3D Printer Fire protection

[paulca]

Hi,

tldr; could I use an IR sensor and smoke sensor to provide a "last man standing" fail safe against unattended 3D Printer fires?

So it turns out the 3D printer I bought does not have "thermal runaway" protection or the manufacturer disabled it in the firmware.  This isn't a big deal I can reflash with firmware that enables it.

3D printer thermal run away occurs when the "hot end" heater for the nozzle and it's thermistor become detached somehow.  Either the heater or thermistor falls out.  The software see the temperature is low so it powers the heater and eventually the heater becomes hot enough to ignite plastic parts, tape or wire insulation etc.  You have a fire.

The same can occur with the heated bed heater, possibly.

The only 100% solution is to never leave the printer unattended.  However it is not uncommon to find larger prints which can take well more than 16 hours to print, so either you need a buddy and take printer sitting in shifts or you have to leave it unattended at some point to sleep.

The solutions for thermal runaway and ensuring the PSU is sane as well as looking for tear down reviews by experienced electronics folks confirming the electronics are rated for the amps they will be pulling etc.... are all very well, but there can be causes of fire which are simply component failures.

First I intend to update the firmware to include said thermal runaway protection.  I also intend to move the printer to the garage, this is more due to the noise of the thing.  The printer will then be monitored by web cam remotely.  If it did go on fire in my garage, as it's just concrete floors, bare stone walls it would need to get pretty intense to get to the next combustible thing, the bare 2x4" wooden roof beams.

However I was considering mounting either (or both) an IR heat detector and smoke detector directly above the printer which would, if hitting thresholds pull all power to the printer.

I know IR detectors exist for kitchen flame detectors and I'm sure hunting for one that would trigger at a temperature of say, 350*C (well higher than the operating temp, but below the ignition temp of plastics) is plausible.  I also know you can get RF 433Mhz smoke alarms, from which you could remove the sounder (or just leave it) and trigger the shutdown off the RF signal. 

Plausible?  Worth doing?

[bd139]

Mine is arriving tomorrow and I worry about this. I am going to not leave it unattended. I can squeeze some prints in during the day fine and long ones can wait until the weekend

[Siwastaja]

Classical single-time-blow thermal cutoff device (thermal fuse) in series. Very commonly used in all thermostat-controlled heaters, because the problem isn't new at all. It's beyond me why it's not employed by default, it's a total no-brainer.

https://www.digikey.com/products/en/circuit-protection/thermal-cutoffs-thermal-fuses/146

Of course, it needs to be mounted properly to the heater element. No iffy kludges.

[Monkeh]

Classical single-time-blow thermal cutoff device (thermal fuse) in series. Very commonly used in all thermostat-controlled heaters, because the problem isn't new at all. It's beyond me why it's not employed by default, it's a total no-brainer.

https://www.digikey.com/products/en/circuit-protection/thermal-cutoffs-thermal-fuses/146

Of course, it needs to be mounted properly to the heater element. No iffy kludges.

The highest rated holding temperature there is 220C. I regularly print at 235C - irregularly at 260C. So, brainer required.

[Siwastaja]

Classical single-time-blow thermal cutoff device (thermal fuse) in series. Very commonly used in all thermostat-controlled heaters, because the problem isn't new at all. It's beyond me why it's not employed by default, it's a total no-brainer.

https://www.digikey.com/products/en/circuit-protection/thermal-cutoffs-thermal-fuses/146

Of course, it needs to be mounted properly to the heater element. No iffy kludges.

The highest rated holding temperature there is 220C. I regularly print at 235C - irregularly at 260C. So, brainer required.

You just need to look somewhere else. The world is literally full of thermostat controlled systems at that temperature range, and many certainly do have thermal fuses for safety. They must exist.

[paulca]

This might be the ideal solution:
https://www.amazon.co.uk/AFO-Extinguisher-suppression-Self-activation-All-extinguisher/dp/B078NW8DBS/

Although it wouldn't kill the power to the device, though it's unlikely to reignite

[paulca]

Mine is arriving tomorrow and I worry about this. I am going to not leave it unattended. I can squeeze some prints in during the day fine and long ones can wait until the weekend

What did you get, out of interest?  Some are worse than others.   Apparently, based on a tear down I didn't do too bad with a Creality CR-10s.  It's got a Mean Well 12V 30A PSU still in it's own case, inside the main control box.  It has proper gauge wires with ring terminal connections.  It also has the heated bed mosfet off loaded from the main board onto a heatsink mounted to the side of the case.

However, even without leaving it unattended it is recommended to have a dry powder or CO2 extinguisher on hand.

That said I have yet to find a CR10 that has gone on fire.  ANet A8s or whatever they are called seem to be the worst with quite a few fires and near fires reported.

[paulca]

Classical single-time-blow thermal cutoff device (thermal fuse) in series. Very commonly used in all thermostat-controlled heaters, because the problem isn't new at all. It's beyond me why it's not employed by default, it's a total no-brainer.

https://www.digikey.com/products/en/circuit-protection/thermal-cutoffs-thermal-fuses/146

Of course, it needs to be mounted properly to the heater element. No iffy kludges.

Does this not just move the thermal runaway issue elsewhere?  Usually cased by the heater or thermistor falling out, which would probably decouple it from the thermal cut-off as well.

I also note that some printer firmware has temperature watch dog timers which check that the thermal actions are being taken and if they aren't, due to stuck/missing sensor or some other software issue the device is put into lock out and effectively powered off.

[Monkeh]

Classical single-time-blow thermal cutoff device (thermal fuse) in series. Very commonly used in all thermostat-controlled heaters, because the problem isn't new at all. It's beyond me why it's not employed by default, it's a total no-brainer.

https://www.digikey.com/products/en/circuit-protection/thermal-cutoffs-thermal-fuses/146

Of course, it needs to be mounted properly to the heater element. No iffy kludges.

The highest rated holding temperature there is 220C. I regularly print at 235C - irregularly at 260C. So, brainer required.

You just need to look somewhere else. The world is literally full of thermostat controlled systems at that temperature range, and many certainly do have thermal fuses for safety. They must exist.

Well, feel free to find one.. And while you're at that, come up with a compact, reliable way to integrate them into the hotend (which needs to be small!) and wire them in such a way the heater cartridge can be replaced easily.

ANet A8s or whatever they are called seem to be the worst with quite a few fires and near fires reported.

A handful, one clearly due to a loose heater (okay, firmware can avoid this, but still.. 'someone' (I wonder who) didn't secure the heater properly), others harder to pinpoint - one looked like it went at the main board, perhaps the monkey with the screwdriver shouldn't have damaged it (no no, clearly, it was the MOSFET - they're always catching fire you know).

[Monkeh]

Dispensing with some of the salt..

A simple smoke alarm in the same room as the printer will warn nicely if it goes horribly wrong. There should be options available which can be used to disconnect power via a relay, which would be beneficial.

Those fire suppression ball things should be quite effective, but also extremely messy.

[bd139]

Mine is arriving tomorrow and I worry about this. I am going to not leave it unattended. I can squeeze some prints in during the day fine and long ones can wait until the weekend

What did you get, out of interest?  Some are worse than others.   Apparently, based on a tear down I didn't do too bad with a Creality CR-10s.  It's got a Mean Well 12V 30A PSU still in it's own case, inside the main control box.  It has proper gauge wires with ring terminal connections.  It also has the heated bed mosfet off loaded from the main board onto a heatsink mounted to the side of the case.

However, even without leaving it unattended it is recommended to have a dry powder or CO2 extinguisher on hand.

That said I have yet to find a CR10 that has gone on fire.  A8Net's or whatever they are called seem to be the worst with quite a few fires and near fires reported.

I bought a Creality Ender 3. It has metal frame and very few plastic parts like the CR-10. I suspect it shares many similarities. I will be running it on the fire place. I have a powder fire extinguisher as well that lives in the hall. The main thing that worries me isn't the thermal overrun but the wun hung lo power supply. Hopefully it turns up with a decent one . Also I intend to make sure that the lapping flames don't end up setting fire to the reel of PLA on top of the unit which is where I reckon most of the serious fires come from.

Apparently since end of 2018, all Ender 3's ship with hotbed and extruder thermal runaway protection enabled as well. I will test this though.

A8's are nasty. The whole frame is flammable!

I've got a smoke alarm in the room so plan is:

1. Visual or audible warning of fire.
2. Isolate it immediately (pull the plug out of the wall)
3. Evaluate. If it's small, grab the fire blanket from the kitchen, remove PLA spool first if possible. If it's large, powder it.


Edit: anyway I thought I'd mention that it's probably ranks up there with the other silent killers: the fridge and the dryer: https://en.wikipedia.org/wiki/Grenfell_Tower_fire (started with a fridge). Also to note, a shop I used to go to as a kid burned down because the proprietor decided to nip down to the pub at lunch, got pissed and just went home for the afternoon leaving his old fire stick soldering iron running. Woosh!

[paulca]

The Ender 3 and the CR-10 are nearly identical.  I think the CR-10 has a few upgrades to reflect the higher price.

It seems the Creality printers just buy off the shelf Mean Well supplies, which aren't too bad, as I understand.

You can print yourself a cable reel stand and filament guide and put it off to the side instead of on top.

Apparently since end of 2018, all Ender 3's ship with hotbed and extruder thermal runaway protection enabled as well. I will test this though.

 

Good to hear, but I might flash the latest "Marlin" anyway.

Not sure how much you have researched yet, but the major pain the rear end with these printers is getting the print to stick and stay stuck.  I'd do quite a bit of pre-research on "Bed Levelling" and "Bed adhesion" and consider buying some cheap options like:

Glue stick - I found Prit Stick did NOT work.  It hardens into a solid with the heat and is an absoltely nightmare to remove, not alchohol soluable either.
Blue painter tape (the printer came with yellow stuff which does work)
Hairspray - makes the surface nice and tacky but can make a mess with overspray (at least it's alcohol soluable)

Apparently people swear by PEI sheets but they are expensive.  £10 for a 300x300 sheet.

I would highly recommend following "Makers Muse" tutorials on "First Layer" settings.  The first layer is critical.  It's really disappointing to get 1 hour into a print and have it come unstuck.... 4 times in a row!  Ask me how I know.

[paulca]

the other silent killers: the fridge and the dryer: https://en.wikipedia.org/wiki/Grenfell_Tower_fire (started with a fridge)

I thought it was the tumble dryer.  Anyway I have one of those hot point dryers.  It has been modified by the service technician but some engineers have said the modification is not enough and fluff can still build up on the element and eventually cause fire.  The modification was apparently to install a plastic screen to try and stop the fluff reaching the element.

It's in the garage which is detached at least.

[bd139]

Threw mine in the trash instantly and bought a Bosch one 

Thanks for the advice with getting into it. Much appreciated. Writing up some extensive notes so have added to that. So much to learn  .

I live with a hair dresser so I have a fine collection of hair sprays to steal 

Meanwell supplies are excellent usually. I have used a few in other projects with no problems and the engineering is top notch.

[paulca]

On the topic of dodgy electronics, I discovered if I move the extruder head across the arm while the printer is powered off that the power generated from the stepper motor powers the display back light.   

[bd139]

That's "interesting". I suppose that's probably supplied off a resistor from the 24V supply then.

[Monkeh]

Glue stick - I found Prit Stick did NOT work.  It hardens into a solid with the heat and is an absoltely nightmare to remove, not alchohol soluable either.
Blue painter tape (the printer came with yellow stuff which does work)
Hairspray - makes the surface nice and tacky but can make a mess with overspray (at least it's alcohol soluable)

Glue stick is water soluble.. It's PVA.

Apparently people swear by PEI sheets but they are expensive.  £10 for a 300x300 sheet.

How is that expensive? If you don't damage it it'll outlast most of the printer.

On the topic of dodgy electronics, I discovered if I move the extruder head across the arm while the printer is powered off that the power generated from the stepper motor powers the display back light.   

That's "interesting". I suppose that's probably supplied off a resistor from the 24V supply then.

The LCD backlight is fed from the 5V rail. The protection diodes in the stepper drivers will feed the 24V rail if you move the steppers unpowered, which will feed the 5V reg and off we go.

[janoc]

On the topic of dodgy electronics, I discovered if I move the extruder head across the arm while the printer is powered off that the power generated from the stepper motor powers the display back light.   

Don't do that - you can blow your stepper drivers. If you need to move the extruder or the axes manually, do it slowly to limit the issue (or unplug the motor, if it is accessible).

[dnwheeler]

Working on medical devices for a living has taught me that most consumer products (especially cheap ones from China) don't have even the simplest forms of fault detection systems.

The 3D printer firmware should be designed to notice that the temperature isn't rising when the heater is on, and shut down the system. They should also be detect an open or short in the thermistor circuit.

[bd139]

That's what the thermal runaway protection does. It notices that the control loop isn't functioning and shuts the thing down.

To be honest it's no more dangerous than a Weller TCP with a stuck thermostat. Been there! 

[janoc]

Working on medical devices for a living has taught me that most consumer products (especially cheap ones from China) don't have even the simplest forms of fault detection systems.

The 3D printer firmware should be designed to notice that the temperature isn't rising when the heater is on, and shut down the system. They should also be detect an open or short in the thermistor circuit.

That's exactly how it is implemented in the Marlin firmware (which most of these cheap machines use). Any abnormal value or the temperature rising slower than expected/falling when the heaters are on will trigger a panic and kill the print, requiring hardware reset.

However, thermistor faults are only one way how a printer can catch fire. For example, if a short develops you have a problem - the power supplies in these things routinely deliver 20-30A so something can go up in smoke easily way before a fuse blows (assuming that there is one - many machines don't have any!) or the overcurrent protection trips. Or the power supply could go bang on you. Or an axis stops moving for some reason (stepper driver failure, belt failure, ...) and the hotend cakes itself into a pile of molten plastic, eventually catching fire. Etc.

Running a cheap consumer machine unattended somewhere where a fire could cause major damage or a loss of life (i.e. home) is a terrible idea, no matter how many smoke detectors, fuses, cameras and similar are added.

That a print requires 16 hours to do is not really an argument anymore - most new machines are capable of recovering from a loss of power (Marlin 2.x feature), so you can just turn the thing off, go to bed and come back to it in the morning and continue. Of course that works only for materials that don't pop off the printing bed when it cools (e.g. PLA on the blue painter's tape) but it is still better than having a fire overnight.

3D printer is a machine tool and should be treated like one. CNC mills and lathes are not left running unattended neither and those machines are typically built much more robustly than a $200 Chinese 3D printer.

[exe]

I'm going to put a fire alarm above the printer.

[BillyD]

Domestic burglar alarms can support smoke detectors. I imagine it operates like any of the other devices (pir/reed switch/etc) on the system where it runs on 12V and opens a relay if the sensor is triggered.
One of those could fairly easily be linked to the printer supply to shut it off, sound an alarm, etc.

Whether it also needs to activate a sprinkler system, call fire brigade, etc is a whole other dilemma as your home-made device is now at least responsible for protecting your property and everyone in it!
How much do you trust it, is it rock solid, etc!

[nixxon]

I recently bought a Sidewinder X1 3D printer made by "Artillery3D" http://www.artillery3d.com/. It is said that their designers previously used to work for various, competing 3D printer manufacturers. However the designers were not satisfied with how user feedback and reported issues were handled by their previous employers, so they decided to make their own gamechanger 3D printer that is supposed to punch above it's belt for a $380 3D printer (now $399). The X1 has a fast heating 110/220V heated bed (30x30cm), volcano nozzle, direct filament feed and thermal runaway protection and silent stepper drivers/motors to mention some of the good stuff. The chassis is also actually grounded The item was delivered at my door by DHL Express within a week (from Hong Kong) after placing the order, for a shipping cost of $8 (+ a local DHL customs declaration fee of $30, though (stuff is expensive in Norway)).

The 3D prints look really good, using some random PLA filament that I got from my local general hardware shop (Flashforge 3D printer spool of 0.6 kg PLA). Link: https://www.clasohlson.com/no/Filament-PLA-til-3D-skriver-Flashforge/38-7721-4

If you consider buying a 3D printer using Fused deposition modeling (FDM) I would check this one out. I got version 4 that has developed quite a bit compared to the initial release version that was introduced 1 year ago. If you buy, make sure you get at least version 4. Here is a link to the seller that supplied my unit: https://www.banggood.com/Artillery-Sidewinder-X1-3D-Printer-Kit-with-300300400mm-Large-Print-Size-Support-Resume-PrintingFilament-Runout-Detection-With-Dual-Z-axisTFT-Touch-Screen-p-1411501.html?rmmds=search&ID=48036&cur_warehouse=CN

They also sell it on Amazon and various other places.

[CatalinaWOW]

While a fire is a theoretical possibility, it requires a sequence of failures.  Not something I am going to lose sleep over on my Ender 3.

1.  Initial failure of thermocouple etc.

2.  Software failure (this one isn't actually totally unbelievable.  While I believe the existence of temp slope detection, this must shut off after stable temp is reached.  Then some stack overflow or the like could open the control loop.)

3.  The hot end has to get to ignition temperature.  Again probably not a very low probability.

4a.  Fire has to propagate through bowden tube and up to filament reel to get enough flammable material to spread beyond printer.
4b.  Print head has to get hot enough to radiation or convection heat material around printer.

It is easy to eliminate 4b. by keeping flammable material away from the printer.  Far easier than adding yet another protective circuit.  So I have done that and stopped worrying.