I’m not sure if anyone’s still interested in electronic load projects anymore, one could suggest it’s kind of been done to death by now. However, I’ve had some good success of late in the physical design side of fitting electronics into enclosures, so it’s brought me back to finishing off the eload project I started probably close to a year ago. Albeit with a totally different outlook on the form factor.
In fact, I recon this could be a good candidate for a kit.
So what I’m doing is to package the components so they can make use of cheap COTS IT hardware.
First off, I’m using standard COTS CPU coolers as have a few others on the forum. I’ve designed a PCB that will mount two TO247 case MOSFETs in the same way as a cpu does. Basically it’s nothing more than physical mounting block that routes all connections to a convenient location where it can be connected via pluggable headers or terminal blocks. First the two MOSFETs are tacked in place using CA glue, soldered, then everything is bolted on around them.
I’m using a Cooler Master Elite 120 case (albeit these parts would fit it most cases.)
http://www.coolermaster.com.au/product.php?product_id=6777So here’s some snaps prototyping, this case will hold two of these load cells:
http://i1085.photobucket.com/albums/j432/harvie256/2013-04-24150311_zpsdf5b5c49.jpg[/img]And the board that’s ready to head off for manufacture.
The control boards are being designed to fit into 3.5” drive bays. It’ll be a single fully isolated control board per load cell, allowing the system to be directly scaled depending on the total power, current or channel requirements.
I’ve been using the Thermaltake contact 21 cooler with two fans fitted and arctic silver thermal compound. A quick survey of the local computer store showed this to be a very high bang for buck price point, at just $25ea for a 140W rated cooler. My design goal was 150W per load cell, I’ve load tested this at 175W for an hour, with the maximum device case temp getting to just under 30 degrees C above ambient. So very manageable as long as 300+W of heat can be extracted from the case (yet to be seen, and I’m open to any good ideas here.) I've also tried it with a water cooler, but I don't think the advantage anywhere near justifies the price and hassle.
The control boards will be pretty much what I’ve built before but repackaged to fit the drive bay form factor, and the software re-jigged to fit. They’ll consist of an isolated supply, Cortex M0 (STM32), isolated asyc link, 16bit DAC, 16bit ADC with 0.05% reference, auto-ranging voltage, etc. With a powerful uC and its fast on-chip ADC, constant resistance, voltage and power can be implemented digitally with a relatively high bandwidth.
I’m also putting together a separate board with basically two isolated auto-ranging DVMs. Again just using the circuitry I've already prototyped, just repackaging. As pretty much whenever I’m loading a supply I’m always also log multiple voltages around the circuit, which is a bit of a PITA at the moment.
Front panels will be in the form of laser cut acrylic panels with banana jacks etc that would fit into the standard 5.25" drive spaces.
The current idea is then to bring this all back to a central board with USB link (and/or Ethernet etc.) That final board is pretty low on the priority list, since there really won’t be a lot to it. Maybe even a linux SBC with plenty of serials.
So from a kit perspective, it would be pretty straight forward to supply the boards assembled and tested. Then the builder could locally source all the relevant IT hardware to make a complete system.
Any thoughts?