Controller options for dual Joby JM1S-2TY

pna

1 µW
Joined
Aug 31, 2013
Messages
3
Hello,

I'm building a vehicle with dual Joby JM1S-2TY motors and am thinking about the controller options. The two obvious candidates I know of are the new 75V, 300A VESC and the Mobipus 72200. They have some pros and cons and questions each, so I thought I'll query the collective brain regarding them and other possible options.

The known (to me) specs of one motor is:

- Winding 2t wye (which will result in a rather low inductance)
- Kv 160 RPM/V (and thus something like 0.060 Nm/A)
- 22 poles, 11 pole pairs
- Nominal bus voltage 45-55 V
- 6000 RPM nominal, 9000 RPM max
- 13 Nm nominal, 20 Nm max
- Mechanical shaft power 8.2 kW continuous, 12.6 kW peak

I haven't found any current specs for this motor/wind combo, but by dividing torque by the (calculated from the specified K_v) torque constant one would arrive at 220 phase amps nominal, 335 A peak. This way the voltage back-calculated from power would be low, around 40 V assuming an electrical-to-mechanical 0.9 efficiency. 0.9 efficiency seems reasonable (if not even conservative) for these motors. If one calculates current through estimated input electrical power divided by specified ~50 V bus voltage one would end up with around 180 phase amps nominal, 280 A peak. But for this to add up with the other specs the K_v constant would have to drop to around 135 RPM/V at full load. Is this reasonable? Expected?

I don't know the exact inductance either. I do have a nice big lab-grade HP 4274A LCR meter, but it's currently stowed away under about 500 kgs of other electronic measuring equipment as a consequence of moving my lab... so currently I can't measure the true inductance with any degree of confidence I would trust. However, toolman2 gives an inductance of a Joby JM1S with (to me) unknown winding as 28 uH. From the other specifications for that JM1S I would guess it's either a 3T wye or a 4T delta wind, and if I understand it correctly the inductance scales like the square root of the turns ratio (ignoring other effects on inductance). This would make my motors have around 20-23 uH inductance. Sounds reasonable?

The main requirement on the controller is obviously that it needs to be able to run the motor. This implies:

- Managing at least 70k ERPM, preferably a bit over 100k
- Being able to run ~20uH motors
- Phase current around 200-250 A cont, 300-350 A peak
- Voltage tolerance of at least around 75 V for me to feel safe. I will probably start running at 12s and perhaps move up to 18s in the future
- Torque control throttle
- Buying something off the shelf. No PCB/electronics designing/building for me. I love doing such things and do it for a living, but that's not the focus for this project.

Nice-to-have features:

- Some kind of dual drive functionality
- Possibility to modify settings/firmware without external assistance
- Configurable and adjustable regen. Possibility to integrate some kind of brake chopper control/interface would be even better
- Possibility to read out all kind of stuff to a custom dashboard
- CAN, RS-485 or similar rugged/industrial/automotive interface with interface specs available
- Robust and reliable
- Efficient
- Small
- Cheap


The two good controller candidates I know of are the 75 V, 300 A VESC and the Mobipus 72200 or perhaps the next size up Mobipus.

Pros for 75/300 VESC:

+ Appropriate current and voltage rating
+ Known to handle high ERPMs
+ Known to handle low inductance
+ Very small
+ Reasonably good thermal interface
+ Rather open. Being able to dig around in and possibly modify the firmware is a huge plus to me.
+ Ben Vedder. Does amazing work and I would like to support him.
+ Potentially very good support
+ The low power VESCs have quite an user base and forum presence
+ Available in Europe

Cons for 75/300 VESC:

- Not quite on the market yet
- Not field proven
- Not as environmentally protected as Mobipus
- Expensive compared to Mobipus, even with the pre-order deal
- Delivery times for the preorders are highly uncertain with the VESC4 and 6 history in mind
- Potentially rather hard-to-reach support
- Supplier without extensive industry experience

Pros for Mobipus 72200:

+ Appropriate current and voltage rating
+ Smallish, but way bigger than VESC
+ Reasonably good thermal interface
+ Seems to be very rugged, fully potted and sealed
+ There seem to be some kind of communications bus interface
+ Cheaper than VESC
+ Field proven, I think(?)
+ Supplier with extensive industry experience (it seems)
+ Shenta. Does amazing work and I would like to support him.
+ samD. Seems to be helpful, communicative and active here on ES
+ Might be possible to get commercial pro support to get up and running

Cons for Mobipus 72200:

- Unclear whether it can do enough ERPMs for me
- Unclear whether it works with RC-style low inductances
- Not available locally in Sweden or in Europe
- A bit unclear what settings I will be able to manipulate without supplier support
- Support might be hard to reach
- Support might be expensive, if at all available
- Not open source firmware
- Control interface specifications are unknown to me, which means at least semi-hard to find on the internet
- No significant online user community/forums


Currently I tend to lean towards the Mobipus, even if I really like the VESC controller, the VESC project and Ben Vedder. I need to straighten out the ERPM and inductance questions regarding the Mobipus, though. @Mobipus official and @samD, do you know the max ERPM and min inductance?

@Trampa/Frank or even Benjamin, if you happen to read this: If I can get a good deal on two 75/300 VESCs I can lend you a Joby JM1S-2TY to test with! =)

And everyone else, any comments? Other suggestions for controllers?
 
I emailed mobipus a couple of months ago asking if it could do 60k erpm in foc. The response was that it was not possible in foc, but possibly in six step commutation. I've never used a mobipus controller personally so don't know if it will work.

Would you be using sensored or sensorless? I don't think mobipus does sensorless either?

75/300 seems to fit your needs, also has documented CAN and UART protocols. There is a slightly cheaper variant of the 75/300 available from http://teamtriforceuk.com/a200s-v1-3/. Current and voltage handling should be quite similar.

The "Nucular" controller is made by a Russian forum member and should handle the motor but I think it will need sensors. Not sure if there is a proper sensorless start algorithm. I think it just transitions to sensorless once the motor gets up to speed.

Another option are the controllers from ASI (accelerated systems). They are foc and can run sensorless or hall sensored. They claim 60uh min inductance but myself and plenty of other forum members have gone lower with no problem (my motor is 15uh phase to phase). Erpm might be a problem. Recommended max erpm is 40k but I've gone to 60k with no issue and Justin of Grin ebikes tested to 90k which seemed to be the max.
 
Well, then Mobipus is out of the game, it seems.

I will run some kind of resolver/encoder position feedback even if the controller-to-be wouldn't need it. The motors has no sensors as it is now, but I'll add some in some appropriate way. Joby says "The JM1S can be used with the RLS RM22 magnetic encoder for position sensing" and they seem to be available from the manufacturer directly for around €70-130 a piece depending on resolution and interface, not at all unreasonable. But what exact encoder I end up with depends on how the mechanics of the motor mounts, transmission and external support bearings and such will turn out.

The RM22 actually look really nice and good value, I might get some even if I end up not using them for the Joby motors... They are magnetic 13-bit absolute position and come in either analog sin/cos outputs, configurable encoder quadrature with index and SSI serial variants. Marcos/paltatech has been doing some resolver work for VESC, I'll look into that before deciding on sensors/encoders. It's usually possible to abuse the MCU SPI hardware to read an SSI encoder, I might try to hack that into the VESC firmware as it already supports AS5047 encoders through SPI. Haven't done SSI on STM32 SPI peripherals previously, though. If necessary for some other controller than VESC I could rig either some kind of 6-step hall emulation with a small cheap FPGA or something off the encoders, or actual hall sensors in the motor if I'm lazy. The fact that I can read and alter the firmware might be the strongest argument for VESC, even if there were other controllers that actually could run the motors...

I'll check out the ASI controllers, didn't know about them, but with Mobipus out it looks like it's going to be VESC 75/300 for me. I've been following the A200s thread on the vesc-project forum, but I think I'll rather go with the Vedder one if I go VESC. The VESC feels rather more sophisticated both hardware- and software-wise than the VasiliSk "Nuclear" controllers, the VESC specs fit my motors perfectly and the openness of the VESC seals the deal, I think.
 
Back
Top