I’ve just built a new Delta 3D printer and I have a weird issue with the heating bed. I keep blowing mosfets, they won’t last more than 30 seconds. The main board is a Ramps, and I’ve changed the factory mosfet with an IRLB3034PBF after it blew up. But it didn’t seem to solve the issue since I just blew 3 of them on my last 3 attemps, even after adding a nice heatsink.
I really wonder why, I have the exact same heating bed on another delta printer, but this one is driven by an old mksbase and it works fine since the last 4 years…
I’ve checked the resistance of the heating bed, it is the exact same as the other one, so there is no short in the heating bed. And no apparent short elsewhere either.
Even the main power wires gets hot, I’ve doubled them and removed the main connector so they are directly soldered to the board.
Sort of stream of consciousness troubleshooting here.
If the wires are getting hot then the circuit is somehow shorted in the bed or at the bed.
How many terminals on the bed? Are they mis-labeled, i.e., instead of 1 positive and 2 negative, you’ve got two positive and one negative, such that when you wire it up you are completing the circuit? Or does the bed get hot while you are releasing the transistor magic smoke?
If it’s not that, then check the bed and make sure the surface is clean - dirty connections are bad and there could be some conductive surface contamination that requires some voltage (higher than an ohm-meter) to breakdown.
I thought it might be shorted somewhere, but the resistance measures I took look totally fine. I’ve got about 1.4 ohm per 12V coils, and 2.8ohm for the 24V coil, which is exactly the same as the other one I have, which works fine.
The bed heats normally, it’s just that the mosfets and wires get hot very fast. The wires are of decent gauge and I use two of them per phase.
The bed I’m using is very similar to this one, you can see different wiring schemes depending if you’re using it on 12 or 24V:
I tried the two wiring schemes they proposed, for 12V and for 24V, but I only have a 12V power supply so both runs were at 12V. The PSU is a 12V 30Amps I think.
On 12V the wires get hot in a few seconds. On 24V the wires stay cold. But in both cases the mosfet heats and burns. On 12V wiring mode the bed heats at a normal speed (1-2 minutes to reach 60C), on 24V wiring it takes forever, maybe 20 minutes, the mosfets usually never makes it to this point.
On the other printer who works fine, I’m using a 15V PSU. Aside from the motherboard (Ramps 1.4 vs mksbase) those are the only differences I can think of.
I’m not very familiar with mosfets. I’ve used the ones that were recommended on this instructable:
1.4Ohms @12V is 8.5A. That’s not too much. With the datasheet linked in that instructible, it should be generating about 0.1W. 12V on the 2.8Ohm would be half that current, 1/4 of the wattage. But that’s not making much sense. Do you have a heat sink on the mosfet?
Have you measured the voltage or the current with a multimeter?
The reason I asked about a diode is if there was something like a zener limiting the voltage, it would short the PSU to drop the voltage. I would guess the 24V side wouldn’t have that problem.
Btw, I tried that site, but I can’t read it, amd when I randomly clicked on stuff, it brought me to the front page. Sorry for my ignorance.
I came to the same conclusions as yours, but somehow it still burns. This doesn’t makes sense, I’m a bit lost. I think I’ll first try to put a mechanical relay between the fet and the heating bed, just to see if the relay burns too.
Not yet, my setup wasn’t convenient to do that, I didn’t have much access. I’ll do it this evening.
I didn’t put any diode anywhere, I’m not sure if you’re talking about a diode that would be on the ramps by default?
Aw, crap, I guess it is only accessible in China now. Used to be accessible last year but the Internet is getting more and more compartimented between countries nowadays, it sucks.
The heat sink should be electrically isolated. It is common with one of the pins (I think it’s usually the drain, but I can’t find anywhere to confirm that).
You’ve got this between the ground and the heatbed, right? With 12V from the heatbed going straight to the psu?
I’d be curious to see the current, but my meter has a 10A fuse. If yours is like mine, it probably wouldn’t be a good idea to use that. You could try to measure the resistance of one of the cables (when it’s disconnected) and measure the voltage drop and do the math to find the current that way. Then you run into the issue of finding the resistance of a cable, which is pretty small… I’m not sure what I’m expecting…
Your PSU shouldn’t even be able to supply enough current to destroy that mosfet.
There’s always monitoring the voltage across the heated bed to figure out the current. Or, put a 1 ohm power resistor in series with the heated bed and then monitor the voltage across it.
Do you have a meter the dielectric strength or a diode check of the bed (measure in both polarities)?
I didn’t put insulation but since the heatsink is not touching anything I doubt it would be a problem.
Yes, at least I think so. I’ve just replaced the original FET with the new one in the exact same orientation. 12V going straigt to the PSU, everything soldered instead of using the crappy green connector.
Yeah, that’s also the fear I have. I’m a bit afraid to blow up my multimeter if it goes over 10 amps. It 's supposed to be fused, but still, it is just an el cheapo multimeter (but I still love it, it is surprizingly accurate and has a lot of neat little functions)
Yes, exactly. That’s one of the things I find super weird too.
Once the FET is fully on it should have a negligible resistance, so unless there is some really insane currents going through it (this one is rated at 195Amps!) it shouldn’t heat at all. But given the fact that this power supply is rated at 30 amps, I don’t see how it could come even close to 195Amps…
I suspect that it might not be fully on, resistance is usually a lot higher when FETs are doing the swiching or whenever they are semi-opened. That’s the only way I think it could get so hot so fast.
I’m not sure to understand. I’ve checked the bed resistance, but I didn’t really pay attention which way. I can try both, but I’m pretty sure it will remain the same.
You can measure the voltage across it, or the voltage across the bed. One of the comments in the instructible suggested that mosfet might not be easy to turn all the way on. That would make some sense, at least.
My thinking on measuring the bed dielectric strength is looking for a possible reduced resistance somewhere in the board when voltage is applied across the terminals. The diode check is for a similar reason - I don’t know if there are zeners or surge suppressors built in to the board to help avoid over-voltage conditions.
Even if the FET is switching slowly, it does not really explain why the wires get hot. It may explain why the 24V configuration works the way it does, though, with the bed taking a long time to warm up.
I’m guessing the problem is in your RAMPS, not in the heat bed. Something is providing a low resistance path from power to ground on the head bed output path, and if it’s not the bed…
Well, I just tried to plug the bed directly to the 12V with my current meter, not using the ramps.
The wires got hot in a few seconds. My multimeter is good for 20 amps and apparently it seems like the current was over its limit, the last reading I got was more than 9 amps.
I really don’t understand what is going on with this bed, the resistance values were lower today than they were yesterday, it was reading around 0.9 ohms per coil. I just realize I forgot to read the voltage, but I bet there is quite a drop since I was hearing the printer fan seriously slowing down when the heat bed was activated.
So I guess the bed is somehow defective. That’s extremely weird. I don’t understand how this is even possible. I could understand if the resistance went higher or if there were no continuity, but in this case it’s the opposite and I cannot find any short anywhere with the aluminum frame of the plate (it’s some kind of PCB emedded in a thick aluminum sheet). There is no continuity between any resistor and the aluminum.
I use this car fuse idea to give a little more protection to my MOSFETs. You can see from one of the remixes of this that I put it under my Delta printer heatbed mount.
I made some progress yesterday. Apparently the Ramps was dead, I guess something fried in the PWM driver when the first mosfet died (my guess would be that the gate resistor got too hot because it was really close to the fet, this probably affected its value too much), so that might explain why the other mosfets fried as soon as I replaced them.
So I changed for an other ramps. This one seems to have a better mosfet, which didn’t heat too much. But on this one, the issue seems to come from the polyfuse. At some point, the polyfuse gets too hot and shuts down the heat bed.
I tried first to use the compressor to blow air on it to cool things down, which worked perfectly. So that narrowed things down to a heat problem only, good.
Next step, I’ve put a fan on the ramps, which did make things a bit better than nothing, but I still could not evacuate enough heat to go from 0 to 60 degrees without the ramps shutting down (at least not when using the heatbed+fans+nozzle+motors at the same time). But once it eventually reaches 60 degrees, it can keep this forever without any problem since it is not constantly ON anymore.
So, in the end, I think the issue is just that this heat bed needs too much power and that ramps board are crappy, there seems to be nothing really wrong with the heatbed itself. So I guess I’ll just buy a SSR or an external mosfet to drive it and call it a day.
I’ll also need to find beefier wires, since the ones I’m using get pretty warm quickly.
Thanks for your input guys, I think the problem is solved now
