Warning: long post. During February and March we were building the power board. Credit for the mechanical design goes to Brian. I'll detail our experience in making this power board, but first here is a picture of the finished item:
Everything that follows took place over a couple of month period. As I mentioned in other posts, this controller is inspired by the Paul & Sabrina's DIY "open source" motor controller. We started with their design and then made some modifications. The mechanical design is quite different. I have to confess I am no longer up to speed on what they are doing. That discussion thread got to be hard to follow if you are not reading it all the time and I have not seen any kind of summary other than the posted schematics and documents.
Anyway, we are using a half bridge for synchronous rectification, instead of using a flyback diode as in the DIY controller. I am also using 5 pairs of MOSFETs paralleled instead of 10 in parallel. I am using MOSFETs with higher current specs and with chassis mount for heat sinking. Here is a simplified schematic of the power circuit:
There are also capacitors across the battery supply that are not shown on the above schematic.
First I will try to describe how this design is put together, then I will show photos of the fabrication process. Think of this as a stack of layers. At the bottom layer is the heatsink. Next are the MOSFETs which are chassis mount to provide for maximum heatsinking. The MOSFETs are IXYS IXFN230N20T. The next layer above the MOSFETs is the copper bus bars. These are bolted directly to the terminals on top of the MOSFETs for maximum conductivity. There are no wires or circuit boards between the MOSFETs and bus bars. At the next layer are copper standoffs which are used to electrically connect the bus bars as well as the MOSFET gates to the power circuit board. Finally is the circuit board itself which holds the capacitor bank and the gate resistors.
The controller logic board will connect to the power board via short wire lengths to the gate resistors.
We started off with the circuit board. This is just a blank board with 2 ounce copper plating on one side. Brian designed the component layout and came up with the copper layout pattern to show all the conducting areas. From this same pattern it becomes clear where the material needs to be removed to make separate conductive areas. He printed the pattern on a piece of paper and glued it to the copper.
Next, Brian used a milling machine to follow the traces and remove the copper as needed.
Then, we drilled out the through-holes:
After cleaning up and removing the paper, we have the finished circuit board:
My lessons learned from this: I think next time I would lay out this board using Eagle and get it fabbed by a board house. Machining this board the way we did worked pretty well but is labor intensive and if we want to build more it is not reproducible. Also, if we had a board made we could have plated-through holes and thermal vias which would help with soldering the components.
Anyway, I think this board came out pretty well.
For the bus bars, we started out with a piece of 1/4-inch thick copper:
One of the bus bars will just be a straight bar, and two of them have "teeth" to be used for connecting to the MOSFET terminals. So it is not just a matter of cutting this stock piece into 3 bars. Instead our plan was to use the milling machine to cut it into the patterns we need.
Like the circuit board, Brian made a pattern on paper and glued it to the copper, to be used as a guide for cutting.
Then we put it on the milling machine and started to remove material:
It turns out this didn't work very well. First, it was a very slow process. We could not remove much material at a time and it was going to take forever the way we were doing it. Second, because the copper is soft, the bit would catch on the edge some and make kind of a messy cut.
So we gave up on the milling machine and next tried using a scroll saw.
Well, this didn't work very well either. The blade kept grabbing the copper and it was also slow going. Finally we ended up using a band saw and were able to make the cuts. I have no photos of the band saw cutting experience.
Here is what the bars looked like after cutting (and before cleaning up). These still have the paper cutting pattern attached:
Next is drilling the holes in the bars where they will be bolted to the MOSFETs:
After that the bars were cleaned up and filed smooth. I don't think I have any lessons learned here. This mechanical work is actually outside my usual capability so the guys who actually did this work may have some other ideas for next time.
To make the spacers, we used some bar stock and drilled out the center and machined the outside to the desired diameter:
The picture above shows our first attempt at the spacers. It turns out this material was not copper even though it appeared to be on the outside. Instead it was copper coated something else. The lesson here is that if that copper bar is really cheap, it is probably not copper :-)
We got some real copper bar and made the spacers the same way. My lesson learned here is to try to find something pre-made and adapt the circuit board design to use that. We had to make these a certain size in order to work with the circuit board. But if we started with some off-the-shelf spacers maybe we could have skipped this step.
This photo demonstrates how the bus bars are attached to the MOSFETs, and how the spacers are used to connect to the circuit board:
HEATSINK AND MOSFETS
At some point we will attach a fan to the heat sink. That detail remains to be worked out. We drilled out the heat sink with the MOSFET mounting pattern and tapped the holes:
Next the MOSFETs are screwed down to the heatsink. For now we are not worrying about thermal paste, but when we get to the point of serious testing we will apply thermal goopy to the bottom of the MOSFETs for maximum thermal connection.
Here is the power circuit board with the capacitors installed, followed by a test fit of the bus bars to the MOSFETs.
And finally, here is one view of the finished controller:
THE MOST IMPORTANT COMPONENT
I almost forgot ... Ed keeps us supplied with his excellent homebrew - Thanks Ed!
THIS WEEK'S/MONTH'S EPISODE
I put together a short video showing the controller board fabrication. It pretty much just shows what is presented above. There is some motion video for some of the work we did. This quicktime movie is about 3 minutes long: