Cheap FOCer (VESC 4.12 based design)

[Samd]

Interesting thread. Subd.


Sent from my iPhone using Tapatalk

 

[qwerkus]

While we're waiting for me to make progress, I have a question. Does anyone have any success stories of fan cooling motor controllers while still keeping the enclosure decently water resistant? A quick search on the internet turns up weird or impractical things for venting an enclosure and maintaining water resistance.

I second that: best way is outside cooling. Thermal compound between FET and aluminium box, and between aluminium box and some alu radiator / heat sink. A small fan will help a lot, but eventually die to the weather, not to mention the need for a separate power supply. The better way to do it would probably be to thermally paste the controller box directly to the bike frame. Given it's surface, it should provide a good radiator.

 

[qwerkus]

How many? You need to stamp the 2mm cap to create a tiny channel for the o ring, about 0.6-1mm deep. For 2pc, you can use a dremel. Than its just a matter of finding he right sized nitrile o ring.

Thanks for the suggestion. I have 20 of these cases, so I won't be able to dremel them all. I'm thinking the silicone sheets will probably work best for me.

Don't want to hijack this thread, but you can always go with a thick ring instead of a discrete one. 3-4mm ring firmly squeezed between cap and box will seal anything. The cap has to be strong enough though, otherwise it will bent. Also, you will need to pick weather resistant o rings, and it won t be pretty.

 

[qwerkus]

PROJECT UPDATE!

Quick question: how do you access the 2x large holes when there is an aluminium bar (the FET Rail) on top of them ?

 

[shaman]

@amberwolf

Thank you very much for the information. It seems like external cooling will be the most practical as pointed out by qwerkus.

@qwerkus

Good eye. I'm hoping that the FETs (and therefore aluminum bar) are stood off from the PCB enough so that a mounting screw can still exist there. It's not ideal, but I don't see another place on that part of the board for a mounting hole. The primary means of mounting the board to the enclosure will be the aluminum bar itself, so one wouldn't even need the mounting holes in the PCB. I can probably have better mounting hole placement on the 12 FET version where I have a bigger board to play with.

 

[shaman]

Ok so I added a 60V TVS diode to Vsupply going to the DRV8302. I've read a few threads where operating the VESC in FOC mode while using a 12s battery caused damage to the DRV8302. It is believed that this is due to the high voltage spikes produced while regen braking, coasting, or whatever. Hopefully the TVS diode will protect the DRV8302 from these voltage spikes when using this controller in FOC mode with a higher voltage supply.

 

[nieles]

if possible change your layout for the high power battery traces to a laminated busbar setup. so BAT+ on top and BAT- on bottom. any stray inductance in your layout where big currents will flow will induce voltage spikes and kill fets and drivers.

some inspiration:
https://endless-sphere.com/forums/viewtopic.php?f=30&t=91689&p=1372301#p1372301
https://endless-sphere.com/forums/viewtopic.php?f=30&t=55641

 

[shaman]

@nieles

I'm trying to make assembly as easy as possible for the community. The idea for now is to simply strip a length of the battery connector cables and solder the bare portion to the exposed tracks underneath the board. If one is using 10awg or 12awg wire then it should provide the necessary power handling capability and minimize inductance enough (hopefully). I agree that bus bars would be ideal but I'm trying to keep this design simple. You'll get to say "I told you so" if and when my design blows up during testing

 

[nieles]

a another option is to slow the switching of the fets way down by using larger value gate resistors. as the voltage spikes are caused by v = L*(di/dt)
but this will create more heat in the fets, increase your switching losses.

bigger wires wont lower the stray inductance. reducing loop area does -> making sure the + and - overlap.

 

[shaman]

Gotcha. I think I might be able to run the large ground trace under the large power trace. This would cause the board to only have one layer per large gnd/power rail but I think soldering on the wire as I mentioned before could compensate for the lack of copper on the PCB.

 

[pavlik1]

Very interesting. Thank you. This is exactly I am looking for, especially with the new DRV for 100V.

 

[Grantmac]

I'd be interested in trying this with my Cyclone. Can anyone compare the assembly difficulty with a megasquirt EFI? I built one of those pretty easily in the past.

 

[shaman]

I'd be interested in trying this with my Cyclone. Can anyone compare the assembly difficulty with a megasquirt EFI? I built one of those pretty easily in the past.

None of these have been created yet. I'm still designing it . I am trying to make PCB assembly as easy as possible. I don't know anything about the megasquirts though. If you are comfortable with soldering SMD components then making one shouldn't be difficult, just maybe time consuming. I imagine assembling one of my controllers will be easier than assembling the original VESC.

 

[liveforphysics]

Love it. Any shot at a 12f layout or a 6f TO247 package version? I would chip in a few hundred bucks to cover some board spins.

 

[shaman]

Love it. Any shot at a 12f layout or a 6f TO247 package version? I would chip in a few hundred bucks to cover some board spins.

I definitely intend on making a 12 FET version that will also handle higher voltage. Let's call this first version I'm working on a proof of concept and me practicing my PCB design skills. A high voltage (HV) version of both a 6 FET and 12 FET controller will then follow after I've tested the initial prototype. liveforphysics would you be able to assemble your own controller if I gave you the gerber files and BOM? If so, no need to chip in. Just help me test the design when the time comes. Otherwise we can discuss financial stuff once I have the 12 FET design ready to go.

 

[nieles]

is your plan to open-source the hardware design?

 

[shaman]

is your plan to open-source the hardware design?

Yep! I'll create a Gitbhub repository for the gerbers, schematic, and BOM soon. FYI this first design won't have the project files but the future versions I've mentioned will and they will be for KiCAD.

 

[shaman]

Project Update

I ordered the boards and BOM. Boards are from JLCPCB and BOM is split between LCSC and Arrow. Most of the BOM is from LCSC with only the current sense resistors and inductor coming from Arrow. Everything is decently cheap so far especially if you are getting from free samples from TI. If you didn't know already, you can make an account on TI.com and get 5 free samples of most ICs. You can save a little money here by getting free DRV8302s, TLV1117 LDOs, and the CAN transceiver.

 

[qwerkus]

@amberwolf

Thank you very much for the information. It seems like external cooling will be the most practical as pointed out by qwerkus.

@qwerkus

Good eye. I'm hoping that the FETs (and therefore aluminum bar) are stood off from the PCB enough so that a mounting screw can still exist there. It's not ideal, but I don't see another place on that part of the board for a mounting hole. The primary means of mounting the board to the enclosure will be the aluminum bar itself, so one wouldn't even need the mounting holes in the PCB. I can probably have better mounting hole placement on the 12 FET version where I have a bigger board to play with.

Ok, sounds fiddly. On the other hand, if you use alu boxes with rails that slot the PCB in (bmsbattery-style) 4 screws in the FET Bar will indeed be strong enough.

Any chance to get a partially populated board ? I don't have a reflow device, and soldering tiny pieces is just a pain.
"Beefing up" the main power line traces is no poblem though. I guess most of us DIY battery builder have some 0.1/0.2mm copper sheet laying around...

 

[shaman]

@amberwolf

Thank you very much for the information. It seems like external cooling will be the most practical as pointed out by qwerkus.

@qwerkus

Good eye. I'm hoping that the FETs (and therefore aluminum bar) are stood off from the PCB enough so that a mounting screw can still exist there. It's not ideal, but I don't see another place on that part of the board for a mounting hole. The primary means of mounting the board to the enclosure will be the aluminum bar itself, so one wouldn't even need the mounting holes in the PCB. I can probably have better mounting hole placement on the 12 FET version where I have a bigger board to play with.

Ok, sounds fiddly. On the other hand, if you use alu boxes with rails that slot the PCB in (bmsbattery-style) 4 screws in the FET Bar will indeed be strong enough.

Any chance to get a partially populated board ? I don't have a reflow device, and soldering tiny pieces is just a pain.
"Beefing up" the main power line traces is no poblem though. I guess most of us DIY battery builder have some 0.1/0.2mm copper sheet laying around...

Yes I think providing some partially populated boards to a select few is possible. You may want to wait until I verify basic functionality of the design before receiving one from me. Additionally I would like those that are receiving these populated boards from me to thoroughly test the controller in their particular rig. I will be doing a lot of testing in a lab but nothing is as good as testing the controller in a real e-vehicle in real conditions. If anyone has a lack of reflow equipment, there are still ways of doing reflow on the stove with a skillet. Its not pretty, but it's definitely a DIY technique.

 

[wil]

Nice design, keen to see a high voltage VESC design, happy with cheap too Subscribing

Aside from that one IC(fet driver?) with a central ground the whole thing could actually be hand soldered. Don't suppose there are any other packages for it are there? I'm guessing not since it will be using the ground as a heat sink.

 

[shaman]

Nice design, keen to see a high voltage VESC design, happy with cheap too Subscribing

Aside from that one IC(fet driver?) with a central ground the whole thing could actually be hand soldered. Don't suppose there are any other packages for it are there? I'm guessing not since it will be using the ground as a heat sink.

I haven't investigated alternative drivers for the cheap vesc. I may consider utilizing separate gate drivers for the high voltage version but as of right now I am looking at the DRV8353 which also has a thermal pad underneath. These packages are still "hand solderable" but hey just require a reflow station. I got mine for about $70usd.

 

[BotoXbz]

if you add a bunch of vias on the thermal pad, heat can be applied from below with a strong soldering iron
otherwise a hot air gun works too

 

[shaman]

Project Update




Boards and components are in. Seeing them IRL I realize the battery and phase holes for the wires are a wee bit too small. I can still barely fit 12awg wire in them though. Gunna populate the board over the course of the next week or so. Let's hope I don't encounter any crippling PCB design flaws.

 

[wil]

Looking at your photo it looks like already had the via thing suggested above?

What is your approximate BOM cost for this iteration?

You can always just drill out the holes to fit thicker wires, it's not like the tiny extra amount of contact gained by the PCB depth being coated really adds much in the overall scheme of things.