Serious FOCer (84V VESC 6 based controller)

shaman

1 kW
Joined
Dec 10, 2018
Messages
343
Location
Fort Worth, TX, USA
TO247.jpg
TO-247 version

View attachment 2
vertical 12 FET TO-220 version

View attachment 1
vertical 12 FET TO-220 version w/aluminum channel heatsink

12 FET with Sink side.PNG
vertical 12 FET TO-220 version w/aluminum channel heatsink

View attachment 3
vertical 6 FET TO-220 version

View attachment 4
horizontal TO-220 version. Image for reference only. Layout does not yet accommodate horizontal TO-220 package.

Announcing the Serious FOCer (working title)

Features:
  • 15V to 84V operation
  • Expected power of approx 5kW with TO-247 FETs and good heat sinking
  • Has primary features of VESC controllers (FOC, configurable current/temperature limits, throttle response, dual motor operation, ect.)
  • Different versions for different FET packages
  • TO-247 version for higher current capability
  • Single-sided dual-layer PCB
  • Small 45mm x 101.5mm board size
  • Each switch node has it's own bulk cap, pwr, gnd and phase cable holes. Reduces need for beefing up long traces for increased current handling capability
  • Utilizes DRV8353RS to achieve higher operating voltage, superior fault detection, and small footprint. Also allows for tailored slew rate control to minimize ringing/spiking in the power stage
  • Easy On/Off capability with switch of choice
  • Wireless Bluetooth transceiver module can be easily added for wireless interface with VESC Android App. Allows for easy configurations of different speed/power profiles that can be selected at will.
  • Design will be open sourced after verification
  • Designed in KiCAD for flexibility and compatibility with the open source community

Donations:
Open Source R&D isn't free! Any help is appreciated!
https://www.paypal.me/shamansystems?locale.x=en_US

This is the higher performing controller in comparison to my Cheap FOCer controller that is about to go into beta testing. All lessons learned from the Cheap FOCer development has and will go into this new controller design. I am open to any suggestions that will help make this controller great for this community!
 
Woot!!! Yes to TO247 based VESC!!!
 
subscribed. Do you have an approximate BOM cost yet?

Was actually waiting for someone to do the hard work on a cheap and higher voltage VESC6 design :D
Once you make this open source (and I have some spare time) I'll probably look into redoing the layout so it fits in the BBSHD like the original controller. Will be open source too ofc.
 
BotoXbz said:
Do you have an approximate BOM cost yet?

Not yet and it will depend on if you use TO-247s or the TO-220s. TO-247s are much more expensive and will be the majority of the BOM cost. FETs aside, the BOM is pretty cheap. I'll try to gather that number soon for kU pricing and single unit pricing.
 
spinningmagnets said:
Adaptto Mini-E

I looked at a teardown of this controller just now. I like the orientation of the FETs. Someone also suggested this kind of orientation on my Cheap FOCer thread. I think I might use that sort of FET layout for a 12 FET version.
 
The VESC 6 uses individual current sense amplifiers instead of the built-in amplifiers of the DRV.

I've set up the DRV8353RS as per the recommendations from the data sheet. It does not call for a low pass filter on the current sense inputs or phase sense. While adding them may not hurt, it may not be necessary either.

Is there any information to support adding them?
 
shaman said:
Is there any information to support adding them?

I've found the current filters help with sensorless start and at low currents where the snr is poor. I have a vesc set up where the amplifiers saturate the adc at 495A. Without the filters at 2A (no load) the motor current just jitters around 0. Sometimes even goes negative. With the filters it is still noisy at such a low current but reads correctly.

Apparently new code is needed to take advantage of phase voltage filters but it should help with low speed sensorless.

Phase voltage filters will interfere with 6 step commutation and the current filters will break foc inductance measurement. So people might need to solder/desolder the filters as needed or you could switch the filters on using mosfets like on the official hardware.

Benjamin talks about this in his 75/300 testing video and on the 75/300 thread over on vesc project.
 
Thanks for the info in this. I'll go see what Benjamin is talking about for the filters.

I'm not sure I like breaking the inductance measurement and messing with the 6 step commutation.
 
I looked at a teardown of this controller just now. I like the orientation of the FETs. Someone also suggested this kind of orientation on my Cheap FOCer thread. I think I might use that sort of FET layout for a 12 FET version

For about a year, the Adaptto controllers were the "next big thing", and they immediately drew a passionate following. However, customer support was weak, as they didn't have a real dealer network to handle the customer issues, and they did not want to handle customer issues themselves.

They would have a burst of production activity, making a large batch of controllers, then a burst of sales activity (globally), then they would go on an extended vacation.

In fact, I believe some of the interest in your Cute Little FOC'r is due to the groundwork laid by Adaptto, and then they squandered it. Adaptto's are small for the amount of amps they produce. As Luke has been saying for a long time, efficient components don't make as much waste heat in the first place.

Adaptto's are expensive, but the FOC'r series will be MUCH more affordable.
 
spinningmagnets said:
customer support was weak, as they didn't have a real dealer network to handle the customer issues, and they did not want to handle customer issues themselves

I wonder what the common customer issues were? If they were setup/operation related or if they were functional/quality issues?

spinningmagnets said:
Adaptto's are expensive, but the FOC'r series will be MUCH more affordable.

Indeed. The 2-layer single-sided PCB makes manufacture and assembly cheaper. It's also all a single board solution vs Adaptto's multi-board solution.
 
:!: Project Update

Added 12 TO-220 FET and 6 TO-220 FET versions to the original post.

The 12 FET version offers higher power capabilities with only 22% increase in board size. This is different than the usual method of simply arranging the parallel FETs right next to each other in a single row causing board length to nearly double. My arrangement results in a board that is the same length and only slightly wider. Aluminum channel is used to act a heatsink for the 2 rows of FETs. The aluminum channel can then be coupled to an aluminum enclosure for increased heatsinking capability.
 
I apologize in advance for my ignorance about electronics. Mea Culpa.

For an 84V max, I am assuming 4.2V (per cell) X 20S = 84V, so we are talking about 48V-84V, etc...

Concerning the "package" size of mosfets (220, 247, etc), are there any efficiency gains to be had if I was to acquire one of these boards (minus FETs) and then supply and solder my own FETs, with a max of 60V? (I am a fan of 13S/14S, with 14S supplying 58.8V max on a full charge)
 
No apology necessary. Ask any time.

Generally a good 60V FET will have lower Rds(on) than a good 100V FET of the same package. The lower Rds(on) will mean less energy lost to heat and therefore will be more efficient. Generally lower voltage FETs will have lower Rds(on) values for the same package.

I wouldn't recommend running 58.8V on a 60V FET. That's cutting it too close and doesn't leave room for voltage transients(spikes). However 58.8V on a 80V FET would be fine. Ideally you want to operate at or under 80% of the max FET voltage.
 
So I take it looking at your renders there are no bus bars in this design, you'll be using direct connection at each FET set to positive and ground?

Very impressed with the size your packing this into, 100x50 means you can hide it under your BB and it's essentially invisible!
 
wil said:
So I take it looking at your renders there are no bus bars in this design, you'll be using direct connection at each FET set to positive and ground?

Yes this is the case. Also the legs of the FETs will be bent to where each cable comes in to the board to help make a solid connection. This eliminates the need to beef up any significant lengths of track to increase current handling. One will have to split battery positive and negative cables into 3 cables each. This shouldn't be to hard and cable is cheaper/easier than bus bar.

This may cause some stray inductance but I'm hoping to minimize it's effects by dialing in the rise/fall times via the DRV. One can adjust how many milliamps are being used to drive the FETs through a particular register of the DRV.
 
shaman said:
wil said:
So I take it looking at your renders there are no bus bars in this design, you'll be using direct connection at each FET set to positive and ground?

Yes this is the case. Also the legs of the FETs will be bent to where each cable comes in to the board to help make a solid connection. This eliminates the need to beef up any significant lengths of track to increase current handling. One will have to split battery positive and negative cables into 3 cables each. This shouldn't be to hard and cable is cheaper/easier than bus bar.

This may cause some stray inductance but I'm hoping to minimize it's effects by dialing in the rise/fall times via the DRV. One can adjust how many milliamps are being used to drive the FETs through a particular register of the DRV.

Hmmm seems like the opposite of the recommended laminated bus bar. We'll see what happens when you build it...
 
thepronghorn said:
seems like the opposite of the recommended laminated bus bar.

Yeah probably. But bus bars sized for PCBs like this aren't easily available to me or anyone else on the DIY level. You can get by with lengths of solid core copper wire and stuff but I'm trying to avoid anything like that.
 
shaman said:
thepronghorn said:
seems like the opposite of the recommended laminated bus bar.

Yeah probably. But bus bars sized for PCBs like this aren't easily available to me or anyone else on the DIY level. You can get by with lengths of solid core copper wire and stuff but I'm trying to avoid anything like that.
Completely agreed to this - real bus bars are not in the realm of something I can realistically accomplish, thickened traces with solder or wire I can.

For a DIY controller I think this is a good approach to take.
 
So in shopping for various FETs to use in my FOCers, I've looked at these FETs here

https://lcsc.com/product-detail/MOSFET_Infineon-Technologies_IPT015N10N5ATMA1_Infineon-Technologies-IPT015N10N5ATMA1_C108964.html

They have really small Rds(on) and decent gate charge. Of course they're SMD which I've talked crap about due to the harder ability to heat sink. LCSC has them for a good price. I'm thinking I could do a 12 FET version that has these and do a sort of heat sinking through copper bar like @galp did on his BESC. It won't be as DIY friendly but could offer high power density. What do you guys think? Worth it or no?

Edit: found these too for even better price, same Rds(on), lower gate charge, and I could run 18 of these. Mouser just wont be in stock with them for a while.
https://www.mouser.com/ProductDetail/ON-Semiconductor-Fairchild/FDBL86062-F085?qs=sGAEpiMZZMshyDBzk1%2FWi%2FD7Em5shE8q2Wwbr9sX%252Bqck5TZDb7JrYQ%3D%3D
 
shaman said:
So in shopping for various FETs to use in my FOCers, I've looked at these FETs here

https://lcsc.com/product-detail/MOSFET_Infineon-Technologies_IPT015N10N5ATMA1_Infineon-Technologies-IPT015N10N5ATMA1_C108964.html

I've soldered a few vesc boards with the IPT015N10N5ATMA1 now. The copper bars do a great job of absorbing heat and permit heavy acceleration. Replacing a mosfet was an absolute nightmare though. Need a lot of heat with all that copper.

Unless I actually needed every bit of power possible, I'd settle for to-247. And save a bit of money in the process. IRFP4468 is not expensive from lcsc.
 
Back
Top