Quote:
Originally Posted by MazdaMatt
Paul,
I know that the bombardment of ideas is nice to read through and ponder for the future, but I'm really interested in how your basic controller is going. I'm a little lost in the 13 pages of progress and discussion - could you give us a rundown of your "done" and "to do" lists with brief description? (and maybe a "might be nice to try" list if you have a few ideas pending)
Thanks!
PS - do you have any CAD drawings of the layout? I could see about a quote from one of our board suppliers. If you have a schematic in a cad program, i may try to copy your layout if i get time (and an internet connection at home... i'm so ghetto).
pps - how'd you come up with your initial schematic?
|
The basic controller is going very well.
DONE SO FAR
-------------
Functioning low voltage controller. I have worked out a lot of the bugs that could happen in the more expensive high voltage case. I am basically done with the etching designs for the high power section PCB. There really isn't much to do in that regard. There are only about 6 possible ways to do it. You can't permute B+, B-, and M- in too many ways! I know how the layout will be, and how the low power section will connect to the high power section. I know it so well now, that I don't need a schematic anymore. I can walk through each detailed little resistor and capacitor and mosfet in my head, and I know exactly why it needs to be where it is. The practice low power version really helped in that regard.
TO DO
------
Change the throttle interface from 3 wire to 2 wire. That way, the standard PB-6 can be used with the controller, instead of a radio shack 5k pot. You can still use a radio shack pot, but you will only need 2 of the 3 leads. The chip necessary to do that came today! That's about 1 hour of work to have it changed over and tested.
I need to change the power supply of the mosfet driver and micro-controller to the fancy isolated 15v-15v DC-DC converter. I haven't ordered it yet. It should be pretty easy. It might be a couple hours of testing to make sure everything is wired right and working properly.
I need to get the dang mosfets, diodes, and capacitors. That will probably be around the 10th of Feb.
I need to get a mill for etching the power board.
I need to transfer the low power section from a breadboard to a prototype PCB, solder everything in place, and test it. I also need to add a nice clean interface for in system programming of the micro-controller onto that prototype PCB. Then, I need to figure out how the real production control board ought to look. I have ExpressPCB and ExpressSCH for that (it's free!).
I need some sort of box to put the controller into. I need a good cheap supplier of aluminum fan heat sinks (to attach to the big dang aluminum heat spreader). I need to compare the heat dissipation from isolating the back of each mosfet/diode from the heat spreader vs. attaching them directly to the heat spreader, and isolating the spreader from the heat sink. I need to figure out which material works best. I'm using mica right now.
I need a good source of copper bus bars. Several questions come to mind: Should the size of bus bars be chosen so that no cutting is necessary? Is that even practical? Open source means people may not have access to band saws.
I need to come up with a good PI loop for finding right PWM duty that gives the correct current. I don't know of a good way right now to observe the unit step response for loop tuning. I'm still learning how to use my oscilloscope, which came without documentation. anger. I'm not sure how to see transient stuff (one shot stuff). I think that tuning the PI loop to the bike motor would not be a good idea. The car motor and bike motor are very different.
I need to remove the car's 72v 400amp curtis controller, and put mine in place, to test the software with the very different motor. The car's motor is 6.7" diameter, 60 pounds series wound DC. The bike's motor is brushed permanent magnet DC. I only have a 72v system, and Ben Nelson volunteered to try it at 120v or even 144v. He has 12 Gel Cell batteries.
For thermal shutdown, I know what temperature the mosfets should be shut down at, but I can't monitor them. I have to monitor the temp of the heat spreader. So I need to find the right temp of heat spreader that gives a safe thermal shutdown temp. I haven't been able to test that, other than with a hair dryer. The heat spreader has been staying cold in this low power context.
I need to reprogram the controller for a new Hall Effect current sensor. The new current sensor will have a range of 0 to 500 or 600amp, and I'll have to rescale things, and make sure it's done right. I don't want to cook $150 worth of mosfets. I also need to test speed differences of using the 10 bit resolution temperature and current monitoring vs. 8 bit monitoring. I would prefer the 10 bit, but I switched to 8 bit because I needed 8 bit PWM duty resolution to get up to 16 KHz switching frequency. So, I changed everything to 8 bit, so I could do a programming speed hack that does a single read of the high 8 bits of a 10 bit measurement rather than 2 reads of a 10 bit measurement. (it's an 8 bit processor)
I think I want to add some hardware current limiting too, and not just depend on the micro-controller. I have a schematic of how to do this from someone on the EVTech list, but I've been told it takes some debugging to get it right in the noisy environment of a high power motor controller.
I don't have CAD, but I do have ExpressSCH, which is pretty nice, and very similar. The original controller design was from a few sources. I took ideas from Ian Hooper's 24v controller for an electric trolley (with his permission), and various readings that I found on the internet on controllers based on 555 timers. He has also helped with possibilities for the power section layout. I don't know which one to use yet. I might go with his (1 of 6 possible hehe) since I know it works. I made the schematics that I posted myself, though. He has also had some really good suggestions on which mosfets, diodes, and capacitors work well.