Here’s how I went about installing the heat-bed sold by Marlark.
I was going to use a new 15 V supply and remove the stock one.
Using Kelvin measurement I got .645 Ohm between the +12 V terminal and minus on the heat-bed. So the 0.65 figure quoted is correct. 15 V and 0.65 ohm will draw 23 A so the power supply needs 345 W for the heat-bed alone. 15 Volt is quite low and voltage drops will be noticeable in the time required to reach the temperature set point. In order to minimize the drops I used 4 mm^2 cables.
For switching I decided to use a low Rdson MOSFET. You can go the DC SSR route but unless you want an alibay Fotek clone/fake/uprated one with unknown on resistance and no datasheet and a yeah-right max amp rating, prepare to spend more on it than the heat-bed.
After looking at some MOSFETs, I decided on IRF1324PbF (http://www.irf.com/product-info/datasheets/data/irf1324pbf.pdf).
Worst case calculations:
Rdsonmax = Pdmax / IDmax^2 = 300 / 195^2 = 7.89 mohm.
RthetaJ = 62 K/W.
Max power dissipation at roomtemp (25 degrees C): Pmax = (Tjmax-Troom)/RthetaJA = 2.42 W (no surprise there, it’s a TO-220).
Max current at roomtemp: I = sqrt(Pmax/Rdsonmax) = 17.5 A.
Now, since we need it to be able to carry 23 A continuously we have to heat-sink the sucker so it can dissipate 23^2*7.89m = 4.17 W.
Assume ambient temperature of 40 degrees C, max junction of 175 degrees C. According to datasheet RthetaJC + RthetaCS = .5+.5 = 1 K/W.
Rtheta junction to ambient = (175-40)/4.17 = 32.4 K/W. Total needed is 32.4 - 1 = 31.4 K/W.
A heat-sink with a rating of <31 K/W is needed for some margin. (Using the ASSMANN WSW V6560W 9 K/W, the MOSFET or sink never gets uncomfortable to hold heating from 25->100 degrees C)
The circuit I made was this. R2 and D1 are placed directly on the heat-bed. R2 was selected according to my preference of light intensity with the LED I’m using.
A PCB will be quite unpractical for this purpose, but certainly doable. Prototyping board? Forget it. The trace width required is… vast.
Using an opto-isolator like the RepRap power expander board is certainly possible, but the main reason for doing it, apart from isolation, is: the bed output pins on the main controller board is low side switched with a power N-mosfet already present. I wanted a very simple wire harness and I didn’t have an available opto-isolator.
Since this can’t be connected directly to the bed heater output, as noted above, SIGNAL is connected by piggy-backing on T5’s gate pin. Ground to power supply minus.
Note that there’s no flyback diode in the circuit. Even though the load is resistive, there’ll be some inductance. If it’s enough to kill the mosfet fast, I really can’t say. Yet.
D1 and R2 mounted on heat-bed. The solder pads are not exactly placed optimally for the 12 V power option.
New power supply installed. The supply screw holes were slightly larger than the previous supply, so I printed some new standoffs. 4 mm^2 wires for the head-bed. I also replaced the 0.75 mm^2 wires for the printer board power with a 1.5 mm^2 one. T5 piggy-backing is underneath the flatcable slightly above the bed heater output pins.
MOSFET, heat-sink, bullet connectors, thermistor wires, control wire and bed supply wires.
Next modification will be the whole build platform; this is just too thick.
Heating the bed 25->100 degrees C just lying on the kitchen table was 2 minutes 13 seconds. In the sandwich above, it’s 4 minutes 2 seconds.
Current draw starts at 23 A at 21 degrees C and goes down as temperature goes up. At 100 degrees C it’s 18 A.