M2 Fire Report

General discussion topics
User avatar
sthone
Posts: 897
Joined: Tue Jul 08, 2014 8:25 pm
Location: Connecticut
Contact:

Re: M2 Fire Report

Post by sthone » Tue Jul 25, 2017 2:29 pm

hybridprinter wrote:Also good idea to have a smoke/CO2 detector with loud alarm sitting next to the printers (which of course would only be beneficial when running the printers when on-site).
Not that I run my printer when I'm not home but I do run them overnight so I have a Nest smoke detector hung directly above my printers and another one up stairs so at least that should give me a bit of warning should the worst ever happen. I also have all three of my printers plugged into TP-link smart plugs too so I can shut them off remotely if the need arises. (The timer feature on these is great too.) I even thought about hanging a few of these stovetop firestop's over the printers too but I haven't been worried enough to do it yet.
____________________________________________________
See my projects at https://www.theneverendingprojectslist.com

User avatar
ednisley
Posts: 1188
Joined: Fri Apr 11, 2014 5:34 pm
Location: Halfway up the Hudson
Contact:

Re: M2 Fire Report

Post by ednisley » Tue Jul 25, 2017 3:39 pm

Tim wrote:Redundant temperature sensors
Already there:

Code: Select all

#define MAX_REDUNDANT_TEMP_SENSOR_DIFF 10

Enable this option to use sensor 1 as a redundant sensor for sensor 0. This is an advanced way to protect against temp sensor failure. If the temperature difference between sensors exceeds MAX_REDUNDANT_TEMP_SENSOR_DIFF Marlin will abort the print and disable the heater.
But, frankly, I wouldn't trust that any further than I could throw a glob of flaming plastic: the most likely cause of a stuck-on heater will be a failed MOSFET that the firmware can't control.

As a first line of defense, sure, but firmware doesn't come close to solving the problem …

User avatar
Tim
Posts: 1205
Joined: Thu Apr 10, 2014 2:19 pm
Location: Poolesville, Maryland
Contact:

Re: M2 Fire Report

Post by Tim » Tue Jul 25, 2017 8:25 pm

ednisley wrote:the most likely cause of a stuck-on heater will be a failed MOSFET that the firmware can't control.
Perhaps. But there seem to be numerous reports of thermistors having problems, which may be due to the tiny wires connecting to it operating in a hot, vibrating environment. There are other problems like having the whole system stop dead due to a thermistor mis-read which could be avoided with the redundant thermistors.

wmgeorge
Posts: 200
Joined: Thu Jun 18, 2015 1:09 am

Re: M2 Fire Report

Post by wmgeorge » Tue Jul 25, 2017 8:51 pm

You can not depend on electronics or circuits to be the fire shut off. The only really approved way and its UL listed is a thermal fuse mounted to sense over temperature and positively break by mechanical means, the power to the heater. Think toaster, coffee pot, space heaters and the like. A UL listed thermal fuse does not exist on the market anyway to take the heat of the extruder. Therefor it will need to be mounted away from the extruder and mounted in such a way that it would activate if the heater malfunctioned.
Retired Master Electrician, Commercial HVAC/R,CNC Router

User avatar
ednisley
Posts: 1188
Joined: Fri Apr 11, 2014 5:34 pm
Location: Halfway up the Hudson
Contact:

Re: M2 Fire Report

Post by ednisley » Tue Jul 25, 2017 9:13 pm

Tim wrote:problems like having the whole system stop dead due to a thermistor mis-read which could be avoided with the redundant thermistors.
That way lies madness: "Hmm, looks like we have us a failed thermistor here, but … hey, the hot backup looks good! Let's use that instead and continue like nothing happened. What could possibly go wrong?"

When you discover a hardware failure, kill the power: everything else is irrelevant.

Indeed, protective hardware must always fail safe, a principle ruling out firmware, electronics, and circuitry that must activate to shut off the power. Most of the clever ideas we come up with do not satisfy the "fail safe" criterion, but if you keep asking "What happens when this part fails in the worst possible way?" and discarding the offending part, eventually you're left with the simplest possible design.

A normally open relay held active by a normally closed thermal switch on the hot end comes pretty close to the ideal: any wiring interruption will release the relay and kill the power. It doesn't cover a failed-closed relay or power switch, but you can test the relay every time with a manual power-on switch and you should periodically test the thermal switch by (slightly!) overheating the hot end.

Consumer-grade 3D printers aren't anywhere close to that level of protection, but it's definitely in the foreseeable future …

User avatar
insta
Posts: 2007
Joined: Tue Sep 16, 2014 3:59 am

Re: M2 Fire Report

Post by insta » Tue Jul 25, 2017 9:33 pm

I kind of like the idea of using a metalic spring-strip and a magnet held against the hotend. If we overheat, we just hit the curie point of the magnet and *pop*. :mrgreen:

Or, more plausibly, a zinc strip bonded to the hotend that the heater cartridge power has to run through. Zinc melts at 420C.
Custom 3D printing for you or your business -- quote [at] pingring.org

User avatar
Tim
Posts: 1205
Joined: Thu Apr 10, 2014 2:19 pm
Location: Poolesville, Maryland
Contact:

Re: M2 Fire Report

Post by Tim » Wed Jul 26, 2017 7:27 pm

ednisley wrote:That way lies madness: "Hmm, looks like we have us a failed thermistor here, but … hey, the hot backup looks good! Let's use that instead and continue like nothing happened. What could possibly go wrong?"
It's not entirely mad if done right. It can easily pick up on an error condition in either thermistor and determine if it is an intermittant error (one thermistor reads low for a few cycles, but the other one remains constant) or a persistant error. And one would never keep the printer running through a persistant error. Although, on the other hand, there are strict physical limits on how fast the heating cartridge can heat or cool, so even with one thermistor it should be possible to know when you have a flaky readback as opposed to a serious problem.

Gwhite
Posts: 372
Joined: Tue Jul 11, 2017 3:38 pm

Re: M2 Fire Report

Post by Gwhite » Wed Jul 26, 2017 7:56 pm

insta wrote:I kind of like the idea of using a metalic spring-strip and a magnet held against the hotend. If we overheat, we just hit the curie point of the magnet and *pop*. :mrgreen:

Or, more plausibly, a zinc strip bonded to the hotend that the heater cartridge power has to run through. Zinc melts at 420C.
The curie point idea is very do-able. Weller uses this approach for their "Magnastat" soldering irons. They have alloys that go non-magnetic at 260°C, 360°C, 370C, 425°C or 480°C

User avatar
insta
Posts: 2007
Joined: Tue Sep 16, 2014 3:59 am

Re: M2 Fire Report

Post by insta » Wed Jul 26, 2017 8:25 pm

ed's point is more that some of the failures (although I haven't seen them for awhile, ed?) are the MOSFET itself sticking on. Since the firmware can't physically power off the power supply, it doesn't matter if it turns off the faucet, the pipe below the sink has already burst. The water is coming through regardless.
Custom 3D printing for you or your business -- quote [at] pingring.org

User avatar
ednisley
Posts: 1188
Joined: Fri Apr 11, 2014 5:34 pm
Location: Halfway up the Hudson
Contact:

Re: M2 Fire Report

Post by ednisley » Wed Jul 26, 2017 11:33 pm

Tim wrote:one thermistor reads low for a few cycles
IIRC, Marlin already averages / filters a dozen-ish thermistor readings to weed out glitches, so those "intermittent" failures have been around long enough to punch through their normal error suppression.

Mad props to the Marlin devs for stuffing all that function into an 8-bit microcontroller, but … adding more code to decide how permanent an intermittent failure must be before it must kill the printer seems ill-advised.

Post Reply