The holidays are almost here, and with that comes the traditional Mass Consumption of Consumer Goods and Gift Exchange. 3D printers are getting really good and really cheap, and it’s inevitable that a lot of 3D printers will be given as gifts this year. Be careful if you’re giving or receiving one of these printers: they can cause fires as [Ben Hencke] found out when diagnosing a problem with a printer he bought this year.
The printer in question is the Monoprice Maker Select V2, a Prusa i3 clone with impressive specs for a $300 printer. This printer is a rebranded Wanhao Duplicator i3, and we’ve reviewed it favorably. It’s a capable printer that beats the pants off of any Kickstarter printer in quality (and for the fact that you can buy it right now). We’re pretty sure there are going to be more than a few of these printers under the Saturnalia tree this year.
After a few weeks, [Ben] noticed a bit of smoke coming from the printer while the bed was preheating. This wasn’t blue pixie smoke, like you’d find from an exploded capacitor. There was a lot of smoke.
After a closer inspection and help from [Elecia White] from embedded.fm, the problem was traced to the power connector for the heated bed. The green, bromine-infused plastic for this connector was charred and there’s little doubt this could have caused a fire.
3D printing is a fantastic tool, and has enabled more hacks and builds over the last few years than we could have ever imagined. 3D printers are getting very good, and very cheap, and of course this will eventually mean someone losing their workshop to a printer fire. Until someone figures out how to build a ‘thermal fuse’ or something of that nature, 3D printers — from the high-end ones to the still very good Monoprice and Wanhao units — have the potential to start a fire.
This is fairly common with cheap printers.
Those of us that have built there own from scratch regularly check and maintain there printers.
Upgrade components for reliability and try and pre-empt such failures.
Ive replaced a number of those power connectors with known, higher rated and good quality replacements.
But there are people out there that are expecting to be able to treat them like a piece of kitchen equipment and just expect them to work.
When you think about it, that’s tells you something about kitchen equipment. Yeah an oven is simpler, but it’s pretty amazing how rarely the stuff I use every day burns my houses given that some of them involve literal fire in normal operation.
Yay for product liability and UL standards, eh?
3D printers will get safer as soon as people start getting sued or told they can’t sell their flammable junk to the public.
The possibility of 3D printer catching fire usually DOES NOT depend on a particular manufacturer of printer, because most manufacturers are using the similar parts. Instead, the possibility of fire depends on the version of firmware that is installed ! More recent firmwares are more advanced and have the additional protection measures – like against a thermistor coming off place. For example, below you can find a commit message 43c298a (dated Jun 30 2014) from a Marlin Firmware repository. My cheap Chinese 3D printer had a slightly older firmware version installed, so I had to update its’ firmware to enable this “Thermal Runaway Protection”! Always update a firmware of your 3D printer!!!
/*================== Thermal Runaway Protection ==============================
This is a feature to protect your printer from burn up in flames if it has a thermistor coming off place (this happened to a friend of mine recently and motivated me writing this feature).
The issue: If a thermistor come off, it will read a lower temperature than actual. The system will turn the heater on forever, burning up the filament and anything else around.
After the temperature reaches the target for the first time, this feature will start measuring for how long the current temperature stays below the target
minus _HYSTERESIS (set_temperature – THERMAL_RUNAWAY_PROTECTION_HYSTERESIS).
If it stays longer than _PERIOD, it means the thermistor temperature cannot catch up with the target, so something *may be* wrong. Then, to be on the safe side, the system will he halt.
Bear in mind the count down will just start AFTER the first time the thermistor temperature is over the target, so you will have no problem if your extruder heater takes 2 minutes to hit the target on heating.*/
*their x2
Your truely doing Gods work hear
*Here x1
You’re* truly* God’s*
But who’s counting.
Kitchen equipment is responsible for its fair share of fires as well ;)
Probably the cheapest connectors they could find, there are some horrible knockoff ones available out of china. I have had that happen with some DC-DC converters I put into use in a car.
I had an expansion card burn up. I suspect vibration over time loosened the power connector enough to cause a bad connection. 3D printers could have a similar failure mode.
As an owner of a Wanhao i3 (literally the same as the one mentioned in the story, but with original Wanhao markings, instead of monoprice markings), and as someone who actually had, and solved the issue, I can tell you what the problem was directly; and it’s not directly the plugs.
The issue is with poor contact to ground, some of the issue is then traces being small and thin, some is that the connectors don’t make good enough electrical contact with the wires, and some of it is that unless the wires were tinned they oxidise rather quickly, all making for a pretty massive resistance build up, leading to the heat damage you can see above (though in my experience, and that of most owners that have come across this issue, the connectors detach well before they reach any near their ignition points).
TL;DR
The issue is resistance, the plugs don’t help because this style of plug makes poor contact, but it’s not because they can’t handle the current
You should never tin a low resistance connector because the resistance of solder is higher than the resistance of copper oxide.
What!? Copper oxide is a poor conductor.
Nonsense!!!!! Every parallel resistance on trace will decrease overall resistance
The resistance of lead oxide is MUCH higher than the resistance of tin plate or gold plate.
If the copper on an edge connector is oxidized, do NOT coat it in lead-tin solder. Clean it to bare metal and use electroless tin plate.
If the point is to thicken and protect a trace on a PCB, then by all means, coat bare copper traces with solder.
Doubt it’s the connector the culprit. It it was, the interior of the connector would have burnt, not the outside.
I bet it’s a copper trace not wide enough to handle the current.
Chinese engineering at it’s best hehe
The trace might be OK when it was done in a 2oz cooper pcb, and now being done in a 0.5oz cooper pcb..
0.5oz is in most cases more expensive to produce because of how hard it is to reliably form thinner copper sheets… actually, in many processes, the outer layers /start/ as 0.5oz and then get plated up to 1oz during the PTH process, so finishing with 0.5 oz is actually somewhat difficult if you still want reliable vias.
I beg to differ. Gone are the days when you could put 3 Amps through a 2 Amp connector. Now days you might be lucky enough to get 2 Amps though a cheep Chinese 2 Amp connector if you cool it to zero degrees Kelvin and use super-conductive wires.
If you look at the picture you will see that they used selves around the wires that are not copper and probably acted as a thermal insulator to some degree.
Remember that copper is an excellent thermal conductor so the insulation of the copper wire itself should show signs of thermal damage if it was raised to the same temperature as the connector. So obviously the copper wire did not have a good thermal connection to the connector itself.
If you then look at the diameter of the sleeve used over the wire and compare that to the maximum radius of copper wire you would expect to fit in that block connector then you realize that wire is undersized. Perhaps it may have a cross-sectional area sufficient to carry the current required with an acceptable voltage loss *but* the design of the block connector and it’s testing probably intended or had larger wire that also performed the secondary function of dissipating the heat through the copper wire itself.
When you design things to %100 of component capacity that you can only expect %100 failure. The old ‘rule of thumb’ was to design to 50% – 75% capacity and even 75% was frowned upon because components like electrolytic capacitors have huge variances not just in manufacture but also in early product life.
I regularly see people design at 100% of capacity. “I only need 1A, so I got a 1A regulator.”
I’m not sure I get all your explanation…
Here’s my detailed mental path:
Let’s assume max current is 10A (I think circa 100W at 12V is in the ballpark for hot beds)
If wires were undersized it’s most likely they would have melted.
BTW it’s probably over 1mm2 (awg16) which has a 20°C rise at 13 amps.
https://wiki.xtronics.com/index.php/Wire-Gauge_Ampacity
-> wire size issue is unlikely.
On the other hand, you need a 7mm wide 35um thick trace to get 13A at 20°C rise.
https://www.eeweb.com/toolbox/external-pcb-trace-max-current
I doubt the pcb designer used traces that wide, but it’s hard to confirm.
-> in my experience, undersized trace is much likely (I saw close to a dozen failed traces in different boards but a single wire size issue)
Connectors usually fail at the contact point(s) between the female and male contact. Basically all the current flows into a section much smaller than the contact section itself, thus locally increased resistance and heat dissipation.
So if it was a connector failure burnt traces would be visible on the remaining pins and interior of the connector housing would be much more damaged, being closer to the heat source.
-> most certainly not a connector issue
That’s what led me to the conclusion: undersized pcb traces
This isn’t about current carrying ability, it’s about thermal dissipation.
For example, how much current can a copper wire carry – you sent generalized tables/formulas but these generalizations or assumptions hide some of the realities.
It comes down to what makes copper fail and the answer is heat or explosive decomposition – these are often a good indicator that the copper wire is too thin. This will occur at an approximate ‘point’ and that ‘point’ is actually a temperature.
For copper that function is rather linear because heat oxidizes copper and copper oxide has a conductivity very close to the copper itself and that is why you can make generalized tables that will be close enough to the reality.
However this is not the case for solder for two reasons. First it is strongly disadvantaged by a great different in melting point: Copper 1080°C (1981°F), Solder 188°C (370°F). And secondly the oxides created by heating solder are far less conductive so as it heats it becomes more resistive which in the case of being heated by current will even further increase it’s temperature iteratively and resistance as thermal run-away.
It’s the second problem with solder above that caused this failure in combination with two other factors.
If you look at this picture –
https://cdn.hackaday.io/images/4344021481231639513.jpg
You will see that the holes are far too large for the pins. So instead of having a copper (PCB plated hole) to Tin (pin coating) to copper (pin internal) connection you have solder between the PCB track and the pin.
Only 5 to 10 watts will eventually fry that solder connection and result in thermal runaway and with some math at 12Volts and 10 watts you only need a resistance of as little as 0.07 Ohms so see the problem?
The other two issues are the type of sleeve used on the wire and the thickness of the copper. I will forget the sleeve for now as it’s out of scope.
The closer you run copper wire to it’s limit the more heat it will generate and in his case that would be contributing to the self heating of the solder joint. This is far more important than it seems.
Copper itself can be a heat sink that will raise by x degrees per Watt of heat. If the copper wire was larger then it would be acting as a heat sink rather than contributing to the self heating of the joint and it would have *most likely* prevented the thermal run-away from occurring in the first place.
So in effect – the size of the copper wire is just as much a mage or break contributor to this failure as the solder in the PCB connection.
And that is why you should NEVER use components (even wire) at full capacity.
There must be hundreds of these PCB’s out there that haven’t failed. They will have thicker wire!
When you raise the resistance of a wire, you drive the current down, and that globally lower heat production as P = RI². So maybe I am mistaken or overlooking something here, but I don’t think it works as you put it. Don’t hesitate to comment to correct me if I am wrong
I derate all no-name chinese components by 50%. Active or passive, it doesn’t matter, if you run them at more than 50% of what they are rated for you’re asking for trouble.
Ooh, things that get hot can catch fire, who would have thunk it. What next things that use water can leak?
Of course, but that is the reason e.g. washing machines have several protection measures against leakage. (*) And there are ideas to make plausibility checks in the firmware.
Which unfortunately does not help against gross user negligence. What gave me once a flooded bedroom from above.
Obviously, the wrong rated nut was used.
Or MAYBE this is a built-in slow-blow safety?
Russian indicator light. It tells you it’s on.
I LOL’ed
Water-cooled computers. ;)
lol
Haha epic!
I work in the railway industry, and sometimes I get to work on locomotives.
Once I was at a client’s workshop, two technicians blew a 630amps fuse but the workshop didn’t had any spares left. (those are used in-line with the 24V electric starter of the 3000hp diesel engine)
Tech A asks tech B: “what should I do?”
Tech B: “slam in a 13mm wrench. It fits.”
They casually slammed the wrench in and started the engine. Amazingly, everything went fine.
This is the day all my stress working on a 3M€+ machine instantly vanished.
In the above picture, copper bar or terminal oxidation prevents the current from flowing directly from the bar to the ferrule. The current has to go through the nut, which normally is only a mechanical feature.
In railway industry bus bards and battery terminals are often coated with special grease that prevent oxidation but allow electrical contact. Not may even be insulated, though not common.
The railroad equivalent of the “penny in the fuse box”.
Seriously, that pic has become a HaD meme! I’ve seen it 4 or 5 times since January, lol.
Is that simplified Chinese? Well I am not surprised.
Also ants like to get in and build nests inside the display.
This is a pain to remove, but it does feel nice to pull a display apart until you even take out the zebra stripes and have it back together working pixel perfect.
Also. The extruder sucks. And the bearings have no grease. But the motors/frame/etc are great! It’s a good starter printer that you can then upgrade!
Happened to me as well, couple years ago, when I just built my first prusa i2.
It was my fault, I made a crappy joint on my molex connector for hot-end, one of the pins came loose a bit.
When I was in another room, something smelled bad, went to look and came across this:
https://lh3.googleusercontent.com/WV7o-tXJhAJOwqGjXI7nV5uNHiqF07AyBdvBWI26qj8_JO2dObpaLOkM-MffmOOpxpCF1KzCSSpYFg=w1920-h1080-rw-no
Like any equipment that gets hot and uses a large amount of current, take care during use and don’t operate when your not there
Heat guns, hearing plates, large lithium chargers exc
I had a hako catch fire last year, the temperature feedback failed in some manor and it heated up so hot it melted the retainer and caught fire in under an hour of being left alone
Heating*
manner*
Heh, reminds me of that digital soldering station i was using a few months ago. The temp feedback failed and told the controller that the tip is cold. The controller in turn dumped enough power into the tip that the latter one became red-hot and not just slightly red, more like “it’s so hot that it’ll start melting if you don’t turn it off immediately”-red.
Which Hakko? Might want to know for my own reference.
Manner*
Part of the problem may be the youthfulness of consumer 3D printers. You’ll note that laser printers use a fuser unit, and they’re not bursting into flames, left and right.
If I were the women in the video I would be more concerned with teaching myself not to wrap my fingers round the back of the mains switch plate when turning it on, that is an easier way to end up dead than waiting for a 3d printer to burst into flames.
What usually happens with that particular connector and why regular checks and maintenance is important, as well as cheap parts being used. Is the wire in the screw down can get loose and need tightening down again after a while. When its not fully tight it doesnt make full contact and gets very hot.
Hot enough to melt and scorch the connector and board.
Which only gets worse till it fails.
If your lucky the printer stops working at worst a short and fire.
Except, I don’t think it is the connector. Look at this photo:
https://2.bp.blogspot.com/-GXvPUmkLXf8/WEWH-3HzzWI/AAAAAAAAFoQ/74Yl8SEJnFwpetUZ8XErlJAdiqPtSYIIgCPcB/s1600/Photo%2BDec%2B04%252C%2B11%2B22%2B43%2BAM.jpg
Look at the pins left of the burned bit. Does that look properly soldered? I think it’s a bad solder connection between the connector and the board.
“Does that look properly soldered?” Nowhere near it. But have you seen the sparks on the PCB in the video when they applied power to the thing?
The good old slap an electrolytic touching the hot mosfet school of design.
Somehow you have to limit the lifetime of the appliance. But you can still exaggerate: Glue some sticky copper tape around an electrolytic cap and solder the TO220 transistor case to this. I found this in the PSU of an external harddisk which failed several days after purchase. It was an altogether crappy design, wasting 3W of power just doing nothing (except staying hot).
Yeah i had this same issue and i traced it to a bad solder on the heated bed connector too
pulses need higher rated connectors tan for example dc or 50hz ac. but looking at the spot where the heat was, this was probably no soldered hot enough or with “solder” with a high amount of unknown ingredients. also I found out that the surface finishing of cheap pinheaders or connectors is horrible. It actually repels solder until heated far above soldering temperature…
Absolutely the pulsing would have made this worse. It’s going to not just be carrying a ton of current when the hot bed is on, but it’s going to be cycling. Hot beds have a long PWM time (or worse, bang bang control) that is going to thermal cycle from fairly cold to fully hot, over and over again. Any mediocre soldering job, or any source of strain, is going to make this crack, and fail, and cause this kind of failure.
i use hi amp vehicle relays for my hot ends and bed noisy but reliable
i use hi amp vehicle relays for my hot ends and bed noisy but reliable
opps i mean high amp
It was not the switching transistor that failed here. You need connectors to your relays too. And I would question the reliability in a high dynamic control loop application, switching at >=1Hz.
Somehow it reminds me on the Galaxy Note 7 -> Next feature for 3D Printers: Does not catch fire
Er, anyone remember that whole Wanhao recall, you know, that was posted on the same blog because of the shorting on the MCPCB due to a lack of spacers? Couldn’t it be just that?
Everyone’s blaming crap connectors, but there’s a good chance damage at the connectors is a red herring – if there’s a solid short on the hotbed itself (which is also quite thermally conductive, so will actually take some effort to force any magic smoke release – even if it is just a small metallic wire short), the first thing to blow is probably going to be connectors that aren’t rated for the job – power, because it needs to take the shorted hotbed component + the rest of the 3D printer load, and then the hotbed connector, which is pushing out slightly less current.
I somewhat doubt the issue is on the control board end – frankly, I’ve seen more sketchily done PCBs in consumer gear than that.
Of course, the real blame is on leaving out any sort of fuse protection for each heater channel from the design (a PTC would easily protect all of the connectors in this case). Given that this happened on two separate printers, I’m also going to suspect some user pattern that’s causing it – perhaps overtightening on the platform screws during initial setup. This is not to say that this means the user is to blame at all – part of the setup is indeed to level and adjust those screws, and it’s a design issue for overtorquing to cause a short in the heated bed (I own one and am reasonably familiar with the unit, but have not experienced this particular issue). If this is indeed the case, there’s also clearly a problem with Monoprice and Wanhao for failing to recall all of their existing stock and repairing the known defect in all new units.
Also, FWIW, it seems as if these units failed immediately on first use, and didn’t magically go on fire partway through a print.
Connectors that are rated for the job* but not rated for the short circuit current. Given the burn pattern matches what you’d expect if there were a short in the hotbed (more damage to the main power connector, which has to handle the most current, and then the hotbed connector), it kind of looks like a bed heater short of some sort.
Also, I’m going to go take mine apart and figure out a good way to bodge (neatly and reliably!) in a PTC fuse on all of those channels…
I’m quite curious now, actually. Any chance anyone who’s experienced this failure can inspect the heated bed or do a low-ohms measurement of the heater resistance? (It’s low enough that in practice most people will not repeatably measure the correct resistance, due to worn out multimeter leads or insufficient probe pressure – try manually zeroing your measurement shorting probes first, or attemping a four wire measurement)
I will take a look. When we tested the bed, it was something like .8 ohms. That seems like the right ballpark for a giant heat-resistor. We didn’t have something that measures precision milliohms though. Maybe I can rig up something passing a current through it and scope the voltage drop as I move the bed around and mess with the tightening screws?
It might be easier to measure the resistance by hooking it up to a power supply and measuring the current.
“damage at the connectors is a red herring”
See the video again, when she turns it on. The hotbed connector is disconnected from the PCB but there are still sparks flying around it for a few seconds. That can’t be caused by a bad connector because…you know…it’s not connected.
I was pre-heating the extruder and bed in prep for a print. The bed wasn’t moving, and the screen showed things heating up just fine, so if the bed was shorted it was a partial short. I’d leveled the bed once, I don’t think I overtightened the platform screws. I had this replacement unit for less than a month and printed several things successfully.
I’m planing on looking for bed shorts. Thinking of passing a low current to it and scoping the voltage drop while moving, pressing, etc. Ideas welcome, I haven’t returned it yet.
After testing that, I’ll take the bed off and look for damage around the screw area and pcb traces.
Actually, did your kit come with these insulating washers? http://wanhaousa.com/pages/duplicator-i3-safety-notice
This would be a pretty huge fail on Monoprice’s part, given that this particular change happened MONTHS ago.
Yeah, I checked mine and it’s quite low as well (somewhere < 1 ohm cold, though I did not get to disconnect the control electronics). However, I haven't had the time to do a proper 4-wire measurement. Basically, you'd hook up a power supply on constant current mode (or manually stabilize it to something so you can fiddle with something else at th esame time, or use a second multimeter to monitor), at a reasonably high current (~1A) to the heater (so this particular heater will get a bit warm). Then, take a multimeter and contact the heater wires (NOT your power supply leads) to measure voltage. This cancels out any contact resistance, lead resistance, etc. you might have had in your power supply leads, and gives you a pretty close to perfect measurement. R = voltage you measure / current you set on your PSU.
I was also tracking another report of the problem, and found out this is not uncommon:
https://www.facebook.com/groups/WanhaoDi3/
I'm starting to suspect that it could actually be the connectors as well, given how marginal the heater seems to be and the other reports on the Facebook group which appear to be happening (some of which looked like they were not on the first run). From what I recall as well, it's quite difficult to put 10A into a 1oz PCB with reasonable traces and not have a substantial temp drop. The connectors may very well be derated from their nominal rating due to the higher ambient caused to nearby hot component (PCB traces and MOSFET)
Many people on that Facebook group are apparently switching their connectors with XT60s, which don't actually fit the PCB layout, so you have to use a big solder blob. My personal opinion is that at 10A peak or so for the heated bed, a solder joint that actually has to span that much distance (more L, higher R, and mechanically pretty flimsy, esp when hot), may in fact be worse and less reliable than the original terminal in some cases. If you or anyone else ends up thinking about doing that (I'm starting to, pending the results of this little investigation), I'd recommend finding an appropriate, equivalent, branded and rated connector from Digikey rather than bodging in another connector with the wrong footprint.
No, neither of the units came with insulating washers, or a warning about them from Monoprice.
I haven’t had a chance to rig a constant current source test the bed for shorts yet, but I’ve inspected the place where the springs meet, and there isn’t any visual damage, not a scratch, but that doesn’t mean a spur didn’t make connection where I can’t see it.
“Until someone figures out how to build a ‘thermal fuse’ or something of that nature,”
Don’t we have reliable sensors for heat, flame and smoke detection? Perhaps the characteristics of 3D printing don’t make it easy to tune control/safety features based on the outputs of these sensors but even a conservative (low) trigger threshold could switch an alarm that, if not overridden by an attendant within a time limit, shuts off the power. Does anyone already implement features like this in their own builds?
You beat me to it :)
“Thermal fuses” are used by the millions to protect transformers (and motors) from overheating.
The transformer one’s are pretty “cost optimised”, maybe cost 20 ct?
But hanging a piece of regular solder in the top of an enclosed printer probably works.
The low temp stuff for desoldering smd probably works a lot better.
Or go McGuyver. Use a piece of string and a (micro) switch.
A Similar solution was on our old central heating oil furnace: Above the burner was a spring loaded strip of plastic, perhaps even celluloid which kept a switch in the on position. As soon as the strip would burn away, it cuts the power to the burner (solenoid valve, oil pump and blower).
But also this failed quite spectacular one time. In summer we pulled the burner out, it was mounted on hinges, to burn wood just for hot water to save oil. For some reason sometimes the main switch was on, perhaps it was not summer but spring and we wanted some slight room heating (by burning wood) and needed the pump to distribute the heat.. Should be no problem as there is the safety switch, isn’t it?
One time we noticed, that the pulled out burner was firing full power onto the side of the oven and onto the oil pipes. Luckily the pipes were copper and we noticed it early enough that there was only slight optical damage to the oven and the house did not burn down.
What had happened?
The safety switch got loose in it’s metal housing, touched the housing with it’s terminal screws and so was shorted by the housing. I don’t know, why this did not trip the RCD or breaker.
After that incident we installed a plug and receptacle into the cable of the burner which got physically disconnected each time somebody decided to pull out the burner.
But some extra safety switches, thermal fuses or even normal fuses to protect against shorted heated beds should be incorporated into 3D printers.
Thermal fuses are like $2 each
Not leaving your 3d printer unattended is not really a hack now,. Is it?
How about putting your printer in a (semi) enclosed box and adding some smoke & heat detection to flip a big relay if anything goes wrong?
An enclosure is a good idea anyway because of temperature control, and noise & fume emisions.
Or go a simple step further and release the content of a small CO2 cardridge in the thing if needed.
“Not leaving your 3d printer unattended” is not an option with longer printing jobs. Even if you are at home, you have to sleep sometimes.
A “small CO2 cartridge” is 8g or ~4 liters of gas. Although an automated fire suppression system is a nice idea, I do not think, that 4l of CO2 gas will do much. You would need a bigger bottle – and of course a reliable power disconnection before.
So, HAD/Brian here’s a challenge:
There are the HackADay challenges/builds etc. but none seem to be interested in safety rather than the latest widget. Why not put forth a challenge to provide an open-source safety system for 3D printers with an eye to bringing safety considerations into the prototype-level things that are posted here? Current draw, temperature and the like can be monitored at critical points. While it’s not foolproof, it can cut the hazard by several orders of magnitude and that’s good for everyone.
Just surround the printer with cam’s and Flir Lepton modules, throw 1000€ in leptons alone to solve what a decently designed pcb, decent connectors and a couple temp sensors could solve..
The challenge would be aimed at a minimum of equipment to take care of the most-likely problems as well as changing the thinking a little. So, yes, temperature sensors, a current monitor and maybe a smoke sensor at the right place….but more than that, thinking it through at the prototype/experimental/one-off level which most of the stuff on HAD is.
Spend $10 more on the production quality of your electronics. This is a case of shabby soldering. The previous case was bad wiring. If you are going to build your fire prevention system in the same shabby way, you don’t get extra safety, you get an extra fire risk.
So the moral of the story is to have a $300 printer and a $200 fire extinguisher?
To protect your $200,000 house.
Fire prevention and detection is so important to any maker or hacker and often overlooked. I learnt this the hard way only last week when my entire workshop burnt to the ground :(
https://youtu.be/e7JThkfpNMU
Having a smoke alarm and fire extinguishers just wasn’t enough. This is where IoT really may have saved the day, or at least some of my kit.
That’s really sad. But it does not really need to be an IoT solution, with questionable reliability and less than questionable data security. Radio connected smoke alarms would already be very good. Also the other way round, if you are in your – hopefully soon rebuilt – workshop and forgot something in the kitchen. I would not want to depend my life on the internet connection to a cloud server and such a server itself.
The fire code is a lot like plane crashes. Better one crash at a time.
I personally have traded off the “ease” of building my printers using the quick disconnect type connectors and just direct-soldering wires to the board, especially after watching my power input connectors melt on my first start-up.
The amount of time saved assembling really isn’t worth melted/shorted connections.
Connectors with proper current rating were no option? directly soldering stranded wire into a PCB is not 100% long term reliable under vibration conditions. The strands tend to break at the point were the solder ends.
Off the shelf solution: z-wave smoke alarm and a z-wave plug. Smoke triggers the alarm and you have the switch shut off. Won’t put out a fire but will kill power to something that might be smoldering. Set it up to send you an email too.
Usually one of the first things that I do when I am working on my printer is to use an external mosfet or ssr for the heated bed. This means that the current carrying capacity of the wire from the power supply to the bed is greater than what the rating on the RAMPS board can handle. (it also lets me use 24v supplies for the hot bed and a 12v supply for everything else.) The funny thing is that the RAMPS board is rated for 16A total input current. Small design changes over time have been pushing the limits of the connectors on the pcb.
I would say that using caution and that adding external switching would go a long way to making things better.
Spot on! And the SSR has nice bulky connectors that aren’t prone to failure/heating/fire due to high resistance connections providing the wiring is done properly.
http://www.futurlec.com/Pictures/SSRDC50V80A.jpg
They’re cheap as chips on ebay.
This would make a nice retrofit kit to take the problematic heater connector out of the equation and turn it into a low current control signal. You’ve convinced me to do this.
The other advantage as that a burnt out $10 SSR is much cheaper to replace than a controller board. Just make sure it has a heat sink or put it inside an alloy box and heat sink paste it to the box inside.
Also – quality copper crush (crimp) connect eyelets/connectors will give you a sound connection. You shouldn’t solder high current connections that have a low resistance load because the resistance of the solder can be more than the resistance of the heater so guess where will heat up first and then hot solder is even worse – thermal run-away.
So CRUSH copper wire so that you get the maximum contact surface area between the copper wire and the connector.
I used to leave my printrbot unattended regular doing overnight prints, but then one day I was eating lunch and smelt a smell from the kitchen and ran into the office to see it on fire.
To add the bed heater, I’d done what I thought was the proper way rather than putting all the current through those tiny bed heater connectors and added a SSR relay board from a reprap open hardware supplier, and it had failed. I’d soldered the bed wiring to the ssr because I didn’t even want the screw terminal connections on the board involved as I felt they’d work loose through vibration, so this is not a bed element connector fastener came loose with mechanical movement story.
My guess is the regulator overheated because it was under load and out of the cooling airflow tucked up inside the printrbot’s base and some emails to the seller suggested the same, however he also said it was supposed to be heatsinked also, which isnt mentioned anywhere in their literature as I went back and checked. I got the feeling that the seller wasn’t interested in any autopsy or diagnosis, which was a shame as I wasn’t interested in ambulance chasing or getting a free unit (I actually ordered a replacement from the same person just before this conversation).
I’ve rebuilt the bot, replaced the wiring that singed and now I have the same ssr board, but I have it inside a diecast metal box sealed in, using the box as a heatsink also so its passively cooled without being a fire risk and even if it lets go, its going to be contained in that box. Its holding up this time. Since then all has been fine but I no longer print unattended for long periods or overnight, and lets be clear this is something I added to the printrbot, the bot itself has superb build quality. But I see others recommending the ssr boards as the fix all without a caveat that they too can fail.
http://gallery.pipandphil.com/d/38908-1/relay_fire1.jpg
I would be curious to know what temp he set the bed at before it caught fire. I have this same printed and have not had any issues – but I have also read the manual and know the max bed temp (100C). The printer allows the Gcode to set the bed temp – as high as you want – but the printer only supports (per the manual) 100C max.
I was preheating for PETG @ 240/80 (layer 2 i go down to 70), but the printer only got to 146/45 before the smoke.
Ahh, I get it….. An article about someone getting what they pay for…. Wouldn’t be surprised if the traces used in the bed circuit are rated for far less than the expected current.
Did the printer Halt before Catching Fire?
Or did they even try to turn it off and back on again … “IT Crowd”
spark arcing on board at 00:22-00:24 in the video.
I think the fire would have a hard time getting out of the control box – sure if you had papers lying on top of it or something else – i think it would of just continued to melt the board – most kit 3d printer don’t have an enclosure for the control board like this one does. If you put the printer on Ebay – let me know.
Heaters of all sorts malfunction eventually. New from manufacturer is not proof of functionality. Nature of the beast. Ammeter. USE it. Relying on poly fuses is not a good idea either. Use a REAL fuse rated the same as heater element or some reasonable over. In-rush(start) currents may exceed operating. REAL Fuse described as a non resettable throw away element that once exceeded; physically disconnects the load. Thermal fuse is a good safety too ,but those are typically a single point generality or overall temperature dependancy. Amps detection best- so far. A good datasheet on the heating element goes farthest. Knowing that the FLAT element that is sent to you folded like an accordian in over stuffed tiny envelope is probably not useable- helps. Especially when manufacturer states on item “DO NOT BEND”. I recently watched a vid of some crazy bending element from kitchen stove. Please dont. Please dont copy that. 240VAC at 40 plus Amps is seriously explosively dangerous. Experience is NOT always the best teacher.
That would take some serious effort to bend the heating element in an electric stove.
Stove top or ‘range’ Squared ceramic are harder to bend but not that much more difficult.. Hence the metal support under them. I have NO intention of linking that crap, but if u want look under ‘ aluminum’ about 4 days ago. It was one of oven parts fwiw. I maintain “DON’T.”
he is right, PhotonicInduction overpowered JUST such an element and it litterally EXPLODED with sparks, it was so bad it blew TWO holes on oppisite sides AT THE SAME TIME, and this “time” was a quick pop
Even the Anet A8 has the same problem. On facebook group 2 users posted pictures of burned printer.
I wish I had found your article before my son bought his 3D printer. It was one of those “cheap as chips” (Alunar Desktop) kits. He spent about $300 and carefully constructed it. He completed some cool projects (a combination Lenin-lemon was his pride and joy). Two months after purchasing it, he was running a model of Mount Everest for a school project. He started it the night before, watching it for about 2 hours to make sure that the extruder was working. I woke up at 4:00 a.m. to the smell of something electrical (no smoke detectors). When I opened the door to the basement, where his printer was set up, I was hit with a wall of smoke (that’s when the smoke detectors sounded). We got out of the house and called the fire department. They informed us that a mere 15 minutes later we would have lost our entire house. My son said that he thought that the warming plate seemed a little hot, and in hindsight he thinks that was what cause the meltdown of the printer, leading to the desk to catch fire, the upholstered desk chair, the wood joist above the desk…. Our insurance company is filing a lawsuit against Amazon and the manufacturer. I’ll post again if there is an outcome.
All this talk of firmware fixes to prevent your printer catching fire is only part of the solution. What if a FET fails on? No amount of firmware will fix this…