Overheating heated bed mosfet

Hi people,

I've finally got a working printer, and getting some good prints off in ABS, as long as they're small. Larger prints are warping badly. Obviously I need a heated bed.

I've got a MK II PCB heated bed, but when I've hooked it up the bed mosfet unfortunately melted. After this I measured the resistance of the bed to be a little over 1.5 ohm, so sounds like a lot of current going through the mosfet.

So... I could put a heatsink on the mosfet, but the sanguinololu boards leave very little room for a decent heatsink, can anyone recommend one that would be enough to keep the mosfet from melting but would fit on a sanguinololu board?

Or, and I may be talking out of my posterior, would using PWM on the bed keep the mosfet from overheating, and just take longer to heat up the bed? If so, is this built into sprinter or any other firmware? I've seen in sprinter the option HEATER_CURRENT to limit current to the nozzle, but nothing similar for the bed.

Many thanks,

Darren.

The MOSFETs on the Sanguinololu are not the correct type for direct logic drive. If you use a MOSFET with a lower gate threashold and a low RDSon then you can switch the bed current without a heatsink. I think the one used on RAMPS is more suitable.

---

[www.hydraraptor.blogspot.com]

Thanks for that.

Is it just a straight swap? I'll get some ordered from RS and give it a try. Is it worth swapping the hotend one too? Its been a looong time since I did any major electronics!

Whilst I'm here, fantastic blog. Been reading it for a few months now, very handy! I'm not that far away from you actually, just in Lancaster.

Cheers,

Darren.

[www.22balmoralroad.net]

Yes it is a drop in replacement.

The power dissipation is proportional to the current squared so the hot end one is not stressed so should be OK.

Actually I was mistaken when I said they were not logic drive ones, I must be confusing them with Gen7. They are logic drive, but the ramps ones have a lower RDSon.

---

[www.hydraraptor.blogspot.com]

Fantastic, that worked a treat. The MOSFET is still getting rather hot, but at least it isn't melting. I've also had to get a new power supply, the one I was using wasn't enough to drive the steppers, hotend, and bed!

Getting some nice quality prints now. Time to try something big!

Edited 1 time(s). Last edit at 05/01/2012 03:17PM by darrenp.

Just had my mosfet unsolder itself on the Generation 7 board. It was a P55NF060.
Driving the MK1 heated bed with a 18 volt supply.

what is the MOSFET that was recommended?
Is it the RFP30N06LE ?

Dwayne

Looking at digikey, I see the RFP30N06LE was discontinuation on14/Mar/2011.

Gen7 uses the IRFZ44N, which works well. Other recommendations are the more expensive IRL 3808.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Hi, I'm fairly new to this, with very little electronics experience; so some advice would be greatly welcomed.

I've just finished my Prusa build and I'm starting my initial tests. I have a Sanguinololu 1.3a (with ATmega1284p running Sprinter) with a Mk2 heatbed.

I had tested my motors, my heatbed at 40oC and hotend at 50oC; just to be sure that they were heating and reading temperature, and had just uploaded my first print from Pronterface. It appears that on the initial heat up, I've managed to burn out the heatbed MOSFET, possibly my fault for not having a fan running or possibly something else.

Reading the earlier posts, I went looking for the MOSFET part number used in the RAMPS board as suggested by Nophead but it appears to be the same part number that I'm already using (STP55NF06L); which makes me think something else may be wrong.

My headbed appears to measure 0.85ohms at room temperature.

Is the STP55NF06L the correct MOSFET to re-order? Is there anything else I should be checking for before I just swap a new one back in? Would the IRFZ44N part from the Gen 7 be more appropriate?

Cheers.

P = V^2 / R/ for 12 volts this is 144 / 0.85 = 169.412 watts. This is O.K., BUT...
V = I * R; or I = V / R, so I = 12 / 0.85 = 14.12 amps which is a bit high!

---

Bob Morrison
Wörth am Rhein, Germany
"Luke, use the source!"
BLOG - PHOTOS - Thingiverse

Ok, so the current draw required for the heatbed is too high; hence the nuked MOSFET.

I'll order some new MOSFETs, but I'll need to increase the heatbed's resistence. Any suggestions?

My initial query would be, can this be achieved by putting a resistor in-line; or should I look at cutting some of the tracks on the heatbed? I don't intend to ever print directly on the silkscreen; so a smooth surface isn't essential. Or should I just go back to the original seller? The heatbed was from Snipermand an eBay seller from Denmark.

Cheers

The easiest solution is to mount the MOSFET on a heatsink.

---

[www.hydraraptor.blogspot.com]

Oh, there is no heatsink? Nophead is right, you want one. Even if it's only a sheet of aluminium.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Ok, I'll look at a heatsink. I have a couple of small GPU sinks lying around that could be modified. Can both MOSFETs be attached to the same (larger) heatsink, or will that cause electrical issues?

No, not unless you use insulating washers as the tab is the drain connection.

The extruder heater MOSFET does not need a heatsink as the temperature rise is proportional to current squared.

---

[www.hydraraptor.blogspot.com]

Perfect, thanks

Beware that the connectors on the Sanguninolu are not rated for anything like 14A. I would take the gate signal from there but wire the bed directly to the power supply and the drain of the MOSFET and take the source directly back to the PSU.

Also you need the pretty thick wire for 14A. I use 13A mains cable. Fortunately it will only take that current when cold and will soon drop to around 10A as it warms up.

If you use a PC PSU I would recommend using at least 3 pairs yellow and black wires as a single one gets warm with only 5A.

---

[www.hydraraptor.blogspot.com]

Ok, I'm currently using 1.5mm heat resistant flex, so it should hopefully be enough for the current; but you lose me a bit on the circuit description.

I'm using an old 350W PC power supply which is rated for about 26A on the 12V line.

I'm assuming that I need to run some wire from the main ATX motherboard connector to a connector block? I have a 15amp block available, but I'm not sure how I'd be controlling the bed. Do I need some form of relay?

Sorry, this is the bit of electronics I'm still learning.

Yes 1.5mm^2 wire is sufficient.

When the MOSFET is on the PCB all the current has to flow through the 12V and GND connector and the PCB tracks and they are not rated for it.

The source pin of the MOSFET goes to ground so rather than connecting it to the PCB I would connect it to ground on the PSU. One of the bed connector pins goes to 12V, but rather than connecting the bed to that I would wire it direct to the PSU. The other wire of the bed connects to the drain of the MOSFET, so again I would wire it direct rather then through the PCB.

For commoning the PSU outputs, you could use a connector block but I just solder them together in bunches. The cheap PSU I have has got really thin wires that can't even carry 5A individually without getting hot. Even when using all of them I think the current rating of the PSU is higher than the wires!

---

[www.hydraraptor.blogspot.com]

Ok, I'm going to dumb this down a little further; just so I'm straight on this and before I nuke the Sanginololu as well. The connections I believe I should be making are:

ATX +12V --> Heatbed in
Heatbed out --> MOSFET Drain Ring
MOSFET Source Pin --> ATX Ground

The other two MOSFET pins should remain connected to the Sanguinololu PCB? I'm assuming that the Sanguinololou will apply a voltage to the gate pin, which in turn switches on the heat-bed.

The poor old MOSFET is still dealing with ~14A so I'll still need a (substantial) heatsink.

Sorry for the blithering idiot level of understanding, I just can't seem to get my head round how a MOSFET works; despite reading the Wikipedia article...

Looks like I need to make a new RS Order

T.

Almost right. The middle pin of the MOSFET is the drain and is connected to the tab so there is only the gate remaining to connect to the PCB.

---

[www.hydraraptor.blogspot.com]

Ok, so one pin to re-solder to the PCB, one to cut off (to open up some space and avoid confusion); as we're going to use the ring connector, and one to connect to the ATX ground.

I may also have to look at some form of strain relief to ensure that the remaining pin doesn't exert too much force on the PCB. I have the PCB mounted on some ply; so I'll probably just attach some form of support to the heat-sink.

Thanks for all your help. I'll report back once I have it working.

Back with a quick update. Mounting the MOSFET on a heat sink and changing the wiring has done the trick; I've just spent the last week trying to get it to print; most of what I was getting was just garbage. I even managed to melt my first NTC Thermistor and had to scrape the remains off the J-Head; thankfully I had a spare

For the heat sink, I cut up an old Aluminium CPU heat sink which had a tab at one side for mounting it to the motherboard. I was able to tap it and screw the MOSFET and ring connector to it. Due to the potential of this heat sink carrying current I picked up and modified a small ABS box to put the Sanguinololu in and fitted an 80mm fan and controller to it. As far as I can tell, the MOSFET never gets above room temperature now; and my Pololus stay nice and frosty.

During the desoldering I did manage to damage the trace and pad for the gate pin; I'm afraid my desoldering skills leave much to be desired, however I was able to follow the trace and solder a breadboard jumper lead between the MOSFET and the next component. I also had to remove the 4 pin heatbed connector to get enough clearance for the crimp; however that removed cleanly when I attempted to desolder it; I sort of wish I'd done that before the MOSFET as it would have given me more clearance to remove the MOSFET.

Once again, Thanks for all your help, and I hope this helps someone else as well.

Happy printing

T.

Hi!

First of all, thank you so much for your contributions in this forum: reading your posts i've just avoided many errors and gained a lot of ideas for building my Prusa i2!!!

But back to the topic. Thanks to Taint's question i learned the danger in connecting the heating bed (HB from now) directly to sanguinololu, and thanks to Nophead explanation now i know how to connect the mosfet directly to my ATX power supply.

As suggested, i removed the 4-pin HB connector, and used an ALU heatsink for the mosfet. This heatsink touches the pcb surface: this is intentional, to sustain the mosfet connected to pcb only by the gate.

And this is also the problem: how can i connect the mosfet source directly to PSU? Due to the mosfet orientation, the heatsink closes the way to the outside of PCB, and there's not enough space to solder a 1,5 mm wire in the space between mosfet and pololus (i'm afraid i could break the source pin if i try to bend it).

I admit i have no brilliant ideas (well i have no ideas at all).

Any clue? Can you explain how you solved the problem, or - better - can you post some photos of your solutions?

Thanks again,
M4k

Hi Merlino4000,

Sorry for the delayed response; I've been a bit busy the last couple of nights.

I managed to get back to the printer tonight and I've taken a couple of photos as requested; I'll post the links below.

I found that the easiest way to connect the source was to crimp the wire onto the MOSFET before soldering it to the Sanguinololu. I cut off the drain pin to give me a little space; as I was connecting the drain via a ring connector.

It did make it a little cumbersome to reinstall the MOSFET, but as you can see I went a little overboard with the heatsink.

Oddly enough I've just managed to burn out my first ATX power supply mid-print, but then it was a VERY old PSU

Let me know if you need any more info about my bodge

T.

Picture 1
Picture 2
Picture 3

Hi, Taint.

First of all, thank you for your reply: thanks to your questions and answers, I've managed not to burn my Sanguinololu, for which i think i'm in debt with you ))

I saw your photos: your realization it's exactly what i understood from your explanation. I realized something similar, not as neat as you did but functional at least (i hope so).

BTW, the heatsink i used is smaller than yours, so i'm a bit scared (i don't know the temperatures the mosfet can reach, i hope i haven't underestimated the amount of heat to dissipate). What do you think about placing a fan over the mosfets? in your opinion is it possible to drive the current directly from the sanguino?

I found an article on thingiverse (http://www.thingiverse.com/thing:22202) about something similar, but it's more focused on driving an extruder fan...

But enough with the questions: thanks to you i'm a little closer to my first print, so thank you again for your help.

Cheers,
M4k

Hi M4k,

Apologies again for the delayed response.

I suspect you're unlikely to need such a large heatsink; as was discussed above, my heatbed seems to have particularly low resistance and it does generate quite a bit more MOSFET heat than normal. I also supplemented my heatsink with an 80mm case fan; which I attached to the top of the ABS box I mounted my Sanguinololu in. Mounting the Sanguinololu was a safety decision; as I'm using a plain aluminium heatsink there's a good chance it could be carrying current and accidentally touching it could be quite nasty.

I connected the fan to a Corsair PC fan controller; which was included with a PC case I bought a few years ago. I'm only using the controller because I'm using a particularly noisy 80mm fan and it allows me to adjust the fan speed. That was then connected to one of the 4 pin hard disk connectors on the PC power supply I'm using. The connectors are there, they're spare and it has the bonus that it switches off when you power off the printer. I probably wouldn't recommend driving a fan of that size directly from the Sanguinololu; plus unlike a hot-end fan, you can effectively run it continuously with no detrimental effect on the print quality.

I hope this helps,

T.

Boy - wish I'd read this thread sooner - would have saved me some time. So wifey decided she wanted a 3D printer - hey, that's why I married her - so I decided to pick up a "pre-build & calibrated" Mendel Prusa. Went to pick it up, and the seller toasted the motor controllers, which should have been a sign to walk the heck away, but he manged to get it "fixed", printed some stuff, and the wife had a smile on her face. FIgured I could sort out the electronics, as the mechanics looked solid.

Got it home, and cranked it up, and started having problems - mainly that nasty-smelling magic smoke was coming out of the ATMega, and that everything on the board was hot as crap. Called the guy that sold it to us, and told him I was sending it back (the Sanguinololu), and expected a new one post haste. Wifey was not smiling, so I ordered a Sanguinololu kit and some Pololu's - figured a fresh start was called for.

Put the board together and started calculating the heat loading, and was not happy. Looked at the board, and I agree with Nophead's premise that the traces are too small for the MOSFET's, the pads are super marginal at best - so 100% agree with him that the MOSFET for the bed does NOT belong on the Sanguinololu.

I've also "fixed" my MKII heatbed - as I find out Nophead has reccommended - as the original was inadequately wired, and the pads are not even marginal, they were useless. Measured the resistance, and realized I'd be sinking ! 11amps on startup of the heatbed - through a unsink'ed T-220 - bad idea.

Anywho, to stop rambling - couple of observations -

1] I'm using a 13.89v, 20amp supply. I'm using IRFZ44's MOSFET's simply because they are over-rated for my load, and I have a few laying around as spares.

2] Heatbed is 1.3 ohms, so full load will be 148watts and 10.68 amps max load.

3] Extruder heater is 8.1 ohms, so full load will be 23.82 watts, 1.72 amps max load.

4] Power supply is fine for what I need.

Reference says a T-220's thermal resistance to air is 62 °C/Watt, so heatbed MOSFET is going to require a decent heatsink, and extruder could survive without.

I used a 3-pin Molex KK series to replace the bed MOSFET on the board, and didn't insall the "bed heater" 4-pin. I connected ground and gate, so I'd make my life easy for testing.


The above said, I decided to put sinks on the 7805 and the IRFZ44 for the extruder I've left on the board. I had a couple of slim profile T-220 sinks in the bin - FISCHER ELEKTRONIK FK 220 SA 220 - their thermal resistance is 25°C/W. They need to have a bit of modification to clear other bits on the board. These are cheap as chips sinks, so easy to shape with pliers. Order an extra one if you've never played with these before, they are super brittle.


I fitted these to both the 7805 and the extruder MOSFET using a bit of thermal paste.


And for the bed mosfet, I wanted to get to at least 5°C/W, so went with a MULTICOMP MC33266 rated at 3.6°C/W - a bit overkill, but hey, was cheap, in stock, and looks cool ! ;-). I did not use the "lug" method suggested above - with the MOSFET off the board, there's no compelling reason I could think of - and it would have complicated my wiring plans. Used some adequate wire, and a higher temp on my iron to make sure I had a good solder joint. I didn't take any pictures, but if you don't cut the MOSFET legs, you get a good long surface to solder your wires to. I used some leftover endstop wire for the gate. I popped some heat shrink on it to provide strain relief, and called it ready to test. You can see "version 1" from last night sitting looking abandoned! :-)


Test jig was an Uno connected to the gate, listening for me via serial to turn the gate on / off.



Ambient temperature was 22C.



+ ~ 8 seconds, 51C on the tab where it connects to the sink.


+60 seconds, temperature on the tab/sink connection stablizes at ~ 80c.


+60 seconds, silicon is stable at 60C, so life is good.


So all in all, I'm happy - will mount the sink somewhere out of the way, and keep an eye on it - should not need active cooling, and the sink itself is barely touching 75C, so should be able to have the wifey make a nice little bracket for it.


Finally - and the subject of another rant - the 5vdc line from the 7805 is tied to the USB supply - another potentially "bad thing".. Nophead seems to have noticed this as well - rather than cut the trace on the board, I hacked up a USB cable and cut the 5vdc supply. I've got a couple of surface mount MOSFET's on the way, and will make a proper disconnect for my cable so I can have a safe supply regardless..

Your post has concerned me. I've just been finishing off the rest of my setup and decided to use some aluminium W channel as a heatsink for each MOSFET and the regulator. I just cut a ~3cm length for each one and use some kapton tape (No thermal paste) / nylon bolts to secure and insulate them. I indended to use 12 AWG cable and connect one end to the MOSFET tab and the other would be powered through the board. (I've put cable along the tracks from the power connector)

I thought that would be okay (Not tried it yet) but after looking at the temperature yours gets to with a much larger heatsink, I'm rather concerned. I might have to go mounting them separately or a larger heatsink and switching out the MOSFETs with better ones. Maybe I can get away with a fan instead.

Edited 1 time(s). Last edit at 08/31/2012 07:52AM by yngndrw.

I use an IRF 2804 mosfet. It works rather well.