BESC - beefed up VESC motor controller for high power ebikes

[kct]

G'day,
I'm planning to build a 150v spec'd BESC-G2 and was wondering if anyone knows a reason I can not use 0.5oz copper for the internal layers. I'm only curious as I can get 0.5oz internal layer boards from JLCPCB for less than half the price of 1oz boards from PCBway and am a uni student on a budget.

I'm converting a KTM to electric and using a Kia Niro Hybrid Starter Generator as a motor (IPMSM). I've worked on these kinds of things before though personally have no experience with four layer board design and would hate to fry the thing just for cheaping out :lol:

The BESC look amazing! Thanks for making it open source Galp.
Cheers.

 

[amberwolf]

Half the copper thickness means half the current carrying ability, and half the heat transfer ability, on the traces using less copper.

If either of those is important for the traces that would now be using less copper, it will affect how well the controller performs for you.

 

[kct]

Yeah looking at the layout the high current devices on the board all seem to be interfaced to the bus bars on the back unless I'm missing something. Saying that the shunts interface through the traces on the bottom layer and have vias through into the internal layers. You could be on the right track, I might run into current and thermal problems. I am probably better to air on the side of caution.
Thanks

 

[4πr^2]

I would think your internal layers should only be carrying computer / data signals, no? If you're putting high enough current into the core of the board where 1/2oz or 1 oz makes a considerable difference, might be time to reevaluate the design. The other consideration is putting high current in the center of the board, if you ever pop a trace the whole board is toast. If you pop a trace on the top/bottom, at least there is a possibility for repair.

 

[Jrbe]

I looked quickly at the design. The 2 inner layers are setup as mostly planes with the majority of signal traces on the top and bottom. I didn't spend a lot of time going over it. But it looked like as long as you're using the add on copper buss bars it will be ok. You can go in and increase the internal trace widths if you're worried, there looked to be room around them.

I was considering reworking this design to be compatible with the JLC assembly service. I don't think anyone has done this yet. I checked the forks on Github and didn't see much of anything meaningful but didn't dig deep.

 

[kct]

Thanks for the input. I think you're both right. If the difference between 0.5 and 1oz was critical they'd be bigger issues to tackle. I think the inner layers are mostly there for the low side componentry (I think). I'm going to leave the inner trace widths as I think if they're what makes or breaks it a new layout would be in order anyway (and beyond my time/ability at the moment).

I'm planning on doing the board assembly myself to keep cost down and because I find it somewhat therapeutic. I'm equipped to for basic reflow work and am no stranger to hand soldering small components when necessary.

When you say re-work the design for assembly service what's that involve? Surface mount fets? ICs all on the one side, something like that?

 

[kct]

I'll drop a line back hear and let people know how it goes with the 0.5oz inner layers once I have it up and running and get to test it under heavy load. That'll probably be a few months given my schedule and component availability.

I'll spin up a thread on the bike build once its halfway there as well.
Thanks for the help guys its greatly appreciated

 

[Jrbe]

I'm planning on doing the board assembly myself to keep cost down and because I find it somewhat therapeutic. I'm equipped to for basic reflow work and am no stranger to hand soldering small components when necessary.

JLC typically has very low cost components and very low cost assembly. The builds I've done (not motor controllers yet) were cheaper to have JLC make boards, buy parts from them, and have them assemble them vs. me only buying boards and components. It's a special kind of torture to hand assemble boards.

When you say re-work the design for assembly service what's that involve? Surface mount fets? ICs all on the one side, something like that?

JLC has a limited inventory of parts they have in stock. Though it seems they're expanding into being able to purchase parts from outside vendors as well for their assembly services. They also only do single side component placement currently.
Basically the components have to be in JLC's offerings and have to be single sided smd assembly. So the design gets reworked to fit those. It's not typically a quick or easy thing to do. Sometimes you can have JLC do maybe 80% of the assembly and populate the rest yourself so you don't have to hand solder all the resistors, capacitors, and anything else they have in stock repeatedly on your boards.