# MOSFET trio/quartet viable option?

I need to supply 10-15A at variable voltage without anything getting too hot. Is it good practice to use several MOSFETs to handle larger currents?

I am currently using an IRF9630 P-FET with an Rds_ON of 0.5-0.8 ohms, but will likely use something better for my finished device.

I am aware that FETs aren’t 100% identical, but I’ll check to make sure they output the same voltage down to two decimal points. A quartet handling 15A would each have a W_dissipate of 3, assuming a terrible Rds_ON of 0.8 ohms.

Cluebats?

Edit: Olin has schooled me, but I need more schooling. My FET absolutely must not fail closed, or some things will be destroyed. How do you guys protect hardware from this kind of malady?

Cluebats... my skull is thick.

• What you're asking isn't clear.... What supply voltage are you starting from? Do you want to vary the voltage/current into the load from 0 to some value? Max load voltage? Sensitive to noise? Nature of load? ;) – peufeu Sep 25 '17 at 12:27
• @peufeu 13.8V... I wanted to adjust this at the FET w PWM, and read voltage through a 10V zener. – user2497 Sep 25 '17 at 12:28
• I don't get the zener thing. Can you post a schematic and more explanations? – peufeu Sep 25 '17 at 12:48
• @peufeu I just measure charge- and battery voltages with a 10V zener between battery positive terminal and ADC on my arduino. It works fine. Olin has answered my question. Can’t use a quartet :( – user2497 Sep 25 '17 at 13:03

## 1 Answer

FETs have a positive temperature coefficient of resistance, so can be paralleled. However, that just means one FET won't run away, take most of the current, and blow up. It doesn't mean they will share the current equally. There are some options:

1. Use a single FET and PWM. Depending on the load, you may need a LC filter to smooth the pulses.

Another way to look at this is to use a switching power supply to make the desired voltage. You'd be controlling the reference voltage input to the switcher.

2. Use BJTs with individual emitter resistors instead of FETs. BJTs have a negative temperature coefficient of their saturation voltage, so can't be paralleled directly. You get around this by giving each BJT its own emitter resistor. If you're going to add a resistor and operate them out of saturation, BJTs are probably easier to use in this case due to the relatively tight range of B-E voltage. The G-S voltage of FETs will vary more.

• Thanks Olin. Well, looks like I will have to use a much better single FET. I have some IRLZ44Ns enroute from asia, but I don’t trust them for this. If you can recommend the IRLZ44N I will buy a handful locally, from a reputable vendor. I need this device to not explode. What say you? – user2497 Sep 25 '17 at 11:23
• Also, the FET must not fail closed, since that would destroy some hardware... do you have a trick up your sleeve for this contingency? – user2497 Sep 25 '17 at 11:24
• I have many SMPSs, but none of them can be controlled with digital logic. What would you google for? – user2497 Sep 25 '17 at 11:29
• @user: You'd make your own switching power supply. Actually I wouldn't even make one with a analog reference voltage. I'd read the output voltage with the A/D of a microcontroller, then control the FET gate with a PWM output of the same micro. The rest is firmware. – Olin Lathrop Sep 25 '17 at 11:33
• I have a 300W buck converter... in a pinch I could just replace a trimpot with a variable resistor that could be controlled digitally. Do such resistors exist, if so are they reliable enough? – user2497 Sep 25 '17 at 11:35