High power VESCs for Robot Wars

[Rapidrory]

Ok, so this is a little off the topic of the usual electric vehicles on here I guess, but hopefully I can get some pointers on how best to go about this..

Over the last few months, I've been looking for a way to use brushless motors to power combat robots; up until recently it's been done exclusively with brushed motors due to the need for high starup torque and quick reversing, however the weight and power benefits of brushless are very attractive when you have a strict weight limit. After some initial success with various SimonK based controllers, I came across the VESCs and they were clearly the way to go in terms of low speed torque and control. I got some PCBs printed and put them together to run some early tests ( https://www.youtube.com/watch?v=co9p6lGzBbo ).

However for combat, 50 amps is quite low. The PCBs I sent out for print had a few tweaks to the track thickness's and holes for proper heatsink mounts, and I fitted slightly higher current MOSFETs and 0.5mOhm shunts to raise the safe current limit a little, but at a guess that will only have pushed it up to around 70 Amps continuous at best. Having seen that they work, people are now asking me to make 150A or 200A versions, so I'm looking at upgrading the driver circuitry, and increasing the board to 2 MOSFETS in parallel per channel. What I'm wondering is whether the firmware will still work up to these levels without needing modification, and if so, is there a limit to how far it can be scaled before the physics breaks down? (I know people eyeing up ~1000A brushless, but I'm definitely not promising them anything!). I've seen questions asked about higher current VESCs by other people, but couldn't find any examples of one being made. If someone has built a high power VESC and has details, any info would be great as it would save me some trial and error!

We recently used the ~70A versions in a 110kg robot which featured on the TV show, however we had issues with them cutting out and not restarting until the power was cycled, which isn't ideal when you're surrounded by other machines wanting to destroy you! The current limit was set to 60A on 12S lipo, and it was in PID control mode. Not sure what failure mode would cause this, and we don't think it was a loose connection. Difficult to say without some further testing, but any initial suggestions?

We also had a problem with one of the motor wires becoming disconnected and the controller frying itself. Is there a way to sense if the motor has become partially disconnected and shut down the controller or something?

Cheers for any help

 

[thepronghorn]

Have you contacted the designer of the VESC, Benjamin Vedder (website here) about the driver circuitry?

The VESC has been mostly popular with the electric longboard crowd, so the "Stand-up E-scooters and E-skateboards" section might generate slightly more interest, although the motor technology section is indeed correct for controller questions such as yours. Mostly I think your post hasn't gotten much interest due to the fact that VESC's haven't made the crossover to normal ebikes yet.

With regards to your actual question, I think the DRV8302 mosfet driver on the board is probably a little on the weak side to be handling more mosfets. You could try to make it work, or you could design your own gate driver circuit and use mostly just the logic part of the VESC. You might also want to look into the lebowski controller (search username: lebowski) if you want to go real hard with the custom controller stuff.

 

[toolbag]

I agree that the integrated gate drive unit is the weakest link on the VESCs. I've killed a couple when tuning the PIDs a little recklessly .

To make for minimal software changes you could take the output of the VESCs gate drive through a buffer stage like Zombiess' here: https://endless-sphere.com/forums/viewtopic.php?f=30&t=58341 . This would get you enough current to comfortably drive three mosfets in parallel. Zombiess' layouts are very thoughtfully created, so do try not to make unnecessary changes. There will also be considerations for current sharing of the mosfets on the power stage layout, so keep an eye out for mentions of that. With the freedoms afforded by a more robust gate driver you could switch to TO-247 or similar FETs for better heatsinking.

If you keep the same maximum DC link voltage then the component changes shouldn't be too heavy to accomplish what you're asking with a reasonable amount of effort.