Electronics > Projects, Designs, and Technical Stuff
Need help understanding/calculating single pulse power for MOSFET
jbrownshow:
I'm working on a project where I need to quickly charge up a 1300uF capacitor to 100V from a battery. I'm using a low-side MOSFET that when switched ON, charges the battery through a 5.1 ohm resistor. I'm using LTSpice for simulations and can easily get the energy through the MOSFET. The problem is that I can't convert the waveform into a single square pulse to compare with the datasheets in the SOA and Thermal Transient Impedance plots. Can anyone help with this or have a better way of sizing a MOSFET?
ocset:
which mosfet are you using?
Loads of mosfets should be able to manage that.
Theres very low V across the fet when its on, so the SOA chart wont be so useful as its more about operating in the linear region.
A good low rdson fet would not have any problems. Max current is just some 19A.......even with a 10mR fet thats a instantaneous max power pulse of just 19^2 * 0.01 = 3.6watts
jbrownshow:
Thanks for the reply. I’ve selected a SQJA20EP for the time being but I would like to reduce its size. That’s why I’m trying to understand single pulse power limits better. I know for resistors you can convert an exponential decaying pulse like this to a square wave. Where, for a short pulse, P= Vpk^2*(t @ 37%Vpk)/(2R). But with the MOSFET it seems there would be a different formula because of the linear region. It would be great if I could covert the energy to a square power pulse.
T3sl4co1l:
That's 6.5J, so a lossy charge must dissipate 6.5J for a total of 13J consumed.
You'll need a pretty big transistor to do that all at once in a hurry; or a slower (current limited) charge, giving time for the heat to spread out; or a saturating transistor with a ballast resistor that's similarly rated for the energy and power.
The SQJA20EP is only rated 16mJ avalanche, so good luck doing it pulsed. That's fine; you're already looking at the resistor method. Then, how much does the transistor dissipate?
If the Rds(on) is typ. 50mohm, then it will absorb about 1/100th the energy of a 5.1 ohm resistor, or 65mJ. Still above par.
The time constant is 1.3mF * 5.1R = 6.6ms, so the most energy will be dropped in the first few ms, and essentially all of it after about 20ms.
1ms SOA is almost 600W, or 600mJ total (single pulse, from 25C).
But our pulse is longer, so the total allowed energy can only be higher still (10ms is like 180W, or 1.8J), and it's fine, even if it were a square pulse.
This would seem to be fine. :-+
T_J initial is 25C in the SOA, and T_J final is 175C, the maximum rating. With the pulse delivering somewhere around 1/10 to 1/30 the allowable energy at these SOA curves, presumably the temperature rise is 1/10 to 1/30 of 150C, or 5-15C. So T_J max for your case should be less than 160C.
I would recommend a comparator and latch, to watch for:
- Overcurrent, in case the resistor fails shorted or something else nasty happens to the circuit
- Charge (current * time), or capacitor voltage, in case the capacitor isn't charging (load short?)
- Maybe temperature? It seems you have quite some margin here, so a simple limit at 150 or even 125C should be quite safe.
It seems you could also perhaps use a smaller (cheaper?) transistor.
Obviously, you need to run the same check for the resistor; it's dissipating 2kW peak and 6.4J total, so a modest power resistor will be needed. Wirewound resistors tend to be good at this, with vitreous enameled types preferred over molded types (including metal-case types which are sealed with plastic).
Tim
jbrownshow:
Thank you Tim for your analysis. That definitely helps me understand how to relate my circuit to the datasheet. May I ask your opinion on the Vishay CRCW HP resistors for this application? I was thinking of using a CRCW08055R10FKEAHP for this application as opposed to a wire wound. What do you think?
Navigation
[0] Message Index
[#] Next page
Go to full version