Serious FOCer (84V VESC 6 based controller)

shaman said:
I'm pretty sure it's due to the long gate drive traces passing from the control board to the power stage thought the pin headers.
That's very likely, I also measured that the gate and output waveforms depend on the connector position and orientation, because there is a coupled feedback from the output circuit to the gate loop. I don't even have an optimal solution for that, maybe shielded board-to-board pin connectors would be the best (if such thing exists at all). The single board design probably causes less headache.
 
Instant kills are usually caused by overvoltage or shoot through events, both can be seen in a double pulse test. Before performing any high power test your gate drive should be tuned based on your double pulse test results.

If the long gate traces are too susceptible you will have to reduce the gate drive current to slow down the switching. Miller self turn-ON and overshoots are the most likely mosfet killers. Not sure what would kill a DRV, could be a voltage spike coming into a gate drive pin, which could be clamped with an external pair of schottkys to 12v and gnd.

Low side current sense produces higher EMI due to larger current loops and that one is hard to deal with.
 
@marcos I'll try reducing the gate drive strength and see if that helps. I'll still poke and prod the 2-board design to see what's going on for my own learning. I should have put the gate driver on the power stage board and only passed the digital signals between them.

@nieles Yeah EMC is a special kind of voodoo that can get complicated. It just so happens to be a hot topic in my department in my full time job and I've been diving into it. Yeah loop size or "area in the between the traces" is definitely a major factor in EMC immunity and emissions. @marcos really helped me in how to identify them.

@peters Yeah all in all I think switching to a single board with ideal driver placement, short gate drive traces, and a proper drive strength setting will all come together for a more solid design. I'm almost done designing the 5kW version. Look for a render sometime this week.
 
Feist92 said:
I think this controller will be epic once you work the kinks out so I will patiently await progress. Im really glad to hear you're not giving up on it. Keep up the awesome work :thumb:

Thanks for the encouragement! I'll crank something out. It's just a matter of time!
 
Hello, Shaman.

Looking for controller information yesterday, I found the Serious FOCer thread.

I am building a "Kart-enstein", an AWD buggy using 4 hub 60V 10000W motors of the kind used in the city coco scooters.

I already have the wheels (engines) and part of the chassis, but I'm still looking for the controllers.

The possibility of driving two of the motors with only one controller seems excellent to me. The maximum consumption of these motors should be about 18-20A @ 60V.

If you think your controllers can be used in my project I would like to contribute what I can and buy a couple of units when they are available.

Thank you
 
@olean

To clarify, there is still one controller per motor. The controllers can connect via the CAN bus to work together to drive a dual-motor system. 18-20A @ 60V per motor should be fine for the 5kW controller. Stay tuned for updates here and for when I have hardware available.
 
Top.JPG
TOP
Side.JPG
SIDE
Front.JPG
FRONT
Angled.JPG
Angled

Working on this for the 5kW version. Looking to have the layout reviewed before prototyping. I may ditch the DRV8353 for the 10kW version and switch to half-bridge drivers. 10kW may be just too much energy for that little DRV IC.
 
Hi, shaman. Lot of work beeing done here. I m always impressed by your designs. Now out of cost cutting and speeding things up interest, wouldn t it be possible to simply buy one of those chinese vesc 6, uprade the fets and the drv to get a usable ebike controller that can take 14s - or am I missing some fundamental issues? The main problem with vesc for ebikes aint current, but voltage. Here in europe, 20a is already plenty, given legal restrictions. But everyone is moving to 48v systems or more for obvious reasons... so all we need is a vesc that can work with higher voltages.
 
@qwerkus

Any of the VESCs or knockoff VESCs won't accommodate the new DRV8353. It's a different footprint and is has different registers for setup.There are many other differences between the old DRV and the new one that would prevent a simple swap but those are told in the datasheets.
 
Also FYI everyone I’m just waiting for JLCPCB to come back online 100%. They went down due to complications with the coronavirus and still don’t have their assembly service back up and running. I have plenty of work for them once they do though
 
shaman said:
@qwerkus

Any of the VESCs or knockoff VESCs won't accommodate the new DRV8353. It's a different footprint and is has different registers for setup.There are many other differences between the old DRV and the new one that would prevent a simple swap but those are told in the datasheets.

Thanks for clearing that up. One question that's bothering the electronics noob that I am: why do vesc require that powerful drv's ? Is that a design choice ? Most chinese motor controller can do with much cheaper drivers, like the lishui lsw 675, which has a true FOC capable STM32 chip. A german dude even wrote an open source firmware for it. Too bad it can only take about 15A.

One other question: what do you think of those chinese reflow stations? Barely the price of a good soldering + reflow station.
 
qwerkus said:
One question that's bothering the electronics noob that I am: why do vesc require that powerful drv's ? Is that a design choice ?

Originally I think Benjamin wanted to reduce the component count and cost to make it easier for users to solder their own vescs from blank pcbs. The drv isn't a particularly powerful mosfet driver and it won't drive more than 6 large mosfets.

The vesc design doesn't require the drv. The larger designs like the 24 fet powervelocity one or the official 18 fet vesc use 3x gate drivers, shunts and current sensors as well as a separate buck converter.
 
One ring to rule them all, eh ? One expensive drv instead of 3 gate sub-cicuits. Makes sense.

@Shaman: can you move the usb port to the other side of the pcb ? I'm pondering where to route the phase wires and it seems the only option is between the jst headers: the board itself is already fairly wide, so exiting cables sidewise is no option on a bike. Top is already full with caps and battery wires (on a sidenote: for ebike use, it would be nice to exit all cables though one end) and on the backside of the pcb is kinf of a nogo, because of the mosfets + heatsink. So if one solders a usb header: no more space for the motor cable exit ! Of course this would mean phase wires running over the drv. Not sure if there's going to be an interference problem, or overheating. No idea how hot these things get.

Also I'd consider adding 1mm margin on both sides so we can slide the pcb inside a box fitted with pcb rails - should there be a box of the right size!

EDIT: maybe this box with a little tweaking ?
 
qwerkus said:
One expensive drv instead of 3 gate sub-cicuits. Makes sense.
Yeah. There's pros and cons to this but I'm hoping it will work for this level of power

under.jpg
CFOC2 cables
under2.jpg
CFOC2 cables front entry

So for the motor phase and possibly even battery cables, I think they can be routed underneath the board and to the front similar to how I did for one of Cheap FOCer 2 beta units. Leaving the USB connector where it is and the cables will come through on the bottom side. This would achieve the a configuration with all wires/cables to the front.

Unfortunately I don't see this controller working with one of those typical extruded enclosures. The PCB needs to mounted from the top side in order to compress the MOSFETs to a heatsink. See my Cheap FOCer 2 thread for a similar mounting scheme.
 
shaman said:
So for the motor phase and possibly even battery cables, I think they can be routed underneath the board and to the front similar to how I did for one of Cheap FOCer 2 beta units. Leaving the USB connector where it is and the cables will come through on the bottom side. This would achieve the a configuration with all wires/cables to the front.

Unfortunately I don't see this controller working with one of those typical extruded enclosures. The PCB needs to mounted from the top side in order to compress the MOSFETs to a heatsink. See my Cheap FOCer 2 thread for a similar mounting scheme.

Ok, good to know about the cable routing. Won t there be interferences when running phase wires so tightly between the fets ?

About the enclosure: how about one of those separated style enclosure ? Lower half of the box would be clamped between the board and an external heatsink. On this model, the central groove should be just about the right width to cover all mosfets.

Same discussion we already had with your beta cheaper focer: worrying about enclosures might not be your top priority, but since you target as low price as possible, it will make a difference. For an instance, all those vesc copies usually cost up to 50% more if you need a matching case, because they don't care about enclosure during the design phase, and than need super expensive cnc'ing or even 3-D printing to make a fitting box. Never really understood this. There are sooooo many cheap boxes out there!

On a sidenote: if it's possible to flip the caps over the pcb, total length should fall below 100mm. Won't please the skaters, but could be a solution if you need a shorter box on an ebike.
 
I don't think interference would bother the FETs. If anything this wold bother the control side since it runs underneath that part of the PCB but I'll be accessing that.

I've looked at the separated style enclosures before. No pre-made designs seem to offer the heat sinking qualities I'm after. They would be functional but just not ideal.

Either way I'm staying away from expensive enclosures. Either I'll find a cheap off-the-shelf solution or design my own extruded split-style enclosure for manufacturing. I just would be forced to buy a ton of it.
 
shaman said:
Either I'll find a cheap off-the-shelf solution or design my own extruded split-style enclosure for manufacturing. I just would be forced to buy a ton of it.

You can also get off-the-shelf extruded enclosures in different lengths if that works. For my VESC design I found an extruded case I liked, but it was too short. I asked the manufacturer about it, and they were able to make me a custom length (longer) version as long as I ordered 20 pcs or more.
 
@Addy yes I’m aware of the option of customizable lengths in extruded enclosures. Unfortunately it’s these extrusion designs that don’t optimize heatsinking for FETs mounted to the bottom. These designs tend to optimize heatsinking for FETs mounted to the side walls instead
 
Little FOCer order.JPG

Ordered some prototypes a couple of days ago for the Little FOCer after JLCPCB came back online 100%. I hope these do better with all the improvements. If they pass functional testing, I'm looking at having them double pulse tested. This will be good to really show the gate drive performance.
 
shaman said:
Little FOCer order.JPG

Ordered some prototypes a couple of days ago for the Little FOCer after JLCPCB came back online 100%. I hope these do better with all the improvements. If they pass functional testing, I'm looking at having them double pulse tested. This will be good to really show the gate drive performance.

jeeeeeez: $20 a piece without shipping and without the expensive parts (mosfet, capacitors). I thought jlcpcb was cheaper. What's your target price with the little focer?
 
@qwerkus the pricing is different for 4 layer boards with ENIG and springing for the “extended components” in SMT assembly. Plus it’s low-quantity. If you want cheap, then check out my cheap FOCer 2 thread.

I’m pretty sure I’ve discussed pricing already but I’m aiming for the little FOCer to be somewhere around $100 to $120 at the moment. This number is not a guarantee and is likely to change. Let’s face it, I’ve never taken a product to market before and there’s much for me to learn still
 
shaman said:
@qwerkus the pricing is different for 4 layer boards with ENIG and springing for the “extended components” in SMT assembly. Plus it’s low-quantity. If you want cheap, then check out my cheap FOCer 2 thread.

I’m pretty sure I’ve discussed pricing already but I’m aiming for the little FOCer to be somewhere around $100 to $120 at the moment. This number is not a guarantee and is likely to change. Let’s face it, I’ve never taken a product to market before and there’s much for me to learn still

Yup saw it. Still 12s so not really useful for most ebikers, including me. Lets see how your little focer goes, I think there is definitely some potential without breaking the bank. Maybe you can check out the homepage of the diy spotwelder guy; I really like his option to chose between assembling the stuff yourself or getting the entire kit. In my humble experience, the expensive part is not so much stocking product, but customer service. A strong diy community like you built on esc8 can help you with that and reduce cost a lot. But it requires affordable products. The higher the premium, the higher the post selling service expected by customers.
 
qwerkus said:
Maybe you can check out the homepage of the diy spotwelder guy; I really like his option to chose between assembling the stuff yourself or getting the entire kit.... But it requires affordable products. The higher the premium, the higher the post selling service expected by customers.

Personally I would MUCH rather spend $20 and not have to deal with SMT assembly myself, I think offering partially populated is a great idea with these.
 
shaman said:
This controller is meant to be a balance of DIY and DFM. The DIY kit should make the DIYers happy since they can solder some components themself and even choose which FETs to use if they wish. Yes figuring out a warranty for this type of thing may be difficult. I'll have to see what sort of warranties are available for other DIY electronics kits. A warranty is more of a business aspect.

@qwerkus I've mentioned a DIY kit before. SMT stuff already done leaving the customer to finish out the rest. Customer support for that may be difficult though. This idea is still being considered.
 
Back
Top