Nice thing about switching resistors at low voltages, you don't much care what the gate voltage is, as long as it's enough extra to account for the resistor's drop. :)
That, or use differential sense.
Actually, differential sense is probably a requirement if you want any accuracy (better than 1%, 0.1%?) out of this -- PCB resistances will screw things up, down in that range.
Tim
Yes. We got away w/o differential on the single PCB, with its massive ground, but that can't work with multiple banks.
I'll likely not do more than four banks/section, but might place at least two. We'll see.
How do you heat sink a DPak to 25 watts?
Hang it off the edge of the board onto the heat sink (or route a hole in the board) like the fets.
I figure maybe 5 watts or so max, using regular mounting with some thermal vias.
Can you tell us more about your power resistors, what values and specs, etc. Were these in 1-2-4 steps, etc.?
For high power (kW/MW) dummy loads feeding back the power to the AC network with an inverter makes sense.
However, in this case, we are talking about only 1-2 kW, i.e. electric stove power levels. However already at these power level, you can evaporate a few kettles of water in a long time testing. So getting rid of the heat during long time testing is an issue.
In this thread, there has ben some discussion, should the power be dissipated in a resistor or in the junction of a transistor.
Power resistor can handle up to 400 C temperatures, while silicon transistors can handle only 150 or 200 C at best. From the heatsink design point of view, dissipating the same heat power from a resistor to the environment is much easier.
Now there's a good idea. But it'd take a special kind of inverter control system. Maybe one meant for solar panel roofs. I'm not knowledgeable about how they work.
Another good idea, improve low winter-time humidity.
Updated, to include yours and Arie's suggestions.
ti says LM3914 is still active.
1k 225W x 10. Unary == equal values. Operating range 100-500V 0-4A (and less current at lower voltages).
The linear sinks are ballasted by smaller resistors (100R 100W x 3), which should improve capacity and fault tolerance. At the expense of operating range of course, but I feel the low end is an acceptable sacrifice for a high voltage load.
Output is also slightly filtered (RC), polarity protected, MOV'd and fused. You could literally plug it into the wall, forever (i.e. including rare lightning-induced transients) and not have a problem (fingers crossed).
Tim
P.S. I just noticed, you poor bastard, slumming it on a Yahoo e-mail? Harvard's really fallen these days... :^)
Cute. Not sure how it works tho.
I'd love to make one of those some time, but it's obviously a bit of a pain to put together. :-)
Dumping it into car batteries is another option; I did a project a bunch of years ago, the client was conditioning NiMH cells so I put a lead-acid "accumulator" on the back end and a bunch of synchronous buck converters in front. Discharge one cell while charging the other and so on, and the converters power each other for the most part; the main bus really only needs a trickle charge to account for charging losses.
Resistors are a clear win on size, as you get a heck of a lot more heat into the same volume of air, or heatsink, at 250C+ than you do at 100-150C. If you don't mind that the exhaust is hot enough to ignite some materials...
The budgetary win is less clear. Heatsinks aren't the cheapest to buy or machine or assemble, but resistors need mounting, too. Resistors on custom combo brackets would be best, but that costs NRE; mounting with the standard individual clips, isn't much better than heatsinks in terms of hardware parts count and assembly time. It may turn out a wash.
For Win's case, the low voltage is probably easier to dissipate with transistors, and the desired dynamic range won't work so well with resistors.
Tim
The inductance isn't much of a problem, because it can be exactly nulled. Series inductance adds a voltage that is proportional to what a Rogowski coil picks up, and you can print the compensation element next to the sense resistor on the PCB.
My Harvard email account is clogged with over 100k emails. Mostly all legitimate, sadly. I use Yahoo to get work done.
But heat sinks are a struggle. The cute WA-T247-101E heatsink on my RIS-796 250A pulser is only good for 7.5 watts (100C) or maybe 10 watts (too hot). The pulser can do 60 volts and 400A = 24kW. I set it to its 30A minimum for a test, set the supply to 15V, and turned it on. The TO-247 MOSFET didn't mind the full 450W power level, but smoke immediately poured from the 20mR ballast resistor. Rated at 3W, it couldn't handle 18 watts, even for a few seconds. I had to use under 2% duty-cycle for the tests. In operation with 10V across the MOSFET, 250A pulses are limited by the cute heatsink to under 0.4% duty cycle.
temp just depends on the fan, it's not hard to keep 2kW at a safe temp.
A combination of both has advantages.
Someone mentioned power R cost somewhere upthread: it can be cheaper to use an array of lower P devices, if you have the space. Not that that's news, but it can wipe out a lot of the R cost.
NT
Have something to add? Share your thoughts — no account required.
Ask the community — no account required