Has anyone placed a thermal switch or thermal fuse on the input power wires for a cheap controller board to shut down if it or a neighbor (the bed probably) gets too hot? I was curious about the Anet A8 and similar printers that have the MKS Base 1.x boards since I just got a 1.6 after my Rambo board died.
The whole point of the thermal feedback into the regulation circuitry makes that unnecessary. If your wires get too hot then something else is going wrong. The main power supply cable will get warm in most cases, but not hot. If any wire gets hot to the touch investigate what is going wrong. Again in most cases the kit supplier gave you a wire that is too small to handle the current, so upgrading to a larger wire solves that.
The thermal fuse/switch would take care of the event of stuff you do not notice and act as a last defense against fires. Hopefully it never trips but if it does then it may save you a house. For less than 2 dollars there is little reason not to add one.
Can the transistor/mosfet/ssr switching the power to the heater fail in an ON state?
If yes, then I don’t know how any 3D Printer can pass any modern safety/compliance test without a thermal fuse inline with the heater.
(I honestly have no idea what the safety rules are to get the little stickers applied to stuff sold in stores, but I did a lot of copier/printer repairs in a previous life and the fusers were all protected by inline thermal fuses.)
@Alex_Wiebe I heard something suggesting that the people whole use a MOSFET from the board to power an external MOSFET open them up for the possibility of the external one getting stuck in the on position.
When some printers have heater cartridges or thermistors that fall out causing run away heating, companies turning off safety features against runaway heating and a simple short or fraid wire being able to heat up ridiculous amounts, I think any additional safety measures that can be taken would be a good thing.
Sorry, dumb question: If the off-board MOSFET can fail closed, why can’t the on-board one fail closed as well?
Also, I’ve had the CPU on my +printrbot board lock up when I turned a cheap LED lamp off (the desk lamp had a metal shroud and I had the lamp pointed directly at the printer - I’m guessing some kind of EMP was generated in the LED as the current fields collapsed and the metal shroud focussed that at the board causing everything to immediately stop moving). The hot end stayed hot - I was right there and shut down the power manually. I have no idea if the hot end was being actively regulated by the now locked up CPU, but I’m assuming it was not and was basically on 100% duty cycle until I cut power.
@Alex_Wiebe I do not understand the failings of MOSFETs myself. (shrugs)
Yes, a fail-on MOSFET state is entirely possible, and very dangerous. We address this very minimally by using highly over-spec’d MOSFETs for those loads. EG a 60A FET for a 12A bed load. (Although there’s de-rating for gate drive voltage to consider.) Not a substitute for a heater fuse, and no, I don’t know how people get certifications for these things. My best guess is that it’s a low voltage exemption… electrical safety standards largely do not seem to have been written for 12-24v high-current heaters, because that’s not a design choice most consumer electronics engineers would willingly make.
Firmware lockups are also capable of sticking heaters on, but the Atmega’s watchdog circuit is supposed to prevent that by resetting upon firmware lockup. Older Marlin builds had the watchdog disabled by default (very irresponsible) so this used to be a big safety issue. These days, we have successfully shamed most firmware devs into making sure the MCU watchdog is operable. There are still some exceptions though, like Arduino IDE firmware for Due — a very specific hack is needed to get the watchdog working, so if you don’t know to do that, you can think the watchdog is enabled but it actually is turned OFF by the compiler.
Ground faults are also highly dangerous because we only control heater power on the ground leg of the circuit. If the heater wiring shorts to frame ground (such as from fretted insulation or a fatigued solder joint breaking off) the heater will run away and there is nothing the firmware can do unless you’re running an ATX PSU with the PSU_ON line wired up to the controller, or some similar solution.
@NathanielStenzel , curious why you would add the thermal switch versus an inline fuse for the hot end/bed? Thomas’ recent vid on printer safety had what I thought was a good idea on inserting the fuse inline between the PS and controller board. Seems like that would catch a runaway situation as well…either way, I’m open to other ideas cause I agree and want to add the additional measures.
@Tim_Sills because a fuse is a single use item. If you can have the power shut off and then correct the issue and manage to still use the board, that would be ideal. At least I think that would be ideal. The switch cooling off, allowing current to flow again and then heating up due to a short that is not corrected would possibly be a problem. It is about like we need a breaker and a thermal switch. hahaha
Unfortunately, cheap self-resetting thermal fuses are pretty unreliable. They take WAY too long to trip, and don’t always go back to the same tripping temp.
There’s a nice style of snap-action bimetallic thermal switches that I might put on my next build plate. You can get those made at any trigger temp. Not cheap in small quantities though.