I'm trying to put something together to control the heated seat in my car, and have come up with this:



The idea is to switch the heating element off once a preset temperature is reached.

R1 = NTC thermistor attached to the seat (5-15KΩ in the temperature range I'm interested in.)
R2 = Pot for setting the cutoff point
LOAD = The heating element - a resistive load of approx 2.6Ω @ 12V nominal (automotive, so really 13-14.5V)

The circuit I based the design on used a relay to switch the load on/off, but I wanted to avoid this as a relay clicking every few seconds would be annoying, so I've replaced it with a MOSFET. I'm not really that familiar with using MOSFETs, so the question is: Does this look right?

Also from reading the datasheet for the IRF740 (Q1) I've calculated the heat disipation as being approx 1W - this doesn't seem much considering it will have 5-6A going through it.

 

It would be better to use a comparator IC rather than an amplifier IC. With this circuit it is more possible to drive the MOS half on, overheat it and burn it.

Also, replace R2 with a short and R5 with a pot to set the trigger point.

 

I should have mentioned that I'm replacing the original controller, which is broken.

R1, R2, and R3 represent how the seat is wired internally, and I don't have any choice about that without pulling the whole thing apart. The seat simply has 3 pins for connecting the controller: 12V, ground, and the output from the voltage divider formed by the thermistor and pot+R3.

As for the comparator, something simple like the LM393?

 

Yes it will do the job with a few modifications.

First, remove R6 completely.

Second, connect the gate of the MOS directly to the output of the comparator (remove R7).

Finally, connect a 1K resistor between 12V and the gate of the MOS.

 

Won't removing R6 cause it to switch on/off rapidly with no hysteresis?

 

Hysteresis is caused when you introduce positive feedback (i.e Schmitt trigger). Don't worry about the switching times for your application, you don't need to switch really fast.

 

OK, that's great, thanks.

One last question. You mentioned adding a 1K resistor from 12V to the FET gate. Why is this? My understanding was that R8 was a pull-down to stop noise affecting it, so is the 1K instead of, or as well as R8?

 

Comparator ICs output are open collector outputs. Thus to charge the gate capacitance of the MOS and turn it on you need a pull-up resistor. A 1K is a reasonable value.

The 10K pull-down resistor is used to discharge the gate capacitance of the MOS and turn it off in case the comparators fails.

 

If the comparator output fails, the 1k and 10k resistor will comprise a voltage divider, holding the MOSFET's gate at 10/11 times Vcc. In an auto with the engine running, that's 12.54v; the MOSFET will be "stuck" ON.

Without hysteresis, it is VERY likely that oscillations will occur, causing the MOSFET to be operating mostly in the linear region, generating lots of heat.

The LM393 is not rated for the automotive application temperature range.

The IRF740 is a high-voltage MOSFET. It would be more appropriate to use something like an IRFZ24, IRFZ34, or IRFZ44, rated for 55v but much higher current and lower Rds(on).

IRFI1310 is another possibility. There are quite a few MOSFETs out now with lower gate charges and lower Rds(on) values.

 

If the comparator output fails, the 1k and 10k resistor will comprise a voltage divider, holding the MOSFET's gate at 10/11 times Vcc. In an auto with the engine running, that's 12.54v; the MOSFET will be "stuck" ON.

That was how I understood it as well, but not being sure I assumed I was wrong.

Without hysteresis, it is VERY likely that oscillations will occur, causing the MOSFET to be operating mostly in the linear region, generating lots of heat.

Do you think I'm better sticking closer to the original then - a 741, with R6 for feedback to stop it oscillating?

 

The 741 is problematic, because it is so old and slow. It is also not capable of getting very close to either power rail, which means that the MOSFET may be left in a partially-conducting state when it's supposed to be OFF.

This will cause heat and wasted power.

I'll have to re-visit this.