Dual OEM AC EV motors, one drive?

[cajunfj40]

Hello all,

Been keeping up with the development of MPaulHolmes' "Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control" and the gears are spinning merrily in my head.

I'm still a ways off from being able to afford a conversion, but using 1 or more Nissan Leaf motors with this 200kW capable controller looks like a winner from a bang for the buck perspective.

I have what may be a motor or controller theory question, and please keep in mind that I'm a Mechanical Engineer, not an EE.

I would like to run two Leaf motors off one controller, based on my assumption that it doesn't seem likely that the Leaf motor is overbuilt enough to put out much more than 100kW without increasing the voltage above a stock Leaf battery pack level, due to hitting the current saturation limit (or something). A single motor is "enough" (80kW constant rating at 7000RPM will theoretically push my theoretical EV up a 6% grade at 60-70mph), but I want more, of course. If a single leaf motor can handle 200kw short term (or longer with additional cooling), then no need to pair up.

Theoretically, this can be done - at least, it is common in industry to run multiple AC motors off a single VFD. However, the Leaf is an Interior Permanent Magnet type AC motor, thus if I understand correctly it is a synchronous type and operates without "slip", and this complicates the process of trying to parallel two motors.

I have an idea of what I would need to do to make two IPM motors work on one drive, so please follow along and let me know where I made my mistakes, and correct them if possible.

Poking around on the web I found some quite good cutaway views of the Leaf driveline, one included a teardown of the motor itself, showing that the rotor lamination stack is keyed to the main shaft (more like two very long slots on the shaft, 180 degrees apart, and the laminations have an integral pair of tabs that fit into said slots). Thus one could "easily" make a longer shaft and press two rotor lamination stacks onto it, and expect them to be very well aligned.

The stator windings/lamination stack is press-fit into the cooling-channel equipped main motor housing - it is unclear whether there is any keying there. The "simplest" way to pair them up would be to machine an adapter that would connect the "front" of one stator housing to the "back" of the next housing, with the stator slots aligned. Obviously the stator spacing would need to match the spacing of the rotors on the new common shaft.

Re-assemble, and you have a double-length motor with two rotors on one shaft, two stators in paired housings, each with their own cooling jacket and stator wiring connections, and one can use a factory front and rear housing end-plate and bearing setup on the ends. Custom work can be done to add additional cooling to the end-turns if desired.

Is this sufficient to align the two motors well enough to achieve proper operation when the stators are connected in parallel to the controller, or do I need to set up my "stator housing coupler" to be adjustable and play with the alignment until I get some measurable electrical output from the two stator windings to match? If I need to play with it, what sort of electrical output do I look for? I thought I could set it up and spin the rotor and watch the AC waveforms from the two stators on a scope and adjust the two housings relative to each other until they match as closely as possible. Or, I could run the paired motors off the controller at a low current and adjust the relative position of the stators until max torque is reached - but this is a much harder setup.

On the electrical side of things, once the two motors are aligned, do I just hook up the two stators in parallel - making sure the correct phases are connected so the unit doesn't fight itself? Do I need to carefully adjust the wiring resistance so they are equal, or is inductance more important, or both? How critical is this adjustment?

I would then use only one encoder/resolver for the paired motor.

Thoughts?

This seemed more straightforward than a custom gearbox to accept two Leaf motors and combine the output into a single output shaft - trying to align the two rotors through the gears adds the problem of slop and having to re-do it every time you pull a motor or strip a gear.

Alternately, if the two motors cannot be paralleled on the same drive for electrical reasons, can the twinned motors be run from a pair of inverters in master-slave mode, assuming the motors are aligned well enough that the single resolver is accurate enough to provide rotor position information for both motors? How about if two standalone motor/encoder/inverter sets are mechanically connected? What would be awesome is a way to run two stock Leaf inverters this way...

Note that both ideas can be extended to pair up two or more of virtually any available wrecked OEM EV motors and possibly their inverters, depending on motor mechanical setup and inverter electronics/control scheme.

Thanks!

[freebeard]

http://techon.nikkeibp.co.jp/english/NEWS_EN/20110724/193591/?SS=imgview_e&FD=47620101

The other thread you might look at is Hot-rodding the Toyota MGR. The OP, e*clipse, is working on combining two of the MGR in a custom case. It's rotor splines onto the shaft. He might be interested in or have answers to your questions.

Arcimoto use two electric motors in front-wheel drive. It's not clear to me if there is a differential or the controller[s] provide the differential action, driving each wheel separately.

To what will you custom shaft connect. IOW base vehicle, front or rear drive, lateral or longitudinal, etc. Have you considered putting one MGR on each end of the car for 4WD?

[cajunfj40]

Hello freebeard,
I've been keeping up with that thread as well. Intended application is an old 4x4 FJ-40 Toyota Land Cruiser. If axle-mounted motors get a sufficient torque/speed spread without having to change gears by the time my budget is available, I could go 1 motor per end or per wheel, but at this time it is far simpler to run everything through the transfercase and driveshafts to the existing front/rear live axle setup.

For off-road application, the setup to meet or exceed is ~6500 ft-lbs at the axle (with locking diffs, that can be all at one wheel, or divided amongst all 4, depending on how many are in the air/on a slippery surface) in low range. (250 ft-lb gas engine, 3.555:1 first gear, 1.96:1 low range, 3.73:1 axle gearing - total of ~26:1 gearing). Gearing lower (which is cheap down to about 66.5:1) becomes less important so long as sufficient torque is available. Gas engines need gearing to keep enough torque available at slow speeds without stalling the engine or burning up the clutch or torque converter. With insufficient gearing, you end up going too fast sometimes, which means more things busted/broken/bent. With an electric motor being a "constant torque" source below a certain RPM, stalling isn't an issue.

For on-road application, acceleration "feel" must meet or exceed the stock 125hp/250 ft-lb gas engine with 4-speed truck-type manual trans and average-ability driver. Pretty low bar to clear. If it isn't fun, it isn't worth it.

Self-contained range requirements are low - a stock used (cold-climate, so 9-10 bars showing on the battery still) 2011 Leaf exceeds my needs at present, even in winter. I'll figure out how to get it to remote off-road locations and how to recharge it there later. It'll be a fun commuter first.

[freebeard]

Congrats on having an FJ-40. I have some friends that keep saying they're finally going to sell the family Scout II.

You may be able to extrapolate from putting an AC motor on a VW transaxle. First gear is useless. Second is a good low end, for pickup. You can drive in 3rd all day if you don't mind oozing away from stop signs. 4th is like an overdrive, but the motor hits the knee in the torque curve and chokes.

I'm no help at all when it comes to machining shafts and adapter plates.

[thingstodo]

Originally Posted by cajunfj40

Hello all,

Hi!

I have what may be a motor or controller theory question, and please keep in mind that I'm a Mechanical Engineer, not an EE.

I would like to run two Leaf motors off one controller

As far as I know, Field Oriented control is done per motor. HPEVS has a dual 35 KW motor, and it uses 2 Curtis controllers, one gets the throttle signal and the other follows in torque mode. I have no details - but if they could do it with one controller it would be much more competitive in price.

Theoretically, this can be done - at least, it is common in industry to run multiple AC motors off a single VFD.

In my experience, multiple induction motors on one controller are done in Volts per Hertz mode. Every output voltage has a corresponding frequency output. It gives you more than one motor`s output in vector mode, but less than a controller per motor.

To my knowledge (please - I`d love to be wrong on this one!) no one does sensored or sensorless vector control, Field Oriented Control, etc with more than one motor per controller.

However, the Leaf is an Interior Permanent Magnet type AC motor, thus if I understand correctly it is a synchronous type and operates without "slip", and this complicates the process of trying to parallel two motors.

Actually, it`s worse than that. The control is like permanent magnet for low speed, and has more contribution from reluctance as the speed rises, so the angle that the current is driven changes with speed and with load (!!) Lots of calculations done by the controller to figure out what signal to send out for maximum torque.

So making both motors put out full torque with the same signal when they are coupled is quite a challenge. I won`t say impossible .. but I will say much more math to prove one way or the other than I`m up for!

Thus one could "easily" make a longer shaft and press two rotor lamination stacks onto it, and expect them to be very well aligned.

I think this is the crux of the issue - figuring out what to measure to make sure that they are `well aligned`.

On the electrical side of things, once the two motors are aligned, do I just hook up the two stators in parallel - making sure the correct phases are connected so the unit doesn't fight itself? Do I need to carefully adjust the wiring resistance so they are equal, or is inductance more important, or both? How critical is this adjustment?

As I understand it ... to get full torque, the resistance (wire length, mostly), inductance (cross sectional area across steel, tightness of the wire coil, consistency of the laminations), and reluctance (resistance to magnetic flux .. sort of .. dependent on consistent laminations of the steel, thickness of the varnish between laminations, etc) would all have to match at the same time. I believe this would be a challenge.

Alternately, if the two motors cannot be paralleled on the same drive for electrical reasons, can the twinned motors be run from a pair of inverters in master-slave mode, assuming the motors are aligned well enough that the single resolver is accurate enough to provide rotor position information for both motors?

Yes.

[cajunfj40]

Hello again freebeard and thingstodo,

freebeard, I currently don't have an FJ40 - haven't for over a decade. See, I had one as my first and second vehicles (rolled the first one, put over 100K miles on the second) and then mostly parked it after graduating from University and needing to work. Eventually I moved, and took it with me, and by then it needed more than just some repair. So I took it apart to "restore" it but had no money, so it got piled up in the back yard and sat for ~7 years, and then I sold it and started a family and all that. I want back in, but I DO NOT WANT the ~12mpg mine had or the ~20mpg a diesel would get. A diesel conversion done right would likely be close to the cost of an AC EV conversion, if I can use junkyard OEM EV parts. Plus I want to be different. :-) Scouts are cool, too, but also suffer from Demon Rust - and I imprinted on FJ-40's, so they're not really what I want. if one has the cash, fiberglass body parts are available for both, and Aluminum for the 'Cruiser.

As for how the AC motor feels through a transaxle, that is good to know. Do you have an HPEV's AC setup? When you mention the torque hitting the knee and the motor choking, I recall from the peak power graphs for their motors that HP does actually peak on those, with the downslope almost as steep as the upslope, whereas on the OEM EV motors I'm thinking of (Leaf, Volt, etc.) they do a flat constant power region after the flat constant torque region. Should help some. There'll be a point at which it makes sense to shift up after the torque falls off too much, but it'll be a wider spread. I want liquid-cooled AC so I can sit there "stalled" on a rock without burning it up, and if it can keep liquid in I can seal it well enough for a dunking. Air-cooled brushed DC would be relatively easy/cheap, but keeping water out is much more of a challenge.

thingstodo, thanks for the insights! For aligning the two motors, I considered the rotors aligned by way of their built-in keys good enough, with the rest of the work being in aligning the two stators. As for possibility, AMR appears to sell "dual core" motors with single inverters, and IIRC those are IPM type REMY motor cores. No idea of the math needed, though! Sounds like trying to run two IPM or SPM or similar motors off the same drive may not be worth the hassle.

How much less critical is it to get two stator windings for induction machines equal in terms of resistance/inductance when running them in parallel off the same drive?

You said it is possible to run two OEM AC motor/drive combos in master/slave mode on a common (or geared/belted together) shaft. Any ideas as to how to control those drives? Mostly thinking OEM EV drives here - for common junkyard availability. For the aftermarket Curtis controllers, and many common industrial VFD's, they have that feature built in - just connect them up and put in the right settings. Not sure that is the case for OEM EV drives - and it seems like it would be an extra-cost thing that wouldn't be worth it to the OEMs to implement.

Hmm. I guess I could send the throttle signal to a separate "controller controller" and have it divvy up the torque demand to the two OEM drives, possibly with a current sensor on a phase of each motor to have a feedback for how well balanced the setup is running, if needed. Or just twin the throttle output and send it to both controllers at the same time, maybe monitor the current on/temperature of each motor for a while in early days to check the balance and see if it needs adjusting.

I've got lots of time to plan this out as the budget builds all too slowly. I'll probably end up buying a used Leaf to replace my currently dissolving ICE car as a commuter (different budget - I need to commute, after all!) for my first EV, as my toy budget may get big enough to buy a running/driving FJ-40 before it gets big enough to buy enough good parts to build one plus the EV bits.

Thanks!

[thingstodo]

*WARNING * another long-winded post!

Originally Posted by cajunfj40

For aligning the two motors, I considered the rotors aligned by way of their built-in keys good enough, with the rest of the work being in aligning the two stators.

If you are using separate controllers, I don`t think you need any electrical alignment. If you are still looking to do one controller - I would be interested in your results.

As for possibility, AMR appears to sell "dual core" motors with single inverters, and IIRC those are IPM type REMY motor cores. No idea of the math needed, though! Sounds like trying to run two IPM or SPM or similar motors off the same drive may not be worth the hassle.

I didn`t know about AMR - thanks for the info.

I don`t know of any research on-going at the moment to drive multiple motors of any type in Field Oriented Control or sensorless vector. If the nerds are not arguing about it - there must be an obvious problem or two.

How much less critical is it to get two stator windings for induction machines equal in terms of resistance/inductance when running them in parallel off the same drive?

You can run the motors in parallel from one drive. I have 4 running in parallel on a machine that runs on rails. There is nothing critical about it - we ran one at a different horsepower for a month when we ran out of spares. But it is volts per hertz. You don`t maximize torque per motor, you waste a bit of power, the motors heat up more than they *REALLY* need to. It DOES give me 4 wheel drive, one motor and gearbox per wheel, with no need for limited slip or differentials. Each motor gets the same frequency, so if it has no traction .. even if it`s up in the air ... they spin the same speed and whichever wheels have traction take the available power. The others just spin the same speed .. or about the same speed. A motor under load spins slightly slower than a motor spinning unloaded.

You said it is possible to run two OEM AC motor/drive combos in master/slave mode on a common (or geared/belted together) shaft. Any ideas as to how to control those drives?

The throttle signal is sent to the first controller. I believe it is a Curtis controller. It outputs a torque signal to the second controller, which follows along.

With a modern dual-signal hall effect sensor, you could .. in theory .. send the same throttle signal to each controller. The controllers *COULD* fight if one of them saw the signal as slight regenerative braking while the other saw the signal as slight acceleration .. but the calibration curves for the throttle should take care of that. There`s always a bit of deadband around `coast`. That should take care of it.

Mostly thinking OEM EV drives here - for common junkyard availability. For the aftermarket Curtis controllers, and many common industrial VFD's, they have that feature built in - just connect them up and put in the right settings. Not sure that is the case for OEM EV drives - and it seems like it would be an extra-cost thing that wouldn't be worth it to the OEMs to implement.

The OEM EV drives (besides Tesla) appear to use CANbus to give them their throttle reference. Sending the same signal to multiple controllers should be quite easy to do.

Going back to a Jack Rickard (EVTV) video - you would need some sort of limited slip arbitration, or ABS maybe, so that you do not break loose one rear tire while accelerating through a turn.

Hmm. I guess I could send the throttle signal to a separate "controller controller" and have it divvy up the torque demand to the two OEM drives, possibly with a current sensor on a phase of each motor to have a feedback for how well balanced the setup is running, if needed.

EVTV has such a controller - they call it GEVCU - and it`s open source so you can see how they do it, or write it yourself.

Or just twin the throttle output and send it to both controllers at the same time, maybe monitor the current on/temperature of each motor for a while in early days to check the balance and see if it needs adjusting.

As I guess I did not really explain above - the throttle is a torque setpoint. If you lose traction, it still tries to drive the motor to the requested torque. That`s what the ABS is for. Or limited slip algorithmn.

I've got lots of time to plan this out as the budget builds all too slowly. I'll probably end up buying a used Leaf to replace my currently dissolving ICE car as a commuter (different budget - I need to commute, after all!) for my first EV, as my toy budget may get big enough to buy a running/driving FJ-40 before it gets big enough to buy enough good parts to build one plus the EV bits.

If I lived in the US, I would purchase leaf wrecks for the EV parts. $5K appears to be the going rate. You get the motor, the battery pack, the charger and the DC to DC converter. At least $10K in parts if you were to buy them. If someone figures out the CAN commands for the BMS and the controller - that`s an extra bonus. I think you have room in that beast for 3 leaf motors coupled end to end. You could keep the rest of the drive-train as is and twist drive shafts if that`s what you wanted to do ...

Making things work with OEM parts has some advantages:
- if you sell it eventually, the new owner can find parts and has confidence that he can
- you can go and buy a part off the shelf to get back running ... if you need to .. or you can get another wreck or buy something from a wrecker
- Comparatively little issues with `that company is out of business, or the part is obsolete so you can`t fix it`.

The alternative, for me, is Open Source. I`m not dedicated to building the boards, or soldering, or writing the program myself. But if you have the schematics you can troubleshoot it yourself. If you locate something that died - like an IGBT - you can buy a replacement and put it in, or get someone at a maker space to put it in. You have the power!

The software may not be as bullet-proof as an off-the-shelf product .. but it may be better .. depending on the guys doing the open source. If you see some new feature in a car in 2 years .. and you want it badly enough .. you can add it!

Boy - I do rant on, don`t I?

I will follow your build with interest!

[cajunfj40]

Hello again thingstodo,

*WARNING * another long-winded post!

No worries, that's basically my default mode...

If you are using separate controllers, I don`t think you need any electrical alignment. If you are still looking to do one controller - I would be interested in your results.

I agree. I'd also be interested in my results - it'll take a while to get there if I try it! I'm a MechE, so this isn't my natural turf.

I don`t know of any research on-going at the moment to drive multiple motors of any type in Field Oriented Control or sensorless vector. If the nerds are not arguing about it - there must be an obvious problem or two

.

I've seen a few papers out on the 'net, but without the EE background I can't tell what's BS and what's not.

You can run the motors in parallel from one drive. I have 4 running in parallel on a machine that runs on rails. There is nothing critical about it - we ran one at a different horsepower for a month when we ran out of spares. But it is volts per hertz. You don`t maximize torque per motor, you waste a bit of power, the motors heat up more than they *REALLY* need to. It DOES give me 4 wheel drive, one motor and gearbox per wheel, with no need for limited slip or differentials. Each motor gets the same frequency, so if it has no traction .. even if it`s up in the air ... they spin the same speed and whichever wheels have traction take the available power. The others just spin the same speed .. or about the same speed. A motor under load spins slightly slower than a motor spinning unloaded.

How does the picture change when the motors are mechanically connected? Can you put it into a torque-vector mode of some sort, perhaps by adding a resolver/encoder, to get the maximum torque available and less heat? Does it require a bit more exacting "matching" of the two motors?

With a modern dual-signal hall effect sensor, you could .. in theory .. send the same throttle signal to each controller. The controllers *COULD* fight if one of them saw the signal as slight regenerative braking while the other saw the signal as slight acceleration .. but the calibration curves for the throttle should take care of that. There`s always a bit of deadband around `coast`. That should take care of it.

The OEM EV drives (besides Tesla) appear to use CANbus to give them their throttle reference. Sending the same signal to multiple controllers should be quite easy to do.

That's close to what I was thinking with "twinning" the throttle or similar.

Going back to a Jack Rickard (EVTV) video - you would need some sort of limited slip arbitration, or ABS maybe, so that you do not break loose one rear tire while accelerating through a turn.

This is thinking way beyond my ability to actually do something about, but a lookup table based on steering input, throttle input, wheel speed sensor input and possibly a suspension angle sensor input would give the necessary ratios for splitting up the torque commands to the 4 wheels, if using separate motors. No need if they are on the same shaft, though.

EVTV has such a controller - they call it GEVCU - and it`s open source so you can see how they do it, or write it yourself.

I'll just have to poke around once I am closer to being able to buy parts and see if anyone has sniffed the appropriate CANbus commands to run the donor inverter(s) I decide on.

As I guess I did not really explain above - the throttle is a torque setpoint. If you lose traction, it still tries to drive the motor to the requested torque. That`s what the ABS is for. Or limited slip algorithmn.

Is this true for OEM EV drives, then? If so, I still only need to worry about keeping the two motor/drive setups balanced enough that the "deadband around 'coast'" can handle it.

If I lived in the US, I would purchase leaf wrecks for the EV parts. $5K appears to be the going rate. You get the motor, the battery pack, the charger and the DC to DC converter. At least $10K in parts if you were to buy them. If someone figures out the CAN commands for the BMS and the controller - that`s an extra bonus. I think you have room in that beast for 3 leaf motors coupled end to end. You could keep the rest of the drive-train as is and twist drive shafts if that`s what you wanted to do ...

I am definitely considering the "whole wrecked Leaf" donor route. Not sure I want to go into the "parting out a Leaf" business, as I've no space to store more than one donor at a time. I should be able to sell the rest of each carcass to a salvage yard, though. There's a thread, I think on diyelectriccar.com, with a guy who tore an entire Leaf driveline out and figured out how to get it running without the rest of the car, and paring it down to the minimum. No special CAN controller.

Making things work with OEM parts has some advantages:
- if you sell it eventually, the new owner can find parts and has confidence that he can
- you can go and buy a part off the shelf to get back running ... if you need to .. or you can get another wreck or buy something from a wrecker
- Comparatively little issues with `that company is out of business, or the part is obsolete so you can`t fix it`.

Much like using newer, common Chevy/Dana/Etc. parts to modify an old FJ-40.

The alternative, for me, is Open Source. I`m not dedicated to building the boards, or soldering, or writing the program myself. But if you have the schematics you can troubleshoot it yourself. If you locate something that died - like an IGBT - you can buy a replacement and put it in, or get someone at a maker space to put it in. You have the power!

The software may not be as bullet-proof as an off-the-shelf product .. but it may be better .. depending on the guys doing the open source. If you see some new feature in a car in 2 years .. and you want it badly enough .. you can add it!

I like the idea quite a bit, so long as the 'hassle-factor' of fixing code doesn't overwhelm the rest of it. I like the "built, not bought" ideal, but when time is at a premium and I enjoy the mechanical stuff far more than the code/electrical side... We'll see.

Boy - I do rant on, don`t I?

As I mentioned above - long-winded is my default mode, so no worries!

I will follow your build with interest!

When I finally do get the rig, I'll try to put a build thread on here somewhere. Might start outside the EV forum at first, depending on budget.

I'll likely keep asking questions here and there on this forum as I get ideas I want to check.

Thanks!

[freebeard]

Originally Posted by thingstodo

With a modern dual-signal hall effect sensor, you could .. in theory .. send the same throttle signal to each controller.

If you're talking about a Ford Fusion/Tesla type throttle, they report over two different voltage ranges, and use the difference as a sanity check.

I agree. I'd also be interested in my results - it'll take a while to get there if I try it! I'm a MechE, so this isn't my natural turf.

That's more time to watch back episodes of EVTV (at 1.25 speed on Youtube).

There's a thread, I think on diyelectriccar.com, with a guy who tore an entire Leaf driveline out and figured out how to get it running without the rest of the car, and paring it down to the minimum. No special CAN controller.

Linky, please?

[oil pan 4]

To get full torque on a 3 phase motor rotor inductance and reactance have to be equal, this usually happens at 20% to 30% rotor to stator magnetic slip speed.