Regulator for heatbed (3d-printer)

[LarsT]

I've made an attempt at making a pwm-regulator for heater plate using a 555 timer and an N-channel mosfet (30N06L), but whenever I turn it on the mosfet quickly warms up and within half a minute it's close to 150C, even with a small heatsink attached, and I simply don't understand why. The timer is set up as astable with a frequency of approx 10Hz with a duty cycle from 5-95%. I've looked at the gate- pulse with an oscilloscope and it shows vertical edges up to 11.4V, no problems there. The heatbed is approx 1.4ohm when cold.

From what I can find from the datasheet on the mosfet it should be able to handle 8amps with 11V at gate just fine. What am I missing here?

[Kleinstein]

It might be a good idea to have a series resistor (e.g. 20-100 Ohms)  in the gate line. With some bad luck the MOSFET might oscillate at a very high frequency without the resistor.

[LarsT]

I know, but I don't think this is causing the problem. It seems the pulses are fine, and no oscillation present.

[georges80]

Quick & rough calc:

8A and 0.030 Vds means if you have 100% duty cycle (on all the time) then:
8A * 8A * 0.030 ohms = 2W dissipation in the FET.

How 'small' is your small heatsink...

cheers,
george.

[LarsT]

Quick & rough calc:

8A and 0.030 Vds means if you have 100% duty cycle (on all the time) then:
8A * 8A * 0.030 ohms = 2W dissipation in the FET.

How 'small' is your small heatsink...

cheers,
george.

I don't have a datasheet on the heatsink, but it's aluminum and the size of a stamp. I will try to find a bigger heatsink.

I did some rough calculations before building the circuit and got 2.2W of dissipated power (Rds=0.035, 8A). The mosfet's datasheet specifies a dissipation capacity of 62.5C/W, which should correspond to (175-25)/62.5 = 2.4W.

That's probably too close even with a (small) heatsink attached. Perhaps a relay would be a better choice for this? I might even ditch the 555 timer and use a microcontroller with temperature sensing and pid-regulation.

[mikerj]

I know, but I don't think this is causing the problem. It seems the pulses are fine, and no oscillation present.

Have you actually measured the rise/fall times of the gate drive?  These need to be fast to minimise power dissipation in the MOSFET.

[richard.cs]

It needs a heatsink for long-term reliability, at 2 W the Rth gives a junction temperature of 150 C which might be in-spec but it's won't have a long or happy life. As mikerj says you also need to check that you are switching fast enough to avoid much higher power losses than you expect. Once you've done that you either give it a large enough heatsink to keep the temperature acceptable with your calculated losses (I would aim for 30 C/W or better) or you could change to a different MOSFET with a lower on resistance.

[Z80]

The easiest fix is to use a better mosfet.  You can get jellybean devices with Rds(on) an order of magnitude lower than the one you have.  Yours will be dissipating approx 2W at 100% duty cycle. The junction to ambient thermal resistance is quoted as 62.5oC/W so 125oC temp rise above ambient with no heatsink.  You also need to be aware that as the temp goes up, so does the Rds(on) and according to the data sheet on that device, by the time it reaches 150oC it has doubled so you now have almost 4W dissipation.

[Richard Crowley]

It is not clear to me why something with any kind of thermal mass needs a frequency as fast as 10 Hz?  Is that common for 3D printer heatbeds?

[thm_w]

It is not clear to me why something with any kind of thermal mass needs a frequency as fast as 10 Hz?  Is that common for 3D printer heatbeds?

Marlin is using ~8Hz if you use PID bed control. Frequency <1kHz shouldn't really be an issue unless you are using relays though, right?
For relay controlled beds, you can enable slower modulation (~0.1Hz).

[LarsT]

I know, but I don't think this is causing the problem. It seems the pulses are fine, and no oscillation present.

Have you actually measured the rise/fall times of the gate drive?  These need to be fast to minimise power dissipation in the MOSFET.

The regulator runs at 10Hz and looking at the graph in "millisecond resolution" and I can't recall seeing any slope at all. I can meaure it to be sure, but I'm pretty confident that's not the cause, at least not for such a low frequency.

[LarsT]

It is not clear to me why something with any kind of thermal mass needs a frequency as fast as 10 Hz?  Is that common for 3D printer heatbeds?

I was aiming for 1Hz and 10uF felt just about right when picking parts and then I stuck with it.

[LarsT]

The easiest fix is to use a better mosfet.  You can get jellybean devices with Rds(on) an order of magnitude lower than the one you have.  Yours will be dissipating approx 2W at 100% duty cycle. The junction to ambient thermal resistance is quoted as 62.5oC/W so 125oC temp rise above ambient with no heatsink.  You also need to be aware that as the temp goes up, so does the Rds(on) and according to the data sheet on that device, by the time it reaches 150oC it has doubled so you now have almost 4W dissipation.

I forgot to take the temperature rise into account, but it makes sense now.

Looking around a little I see the mosfets FDP8870 and IRLB8743 recommended several places. Both mosfets have Rds below 4mOhm. I guess they're both better choices.

[LarsT]

How about using two (or more) mosfets in parallell?