[WIP] Smart Roadster (with BMW i3 motor and inverter)

[janosch]

I am in beautiful Cornwall this week, my friend Chris is embarking on converting his '04 roadster!

He has stripped the car of the ICE components already and will have a new subframe made once we spin the motor on a bench successfully.

Battery system and BMS are undecided, might be the first application of the flying ADC BMS?! Let's see how it matures over the summer.

[janosch]

Mocked with rear crossframe for suspension in.

[Pete9008]

Nice

Will be following with interest as I think you will be finished before me!

Do you need to flip or reverse the I3 motor?

Been wondering about doing a new subframe too. Are you confident that the DVLA won't treat it as part of the monocoque (I know they shouldn't but they seem a law unto themselves on these things!)?

[CJW]

Planning for the motor to be installed in same orientation as in the i3.

It's quite a tight fit front to back, looks like some of the motor castings for the air conditioning motor/compressor on the i3 will need to be trimmed back. The back end of the gearbox is also very close to the central bolt nuts on the rear x-frame. With careful positioning, reversing the rear x-frame bolts, and possibly a modification to the x-frame it should be doable.

The rear subframe and suspension will all be original. Having it zinc thermal sprayed, painted, and will build an adaptor frame that links the i3 motor to the original motor mount positions. Will share CAD file for that when it's done.

Hoping to use the original motor mounts, reinforced with Powerflex inserts. Will keep an eye on them and have custom ones made if necessary as the torque from the i3 I think is around three times as much as the original motor .

[Pete9008]

Sounds good!

Bear in mind that the motor mounts see driveshaft torque reaction not motor/engine torque. The higher ratio of the Roadster box 1st gear balances things up a bit and the mounts will have been designed to handle that. The i3 is likely still higher but probably not by enough to worry the uprated bushes.

[janosch]

Ok,

I set up a precharge circuit, connected 24V and wanted to spin the motor with manualid, but not getting a PWM noise.

udcsw, udcmin I set to 0V.
I gave it manualid 1A and 4A in firmware 5.24 and 4.97 (sine and foc).

Old firmware shows 200V no matter if DC is connected or not, makes sense, the i3 sensor scaling wasn't in that firmware yet. Same for temperature which reads too high.

New firmware shows 0V and 24V when its there.

I can get into manualRun but it won't do PWM.

Any thoughts on how to give this a test spin?

[johu]

Set snshs to i3 in 5.24 software, that will give you correct readings for temphs and udc

Edit: you did... That's odd
Still not home, will check back later

[janosch]

Yes, get good readings on 5.24, but not on 4.97.

We are populating the AmpSeal and trying normal `run` mode for today to get some noise or a rotation out of it.

If we need to do syncofs with this motor we would need `manualRun` in the future - I think I volunteer to fix that bug to make `manualRun` work in later revisions so firmware doesn't need to be flashed back and forth, but thats for another day.

EDIT: Ok, udc now reads 0V in 5.24 even though I can measure 24V at the inverter pins, wonder if I damaged the sensor, but can't see how or when.
It comes in through GPIOC3, just checking if I can measure something there on the mini mainboard.

[Pete9008]

Might be worth having a look here viewtopic.php?t=3097. Not tested on real hardware yet, so only use if you are comfortable with that, but it does reintroduce manual mode (but in a slightly different way!).

[janosch]

Measuring 0.92V on UDC pin on mini mainboard, no matter if 24V on DC bus or not, that can't be right.

Trying to trace this now where it's coming from.Is it coming through SPI for the i3? looking at the code.

Edit: Yes, it is coming through SPI of course.

But snshs is set to i3, why does it still not read the voltage?
There is some noise on idc and il1, il2, but not on udc.

Has my flashing backwards and forwards overwritten something?
The behaviour reminds me of the overwritten can map for the Tesla charger.

Params attached.

[CJW]

After a hiatus over the summer I have resumed work on the conversion.

-traced the resolver and motor temperature connections
-ordered some PCB's to simplify their connection to the openinverter drop-in board
-info posted on page 9 of the main i3 hacking thread: viewtopic.php?p=63507#p63507

-i3 motor and transmission weighed 69.6kg, I believe without the inverter or engine mounts. Will confirm and update as well.
-weighed various ICE components upon removal to keep track of net vehicle weight and help with battery design, posted below. Weights of anything over a few kg measured with digital household scales so should be considered approximate. Will confirm and update this post.

image.png (11.29 KiB) Viewed 10106 times

[CJW]

Current thinking for the battery:

-build 96s1p pack from LG Chem E78 pouch cells or similar
-27.5kWh total so around 25 kWh usable
-104kg for the cells, plus additional weight of cooling and packaging
-located above rear subframe where fuel tank was positioned, pending outcome of F/R weight distribution calcs. The height isn't ideal but the i3 motor's mass will be lower than the ICE engine, and a full fuel tank (ie tank and fuel) weighed around 36kg, so the net increase to CoG will hopefully not be too great.
-upgrade capacity in the future as more energy dense cells become available

Specs from LG Chem below.

[Scott166]

Good work! I'll be following with interest to see how you get he motor spinning

[CJW]

Finally got the motor spinning successfully with the drop-in board so next step is to figure out how to fit it in the sub-frame and get a set of mounts made. It's quite a tight fit, hoping to be able to get everything in without having to modify the cross-frame to create more space. Aware that the motor is going to need some space to move in the mounts, and also that some of the suspension components may flex under load.

View from above:

The back of the gearbox is close to the rear cross-frame bolts at the moment. Need to finalise the height of the motor relative to the main subframe box section as this affects the clearance. Should also be able to move the motor further forward once the air conditioning compressor mounts are gone:

Location of the air conditioning compressor mounts - the one on the lower right is actually touching the de Dion tube, preventing the motor from sitting closer to the tube, and further forward:

Close up of the air conditioning compressor mounts:

[CJW]

Air conditioning compressor mounts removed...

[CJW]

With the motor back in the subframe and the aircon mounts out the way, it now appears there's (just) sufficient space for it so will be getting on with designing/fabricating the mounting frames.

[CJW]

Reviewed motor position in the subframe again today. Current thinking for the motor mounts is:

-Run a box section across the two rear mounts
-Run another box section from the rear box section to the forward engine mount, with a laser cut plate picking up the various mounting holes on the side of the motor via bolts through spacer tubes
-Single tube connected to a piece of angle should be able to pick up the three mounting holes on the left hand side

Very rough sketch below.

[CJW]

Got the rear subframe and dedion tube back from the paint shop a couple of weeks ago, now reassembled in typical driving position to start work on the motor mounts.

The clearance front and rear is quite tight, so also had to remove a tab/lifting eye from the back of the transmission housing to clear the rear cross-frame for the dedion tube. Even with this, there's only about 10mm of clearance front and rear so will be stiffening the engine mounts with inserts to try and keep movement within that range. A Powerflex insert is available for the front but the rears would need to be 3D printed or cast in place.

Wooden bar is a placeholder for a steel box section that will run from the rear driver side to offside engine mounts, to pick up a smaller longitudinal box section that will run to the forward engine mount and support the driver side of the motor. Have done some calcs based on the drive unit weight and max torque the tyres could transfer without slipping to size the beam, looking like 40x40mm with 2mm wall thickness.

Will share the torque/speed calcs at some point as they may be of interest to others planning similar conversions. Based on the i3s motor's torque curve, 9.665 gear ratio, and the roadster's weight distribution and wheel diameter it looks like the wheels would be capable of breaking traction with dry asphalt up to about 64mph, with a top speed of 82mph limited by the gear ratio and motor rpm. Planning to keep the traction control as an option, and looks like this would be particularly important in the wet.

[CJW]

3D printed mockup of the offside engine mount to check geometry and bolt spacing. Planning to get these laser cut from 3mm steel then welded together with steel standoff tubes. Driver side mount is more complicated.

[CJW]

Janosch was down this week to check in on progress so we started setting up hardware to get started on sorting out CAN bus communications.

Have ordered a spare MEG (ECU), mainly to salvage the connector for a replacement ECU build to link the roadster's SAM unit to the openinverter and throttle, but will also use it for some CAN interrogation and testing.

There's quite a lot of useful information out there:
-https://www.evilution.co.uk/mod/roadste ... s-pids.htm (some CAN Parameter IDs and byte values, SAM and MEG connections)
-https://www.fq101.co.uk/how-to-guides/r ... electrical (SAM unit connections)
-Forum member Pete9008 who has done some investigation on the ESP
-Forum member seb43654 who has posted dbc files for the roadster
-https://www.researchgate.net/figure/Sma ... _221032785 (paper summarises the CAN architecture)
-https://hackaday.io/project/19931-smart ... ngineering (has done some work on attempting to work out CAN PID's and has downloaded CAN data available from a Smart fortwo)

Am compiling everything in a single spreadsheet and hope to produce a more or less complete guide to the roadster's PIDs and CAN structure to share.

[CJW]

Sounds good!

Bear in mind that the motor mounts see driveshaft torque reaction not motor/engine torque. The higher ratio of the Roadster box 1st gear balances things up a bit and the mounts will have been designed to handle that. The i3 is likely still higher but probably not by enough to worry the uprated bushes.

Did the calcs on this and looks like the peak driveshaft torque from the petrol motor would have been 1455Nm, while the peak driveshaft torque from the i3s will be 2610Nm (ie total for both shafts). In practice my calcs suggest the wheels would likely lose traction at around 1300-1400Nm of driveshaft torque on dry asphalt (possibly a bit higher with rear squat), which should limit the force on the mounts/bushes.

Table of original max petrol motor driveshaft torques below.

image.png (6.05 KiB) Viewed 5410 times

[CJW]

Updated estimate of max tractive force on dry asphalt is 4300N (1270Nm of driveshaft torque) based on the front/rear weight distribution of 44%/56% (max axle loads of 465/610kg) and assuming weight of 878kg from the car's operating instructions (includes allowance for engine fluids and 75kg driver).

Have combined this with the torque and power curves for the roadster engine in gears 1 thru 6, and a drag curve based on aero drag and rolling resistance. This indicates a top speed of just over 110mph, which is around the figure quoted for the car:

The max tractive force increases just over 5000N (1490Nm of driveshaft torque) for the max permissible gross vehicle weight of 1030kg, a useful figure for designing the engine mounts:

The BMW i3s drive unit with its 9.665 gear ratio produces far more torque than the tyres would be able to put down on the road:

A lower gear ratio, for example 6, would be better matched, but the additional top speed pretty useless unless on a track with aero mods.

[CJW]

The plan for the driver's side motor mounting frame is to run a transverse tube between the two rear engine mounts and then a longitudinal bar to the front engine mount, picking up the three bolt holes in the motor housing. Have started cutting the steel parts, will post pics as it develops.

The roadster subframe has no built-in support on the offside front corner, so need to address this. One option seems to be to use the offside front subframe mounting bolt to hold an additional motor mount in place while avoiding welding the subframe.

[KobusI3toKombi]

"Planning to keep the traction control as an option, and looks like this would be particularly important in the wet." Hi Chris- what was your plan to keep the traction control? Also working towards using and I3 Motor and gearbox, but seems like quite a task to keep the traction control.

[CJW]

Using CAN signals from the SAM unit, which already reduce the torque requested from the motor based on the wheel speed sensors if the ESP is enabled. Still working on the engine mounts, once those are done will get back to the CAN testing. Pete9008 has some good insights here: viewtopic.php?p=43659&hilit=esp#p43659