Axiom: a 100kW+ motor controller

@marcos good to see this released! One question though. Why use a digital isolator + non-isolated CAN transceiver instead of an isolated CAN transceiver like the ISO1042?
 
shaman said:
@marcos good to see this released! One question though. Why use a digital isolator + non-isolated CAN transceiver instead of an isolated CAN transceiver like the ISO1042?
Main reason is sourceability, if that is a word. I prefer to avoid part obsolescence and keep supply chain flexible. ISO1050 and ISO1042 come in rather specific packages that you can't easily replace, and they don't make the area much smaller. Its not the main concern, but using a discrete isolator and CAN phy is also cheaper.
Most parts have very similar pin compatible cousins, even the MCU, FPGA and resolver chip, not to mention the opamps, mosfets, LDO and ISO-AMPs. The connectors are usually worth to keep an eye on its availability, they can become the deal breaker.

For every time you can get away with a replacement part I save nearly $200 in tooling costs, so for small batches it pays out.
 
Into the finals!

https://prize.supplyframe.com

Wooot! Happy happy!

Now back to work. Dang SPI driver speedup is taking me ages to get working.
 
Will the Axiom be available as an off-the-shelf controller at all? ie like say the VESC six and 75/300?
 
Yes! Right after we finish validation and we have an enclosure made.
 
shaggythegangsta said:
Hello MArcos, could you please provide some updates on what is going on with Axiom? any updates on ACIM support?

Hi Shaggy,
These last few days we were making the last strides in the hackaday contest, it closes today.
One of the requirements is to make a video so we are spending the bootstrapping stage money in making that video. Crazy video editing days with the amazing folks of Carbono14.

20190927_143251.jpg
(that's me on the right, Lex in the center, Facundo in the back, and Joni the animator next to me)

I would have posponed the marketing for quite a while until we have more base material, but I guess its good to have the marketing side pushed forward.

Behold the intro video!
[youtube]3LqFlcgrxFg[/youtube]
Did you know a voiceiver like that costs $12 on fiverr? Its mindblowing. I didn't even know there is so much stock footage available for free and that a video like this can be made in only a couple of days if you work on-site with them.
No wonder why so many don't bother on making a real hard-engineered product and jump straight to the video :p Kudos for the team!


Besides that I'm mostly providing support to the beta testers and looking for errors or improvements for the next revision. In my free time I try to work on the fpga

Software wise the main task is merging FW/MTPA, I've been developing small additions, but that big one is still not merged and Arlin is going to need it soon. Hopefully Benjamin itself is getting a couple of Axioms next week and we will see some substantial progress in the software side. ACIM support doesn't have a schedule yet, but with the rate of growth I think someone will make it happen sooner than later, things are starting to move really fast.

We are about to test a second batch of DC Links:
DC Link.jpg

And there is one more thing, I've been testing an instrument cluster, should work on any small linux board (Rpi/BBB for example, probably an android tablet):
[youtube]MwYLekmHtdA[/youtube]
 
Good Work guys keep it up, promotional video is really good and also the cluster is very good.
would like to ask also any update on the pcb that would let us adapt let say already built powerstage from some inverter to the axiom logic board?
i think we will never thank enough the talented people like you and many other that are giving there skills to the open source comunity.
 
shaggythegangsta said:
would like to ask also any update on the pcb that would let us adapt let say already built powerstage from some inverter to the axiom logic board?
No updates in that regard. I have the kicad files ready to upload, they didn't change since we defined the general layout and pinout. I just never made the time to create a github repository to make them online, I just emailed the files to the guys that wanted them.

Not sure if it was posted here, but the template looks like this:
axiom template.png

It has the pins correctly labeled and in the right location, as well as the mounting holes locations to make it easy to adapt the logic board to a different powerstage. Most likely a bigger powerstage :twisted:

Thanks for supporting this endeavor!!!
 
I watched one of Arlin's videos on the Axiom he's using in his Honda and he suggests there might be Torque Vectoring possible/coming? I assume Traction Control is already possible through VESC software anyway, but are there plans for Torque Vectoring to be available or is Arlin developing this just for the Honda himself? Thanks,
 
hockinsk said:
I watched one of Arlin's videos on the Axiom he's using in his Honda and he suggests there might be Torque Vectoring possible/coming? I assume Traction Control is already possible through VESC software anyway, but are there plans for Torque Vectoring to be available or is Arlin developing this just for the Honda himself? Thanks,

My Honda at the moment is just 2 motors with open diffs. So torque vectoring is not possible.
But Axiom has traction control and torque vectoring would not be a lot of work to add from that point. You will need torque control of at least 3 points on a vehicle.
 
hockinsk said:
I watched one of Arlin's videos on the Axiom he's using in his Honda and he suggests there might be Torque Vectoring possible/coming? I assume Traction Control is already possible through VESC software anyway, but are there plans for Torque Vectoring to be available or is Arlin developing this just for the Honda himself? Thanks,

VESC has a simple traction control system that should help a lot in straight launches.

Torque vectoring I think implies something more complex than Arlin's setup, requiring separate control of the torque applied in the left and right wheel, the CRX has "only" a front and rear motor, hence it can't steer the body.
So I think it would require at least both motors driving the same axle, for example one motor for the rear left and another for the rear right. This way it can actually sort of steer the car. Of course Ideally you would want one motor per wheel for proper torque vectoring.

I don't plan to implement torque vectoring in the vesc firmware because its not the right place to do so. That's a task for a higher level system in charge of receiving data from gyros, accel, compass, speed, steering angle, etc. This body control unit will send torque commands to each motor controller over the isolated CANbus interfaces. I would love to see Axiom being driven this way, drifting in a powerful EV.
 
Thanks for the responses chaps! I mostly asked only because I was surprised Torque Vectoring was even mentioned. There was no context, but worth asking. I did find some Russian or Czech project a while ago that open sourced a Torque Vectoring EV solution, but I just cannot find it anymore. Way beyond my maths capability even reading it, let-alone implementing!
 
Does this controller have the ability to daisy chain, so if you had 4 motors, and 4 controllers, you could connect the throttle to one acting as the master controller, which would send data to the other 3?
 
atarijedi said:
Does this controller have the ability to daisy chain, so if you had 4 motors, and 4 controllers, you could connect the throttle to one acting as the master controller, which would send data to the other 3?

Everything is possible. The only limit is your imagination and your time. And the flash memory size. And the cpu load.

To be more precise about what's available right now, you could make a CANbus throttle pedal, the pedal will send torque commands digitally to all the controllers in the bus, it could 2, 4 or 120 motors, they will all receive the same command.

If you want a smarter approach, the code allows to connect 1 torque input to the controller A, and over CANbus it will send torque commands to the controller B, but they will take into consideration the speed difference and reduce the torque if there is a loss of traction. Currently this only works for 2 motors. For 4 motors some firmware changes would be needed.
Its not a daisy chain, it would be a bus.

On other news, after 20 Rev0 boards manufactured and shipped, last week we started the fabrication of 10 more Rev1 boards. Main change in the new revision is that current sensing has stronger filtering, we were using 200kHz cutoff and we will be testing 12kHz cutoff so for those who are already testing it would be worth trying the extra filtering. 200kHz allows more accurate motor parameter detection, but it doesn't help during normal operation.
The other upgrade is the isolation ratings. Now full 8mm clearance/creepage is in place, as we had several approaches for 800V.
More info after we receive and test this batch, will take a while.
 
atarijedi said:
Does this controller have the ability to daisy chain, so if you had 4 motors, and 4 controllers, you could connect the throttle to one acting as the master controller, which would send data to the other 3?

marcos said:
If you want a smarter approach, the code allows to connect 1 torque input to the controller A, and over CANbus it will send torque commands to the controller B, but they will take into consideration the speed difference and reduce the torque if there is a loss of traction. Currently this only works for 2 motors. For 4 motors some firmware changes would be needed.
Its not a daisy chain, it would be a bus.


I thought VESC already supports multiple VESC with multiple motors over CAN? Is that not what Vedder describes as Arrays of VESCs or are we limited to an array of two VESCs only in Version 6? How do they implement e.g. 4wd skateboard with 4 VESC & 4 Motors? Are they simply running 4 motors off two VESC in parallel over CAN link between the two VESC?
 
hockinsk said:
I thought VESC already supports multiple VESC with multiple motors over CAN? Is that not what Vedder describes as Arrays of VESCs or are we limited to an array of two VESCs only in Version 6? How do they implement e.g. 4wd skateboard with 4 VESC & 4 Motors? Are they simply running 4 motors off two VESC in parallel over CAN link between the two VESC?

Ah, looking at the code I think you're right. It will take as reference the slowest spinning motor of the whole bus, so in theory you could have 100+ wheels doing traction control with the current code.
 
I think it would be nice to maintain master slave CANbus with Axiom, but I can't see many people needing to use 4 axiom. I guess it would open up to some rather powerful setups though capable of pulling 1200A perhaps?

I must admit to having a rather silly desire to possibly use a two Axiom in master/slave setup for each rear wheel on a car and run two of the larger Neumotors off each one. Could be a fun and much cheaper way to get very high performance instead of going using OEM Tesla & Leaf motors perhaps?
 
the beauty of all this hockinsk, is that its open source. so if people want to try it out using the default code that already supports most of what you're talking about then you're starting from a good spot. a super enthusiastic DIY'r could in theory set up a multi-motor EV, multi Axioms, one gas peddle.. and then fine tune it from there. essentially what you're talking about is adding a VCU to the code base. you can do this because its open source!!

there was mention of ACIM support in previous comments. this is 100% a goal of mine because ACIM motors are super cheap from motor repair shops all over the world (every city has at least one with all sorts of near scrap motors in the salvage bin). if you pickup a new skill of re-winding them yourself for the voltage you want, same power, you have a fairly reasonable motor ultra cheap. is it ideal? no. is it close to ideal? probably not. but will it give reasonable results at a great price for the average DIY'er? yep. i love the re-wound ACIM idea (mounted in a locost7 with an Axiom control board) for an affordable EV. someday this will happen and we'll be there, burn'n rubber and take'n names.
 
True, the opensource spirit allows for anything. I'm not sure how many people that would benefit in the world e.g. creating a Locust7/self builds and altered vehicles etc who will be kit car builders/restorers and ICE mechanics as they just want approved components that simply get cars built to a standard required. As a classic car restorer myself, there's a huge learning curve to doing your first successful EV Conversion and depending on where you are in World, a real lack of approved components too. Open source fills the gap to at least make it all possible, but in UK and I assume most of Europe, it's probably not going to be possible to build your own Electric Car above 48v in any opensource way like ICE for some time. The main hurdle in Europe is the ECE R100.01 inspection typically costs £6-8K. This is why the only industry is EV Conversion because that only requires a basic visual inspection. In some parts of the World, even conversion isn't yet possible legally.
Anyway - a HV Axiom Controller is some step towards addressing all of the above, I can't wait to buy one when they become available, I have many ideas for them.
 
Yeah!!! We are extremely proud to announce that PD-Axiom won the "Best Production" award in 2019 Hackaday Prize!! 🥳🥳🥳

This means that judges think our product is the best preapared for manufacture of the whole 500 projects that entered the competition. That is a huge accomplishment, and we believe it enhances the quality and detail of our design.

Thanks Hackaday!!! 🤝🤝

Also thanks to all our supporters, that were interested in our product. We reached more than 50k views. Thats BIG!!. 💪💪💪

Special thanks to our families, best fans ever!! 👏👏👏

[youtube]f7vcOI7KJ5s[/youtube]
 
Congratulations to the Axiom team for winning that award! :bigthumb:
That is saying a lot about the professionalism of this project. Good luck with advancing the controller and getting it produced!
 
Back
Top