The “center block” is the standard MK5 Themal Core, which I didn’t want to modify before collecting more data. The MBI doc lives on Thingiverse: right there.
The cartridges won’t fit directly into the existing Core, but perhaps I could (firmly!) clamp thinner blocks to the sides and drill offset holes. The end result would be a pair of trenches along the sides, similar to your sketch, with the blocks covering the outside.
Frankly, that may not be worth the effort, because a relatively high thermal mass actually helps stabilize the temperature during extrusion. The control loop really can’t track the start-stop heat load from the filament’s motion, so having plenty of mass is a Good Thing. I manually start the heaters when I begin setting up for a build and by the time the G-Code is ready to go, the Core is at the proper temperature.
I assume the cartridge heaters came from High-Temp Industries. There’s not much in the way of a datasheet on the site, but the dimensions seem reasonably standardized.