High Power VESC

Discuss hardware related to the VESC such as the NRF nunchuk.
User avatar
TheFallen
Posts: 79
Joined: 26 Oct 2016, 10:49
Location: UK
Contact:

High Power VESC

Postby TheFallen » 04 Dec 2016, 16:45

I needed a VESC that could handle 160Amps rather than the stock 60Amps to drive some 6kW motors. I started with these modified v4.12 VESCs from eBay http://www.ebay.co.uk/itm/VESC-4-12-Mot ... 2375880638 Image

It's made of 2 PCBs, a logic one with the STM32F405 and DRV8302 and a power one with the 6 MOSFETs and 2 sense resistors. I figured that its improved hardware design could handle more current. I was not quite right: Image

So eventually I figured I'd have to design my own. I decided to keep the logic and power sides separate. As I wanted to drive much larger MOSFETs I am using dedicated MOSFET drivers capable of driving 4Amps instead of the DRV8302. I also wanted to upgrade the power sensors to ACS758s. So here's my version of the VESC, largely copied from the v4.12 schematic:
Image

I've not quite finished soldering it up, the example circuit for power supplies used 5.13k resistors, who even sells them? I'll post up a picture of the working board once I've done that and had a go at programming it.

The power stage will be quite odd to most people. I'm taking some inspiration from this post on Endless Sphere: https://endless-sphere.com/forums/viewt ... 41#p828490 and he's a preview of what's to come: Image
Last edited by TheFallen on 06 Dec 2016, 00:11, edited 1 time in total.

arvidb
Posts: 230
Joined: 26 Dec 2015, 14:38
Location: Sweden, Stockholm

Re: High Power VESC

Postby arvidb » 04 Dec 2016, 20:01

Hi, sounds like an interesting variant of the VESC! But there seems to be something wrong with the pictures in your post: I can only see the first one (and the rest just shows the text "Image" instead of the actual image)?

User avatar
TheFallen
Posts: 79
Joined: 26 Oct 2016, 10:49
Location: UK
Contact:

Re: High Power VESC

Postby TheFallen » 06 Dec 2016, 00:12

Ahah! Fixed it, alas Google Photos can be funny some times.

lizardmech
Posts: 171
Joined: 19 Jan 2016, 10:54
Location: Australia

Re: High Power VESC

Postby lizardmech » 06 Dec 2016, 01:47

I'm not sure if that will be able to switch quickly enough, possibly it will work at 10khz or lower. At high frequencies a 20mm trace length between driver and mosfet gate is considered long.

rew
Posts: 937
Joined: 25 Mar 2016, 12:29
Location: Delft, Netherlands.

Re: High Power VESC

Postby rew » 06 Dec 2016, 08:49

the example circuit for power supplies used 5.13k resistors

You just need to calculate the resulting voltage if you take something "close", and see if you can tolerate that. For example, 5.1k is an obtainable value. that is only 0.5% too low. Your resulting voltage will be less than 0.5% too low.

On my version the 12V ended up being 11.89 nominal, 11.91 in practice, or something like that.

User avatar
TheFallen
Posts: 79
Joined: 26 Oct 2016, 10:49
Location: UK
Contact:

Re: High Power VESC

Postby TheFallen » 06 Dec 2016, 10:17

lizardmech wrote:I'm not sure if that will be able to switch quickly enough, possibly it will work at 10khz or lower. At high frequencies a 20mm trace length between driver and mosfet gate is considered long.


10kHz is definitely not high frequencies, I've not done the maths but those traces should be good up to a few MHz.


rew wrote:
the example circuit for power supplies used 5.13k resistors

You just need to calculate the resulting voltage if you take something "close", and see if you can tolerate that. For example, 5.1k is an obtainable value. that is only 0.5% too low. Your resulting voltage will be less than 0.5% too low.


Already done, just waiting some spare time so I can solder on the new value resistors. I've set it so my 3 voltage rails (+3V3, +5V & +12V) are a little above nominal.

User avatar
TheFallen
Posts: 79
Joined: 26 Oct 2016, 10:49
Location: UK
Contact:

Re: High Power VESC

Postby TheFallen » 06 Dec 2016, 23:51

Right, got it all powered up:
Image All the voltage rails show the expected values or near enough so I'm rather pleased.

Unfortunately I forgot to get the JST-PH 5 way connector for the programming header. So I'll have to hold off on programming it for a moment. So I started trying to figure out exactly how I was going to mount the ACS758 hall-effect current sensors. IT turns out they're much easier to mount if they're reverse biased.
Image

So my questions are:
  • Can I easily set up for ACSxxx current sensors and then invert the readings in the firmware.
  • Are there any examples/suggestions I can use for help?

pf26
Posts: 294
Joined: 28 Mar 2016, 14:37
Location: FR Valence

Re: High Power VESC

Postby pf26 » 07 Dec 2016, 15:56

I guess you are measuring phase currents directly with the ACS sensors (and not bottom legs currents as with the shunts) ?
Maybe you want to implement 3 shunts, so you can be compatible with the future release of Vesc ? But I fear your board has provision for only 2 ACS sensors.
Most likely you can invert the reading in the firmware if needed by simply having a negative gain for current measurement in your hardware config file. (I did not check, but if it does'nt work, sure it can be solved).

Your board looks nice. Is it a 2 sides board only ? Will you make it public ?

User avatar
TheFallen
Posts: 79
Joined: 26 Oct 2016, 10:49
Location: UK
Contact:

Re: High Power VESC

Postby TheFallen » 07 Dec 2016, 16:38

pf26 wrote:I guess you are measuring phase currents directly with the ACS sensors (and not bottom legs currents as with the shunts) ?
Maybe you want to implement 3 shunts, so you can be compatible with the future release of Vesc ? But I fear your board has provision for only 2 ACS sensors.
Most likely you can invert the reading in the firmware if needed by simply having a negative gain for current measurement in your hardware config file. (I did not check, but if it does'nt work, sure it can be solved).


The current design is v4.12 compatible, so the two ACSxxx sensors are on the low side of channels A & B. I've already started a new PCB revision that will be v6 compatible, I just didn't want to wait on v6 just in case it never happened. Thanks for the pointer to check the hardware config file.

pf26 wrote:Your board looks nice. Is it a 2 sides board only ? Will you make it public ?


Thanks, I spent a few evenings on making the board look neat. Yes, it's a 2 sided board from Ragworm designed in EagleCAD. I may make it public, but it's really just a v4.12 with the DRV8302 replaced with 3 discrete half-bridge driver chips and 3 power supplies, nothing too revolutionary. It's the same dimensions as a credit/debit card.

pf26
Posts: 294
Joined: 28 Mar 2016, 14:37
Location: FR Valence

Re: High Power VESC

Postby pf26 » 08 Dec 2016, 14:38

I think it is a better idea to use ACS758 sensors on the phase currents, because otherwise, the response time of the current sensor might be too long to get clean and accurate measurements upon low duty PWMs. The datasheet says 3us rise time (IPstep = 60% of IP+, 10% to 90% rise time), but this leave quite a large error and on similar projects, I've had measurements where ACS would react even slower (at least don't expect it to be better than the datasheet statement).
I am not 100% sure, but it seems to me that the V4 code should also work with 2 motor phase measurements instead of on Fets leg. If not, I am interested to rework the code for this (although the timer/adc acquisition is the part I am least familiar with).

Again, your version is very interesting: no more DRV8032, low cost pcb, easy to repair, good cooling of fets, 13s batteries or more...
Question remains on the reliability with all these wires (to the Fets and current sensors) and how to make the Fets wirings short and easy to make (such as the "Lebowski busbar" implementation that you linked for instance.


Return to “Related Hardware”

Who is online

Users browsing this forum: No registered users and 1 guest