Electronics > Beginners
DC dummy load circuit calibration
<< < (4/20) > >>
Ian.M:
As I said earlier, below about 20A your MOSFETS have a negative temperature coefficient for Id - it will increase as they get hotter while Vgs is held constant.   That means instability if paralleled, with the possibility of one heating up till it hogs the majority of the current, and if its dissipation is more than its derated rating at the current heatsink temperature, its TJ_max will be exceeded and it will fail.     There's no certain way round this without separate resistors in series with each MOSFET source.   It may appear to work at first, but as the heatsink warms up, it may at any time start thermal runaway, soon followed by MOSFET failure shorting the supply under test.  The odds of avoiding thermal runaway are improved if the MOSFETs are very closely matched for RDS_on and gate threshold voltage and you also derate them significantly e.g.  each to 50% of the single MOSFET rating.

You can use the resistor to test the heatsink.   You'll need a high current 5V supply capable of more than 5A.   Mount the resistor to the heatsink with a smear of heatsink compound, and apply 5V to it, with an ammeter in series and a voltmeter directly across the resistor.  The heatsink should be left in the same orientation as it will be used in in your project.   After a couple of hours, measure the heatsink temperature and the ambient temperature, and from the difference and the power input from the resistor, you can calculate its RθSA.
VEGETA:
Here are the heatsinks that I got from my friend, what do you think?

How about if I mount the mosfets close to each other to make heat the same?

I need a solution to make this work, I cannot imagine why Dave's design works perfectly and mine is not despite being the same.
Ian.M:
Those are fairly small heatsinks out of a SMPSU.  I doubt they'll handle more than about 10W each without excessive temperature rise.   Test one with 25W from your 1R resistor run on a 5V supply and see for yourself.

Whether or not a particular set of MOSFETs will be thermally stable when directly paralleled is heavily dependent on their characteristics, how well matched they are and how hard you push them.  It helps if you start off with MOSFETs designed for linear operation - if there isn't a DC line on the MOSFET's S.O.A graph you'll probably get a nasty surprise if you push it past a small fraction of its rated power even without paralleling.

In EEVblog #102, Dave used a single MPT3055VL, which is rated for DC linear region operation.  Unfortunately your IRILZ44N doesn't have a DC S.O.A rating so you are gambling even with only one, before you even consider thermal runaway issues if you parallel them.

See https://www.eevblog.com/forum/projects/electronic-load-mosfet-balancing/
VEGETA:
I only have the IRIL44Z for now, but I ordered IRL640A which should be here in a week or so.

What if I used 4 mosfets with each one at one of those heatsinks? We have around 30*1.5 = 45 watts. Shunt resistor will get 1.5*1.5*1 = 2.25 watts which doesn't need a heatsink (or maybe one of the small radiators mentioned previously).

So we are up to 42.75 -> assume 43 watts which means around 10 watts per heatsink, or if 2 heatsinks are used -> 20 watts per heatsink. I will use thermal paste to stick parts together to the heatsink and get the heatsink to be kinda flush to the case.

You mention pushing the mosfet beyond its rated power, but I don't think I am doing this. If I used 4 mosfets, with 1.5A maximum... then each one will get around 43/4 =~10 watts. Now, 3.3 *10 = 33 degrees above ambient -> 33+30 = ~ 60 degrees which is nothing especially with a heatsink.

How does Dave's mosfet handle 40 watts of power alone?

I need a simple solution to build this circuit, while the better version is for another day.

Ian.M:
At 10W per MOSFET, you may be OK.  However you'd be much safer if you had separate resistors for current sense for each MOSFET, each with its own OPAMP driving its gate.   3x 1R 1/4W resistors in parallel would give you a 0.33R resistor good for 0.75W, or 1.5A   16x 0.33R resistors in groups of four, four MOSFETs and four OPAMPs would let you run at up to 6A at low voltage, dropping to about 1.5A at 30V to keep the individual MOSFET dissipations low enough for reliability.

You may wish to consider using an Arduino for monitoring, control and data logging with sensors to monitor the heatsink temperatures, and also to read the load voltage and current so you can program it to shut off the load if its safe ratings are exceeded.   However, if you are going to connect it to a PC, I strongly recommend optoisolating the serial data lines between the ATmega CPU and the USB<=>serial chip to avoid any risk to your PC.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod