Recently I was involved in the design of a range of
PCM's for lithium batteries of capacity up to 200Ah. Testing such a PCM involves dumping currents of over 200A. I work on a tight budget so purchasing a programmable dummy load for this sort of current was out of the question. So I improvised. I ended up using a combination of high wattage fixed resistors for loads up to 30A or so, a home-made water cooled load for high current, and a programmable load when I needed to tweak the current.
My programmable load was one of those projects where I did not do sufficient planning and research. I imagined it would be a doddle to make. I should have heeded warning bells at this point as I have "been there, done that" rather too many times in my short life.
Here are a few pictures of the construction of my programmable load. The basic idea was that a cheap MOSFET can dump 10A easily, so 25 of them should dump 250A. Each MOSFET has a 0.5R resistor in the source to help equalise current sharing. All 25 bases were connected in parallel (with 100R series resistors to prevent parasitic oscillation) and driven by an op-amp which compared the voltage on a multi-turn pot with the output of a
200A current sensor IC.
|
Four of the five panels, each has five MOSFET's with their source resistors |
|
The 6th panel with op-amp circuit being assembled |
|
Here are the five MOSFET panels about to be assembled |
|
Almost complete |
|
Details of a MOSFET panel |
A 12V 100mm fan cools the five MOSFET panels. No calculations here: just vain hope that forced air cooling is very effective. But I hadn't bargained for the gate voltage negative temperature coefficient.
If any MOSFET took more than its share of current it would dissipate more power than the others and get hotter: this temperature coefficient would then make the MOSFET conduct even more leading to thermal runaway. My solution was to add a NTC thermistor in the base circuit of each MOSFET and cemented to its body. A bit of a cludge but it sort of works.
In the event I actually rarely used the programmable load. I found that various combinations of fixed resistors generally was adequate. But it was fun building it.
No comments:
Post a Comment