2004 BMW 330 with SMG/SSG Gearbox

[mdrobnak]

This will be a full EV conversion of a 2004 BMW 330 sedan. It has a fairly rare gearbox in it, so I plan on keeping that.

Previous posts were from the now M3 hybrid thread. There's not too much going on there either as I really need to get this done first. I've made good process on the M3 SMG gearbox control recently, but that's about it.

As for the 330, here's the current plan:

Drivetrain:
* Nissan LEAF EM57/EM61 motor. ~$500. [ 2 inputs @ 400A should produce significant amount of torque out of this ]
* UUC Dual-Disc clutch + lightened flywheel (what's on the car currently)
* SMG/SSG 6-speed BMW gearbox
* Stock differential
* Adapter assembly $250 ???
* Leaf spline adapter $300 including shipping

Inverter:
* Chevy Volt Inverter (part # 12643810) - $81!
* Chevy Volt HV Cables (part # 24274910 and 24274911) - $235
* Ampera control board from Damien - 100 Euros (~$112) value.

VCU (Vehicle Control Unit):
MoTeC M130 with custom software. Will control SSG gearbox, and contain torque -> demand calculations to keep everything running happily.
Purchasing from business partner from the first car I did development for MoTeC M1 on.
* M130 ECU
* Logging Upgrade (Up to 2000 items, and up to 1000 Hz rate (not both!))
* Analysis upgrade (Motec has standard and "Pro" edition of their i2 software, I accidentally purchased this upgrade, oops, had to eat the cost..)
* Development package (run your own software on it)
Value: $6487
Cost: $750
Sane configuration: M130, Drobnak Software "DS BMW E46 OpenInverter EV" package, Level 2 logging (200 channels at 200 Hz) - $2295 + 650 + 329 = $3274

Batteries:
* 108 CALB L173F163B from https://www.electriccarpartscompany.com ... -l173f163b - 108*165.50 = $17,874 (Ouch) - 1300A max for 10s!
* Orion BMS 2 for 108 Cells ~$1800
* Battery Box / Assembly - ???

HV Junction box / in battery box:
* 2 GigaVac GV200 Contactors - $140
* 1 GigaVac MiniVac Contactors - $40
* 1 400A 500V A50QS400-4 Fuse - $30
* ISA Shunt - $181 + $15 wire transfer fee = $196
* BUD Clear-top junction box $66

Charging:
* Tesla Model 3 Charge Port ECU $45
* Tesla Model 3 Charge Port + HV cable $210
* Tesla Model 3 Charge Port Door $54
* Tesla Model 3 PCS $265


I will also attempt to break down costs here.

Running value: $81 + 235 + 112 + 3274 + 3110 + 30 + 460 + 140 + 40 + 30 + 196 + 475 + 255 + 45 + 210 + 54 + 265 - 1350 + 300 = $7962
Running cost: $81 + 235 + 0 + 750 + 3110 + 30 + 460 + 140 + 40 + 30 + 196 + 475 + 255 + 45 + 210 + 54 + 265 - 1350 + 300 = $5326

When I have a component in hand, I'll update the cost.

Edit 2022-05-31 - Added value recovered by selling 6 LG batteries. Add new plan for batteries. Moved all outdated info to outdated section below this changelog.
Edit 2020-06-08 - Struck through some text that is no longer applicable.
Edit 2020-06-09 - Removed Leaf motor references, not going with that. I think this is going to be a fun setup when done.
Edit 2020-06-14 - Organized into categories as the list has gotten larger. Added value vs out-of-pocket cost.
Edit 2020-06-19 - Batteries arrived! Updated info on the Remy motor...if someone is interested I'll pass on your info. Costs updated.
Edit 2020-08-18 - Add Tesla Model 3 parts, add temporary Nissan Leaf charging port price into the equation.
Edit 2021-03-23 - Add Tesla Model 3 PCS, Leaf motor, Leaf spline coupler, remove mystery motor reference for now.

Outdated info:
Drivetrain:
* Remy HVH250-115-DOM motor (Motor is repairable, but found a better option...) - was going to be $1800!
* Mystery motor [ ENGIRO GmbH 260W-08011-ABC ] . Gonna be good.

Batteries:
* 6 LG Chem / Chryler Pacifica batteries ($450 * 6) + $100 "crate" fee + $310 Hazmat residential freight delivery = $3110
* Chevy Volt BMS slaves - $30
* Simp BMS - 459.88 (Simp BMS - Volt Gen 1, CANBus Expansion, Connector Set, Shipping - 350 GBP, $5 USD PayPal fee)

Charging:
* Chevy Volt / Lear charger w/ DC-DC - $475 from Jon Volk [ Actually from CODA car ] - Going to use this on the M3 or the bench instead...
* Sell the EVCC charge controller that was included for $???
* J1772 AC charge socket + CHAdeMO socket from LEAF - ~$250 - Round one - $255 inc shipping
-- Need to see how badly the CHAdeMO one is messed up in real life. Bad. Only good for some reverse engineering work.
* CHAdeMO Control board, contactors - $? (Phase 2)

Other specs:
* Torque peak (Is this linear to current?) ends around 4500 RPMs at 85% value (600A?)
* Pack size of ~16 kWh / 50 miles range at 100% usage

Rating is 480Nm at ... 700A. I think the Volt inverter with Arber333's board will do 600A. It's unclear if combined IGBT current was 1000A, or they are 1000A each! If the latter...That's gonna be fun.

[arber333]

Rating is 480Nm at ... 700A. I think the Volt inverter with Arber333's board will do 600A. It's unclear if _combined_ IGBT current was 1000A, or they are 1000A each! If the latter...That's gonna be fun.

Yes, when i was doing my power estimate of Volt inverter i tested it on 200V battery that could go to 400A so i think 600A is a conservative estimate. I got MBB600TV6A IGBT datasheet from mr Kelly and it is rated as 600A part with 1200A surge for 1ms. I do not intend to go over 600A per unit so my goal of 1000A would be reached with both power sections used with single brain and driving the same motor.
See here its datasheet: https://www.diyelectriccar.com/forums/s ... stcount=78

[mdrobnak]

Rating is 480Nm at ... 700A. I think the Volt inverter with Arber333's board will do 600A. It's unclear if _combined_ IGBT current was 1000A, or they are 1000A each! If the latter...That's gonna be fun.

Yes, when i was doing my power estimate of Volt inverter i tested it on 200V battery that could go to 400A so i think 600A is a conservative estimate. I got MBB600TV6A IGBT datasheet from mr Kelly and it is rated as 600A part with 1200A surge for 1ms. I do not intend to go over 600A per unit so my goal of 1000A would be reached with both power sections used with single brain and driving the same motor.
See here its datasheet: https://www.diyelectriccar.com/forums/s ... stcount=78

Nice. Now if I could just calculate expected torque value...
Looks like Hitachi designed a good amount of the inverter. I saw on the Weber Auto teardown that the original control board had Hitachi on it. Didn't realize they made the IGBTs as well.

I looked at your PCB file on Github. File name says dual interface (and it looks like so - 2 resolver / 2 driver connectors), but the Silkscreen still says single. It also looks like there's some unrouted connections?

Thanks for designing the board.

-Matt

[arber333]

I looked at your PCB file on Github. File name says dual interface (and it looks like so - 2 resolver / 2 driver connectors), but the Silkscreen still says single. It also looks like there's some unrouted connections?

Thanks for designing the board.

-Matt

Well you want to inspect this one. This is latest and tested board.
https://github.com/arber333/Johannes-co ... IGBT-board

This can actually run single motor. This is why i called it single. But it can drive DUAL IGBT sections.
Mind you, you can also run whichever section by itself...

On the right side there is Johannes Rev2 design which controls precharge and DC switch. If you look carefully on the bottom near connectors you will see i left some traces cut with holes for jumpers. If you would use both sides you will need to connect those traces tith thin wires.
Also you will need to replace 3K3 resistors on current sensor traces with 6K8 so you will get close to 3K3 in parallel. This should average out both sensors. Then there are temp sensors and Fault pins where i used OR link.
I even anticipated power loss from 3V3 because of wifi module, but it is not the case. Still there is a 3V3 regulator available to power wifi, or you can just steal 3V3 from chip board.

Resolver circuit is made so you can also use encoder if you want. You just omit resolver circuit and populate pullups...

Orientation of Olimex chip is like the text between connectors . Just align the text on the chip with it.

On the left you have Lebowski controler brain. It is supplied by 5V but it should be held in reset by one pin of the precharge relay. I left myself some leeway by not connecting directly to relay. Also relay itself can connect GND on both actuating pins.
This would be an option to use third power section which is hidden under the main cap and can drive AC compressor.

And you will want to rebuild both power layers, i just cleared them so you are able to see details on both sides... on second thought i think i will just correct the file and post it back.

[mdrobnak]

I looked at your PCB file on Github. File name says dual interface (and it looks like so - 2 resolver / 2 driver connectors), but the Silkscreen still says single. It also looks like there's some unrouted connections?

Thanks for designing the board.

-Matt

Well you want to inspect this one. This is latest and tested board.
https://github.com/arber333/Johannes-co ... IGBT-board

This can actually run single motor. This is why i called it single. But it can drive DUAL IGBT sections.
Mind you, you can also run whichever section by itself...

On the right side there is Johannes Rev2 design which controls precharge and DC switch. If you look carefully on the bottom near connectors you will see i left some traces cut with holes for jumpers. If you would use both sides you will need to connect those traces tith thin wires.
Also you will need to replace 3K3 resistors on current sensor traces with 6K8 so you will get close to 3K3 in parallel. This should average out both sensors. Then there are temp sensors and Fault pins where i used OR link.
I even anticipated power loss from 3V3 because of wifi module, but it is not the case. Still there is a 3V3 regulator available to power wifi, or you can just steal 3V3 from chip board.

Resolver circuit is made so you can also use encoder if you want. You just omit resolver circuit and populate pullups...

Orientation of Olimex chip is like the text between connectors . Just align the text on the chip with it.

On the left you have Lebowski controler brain. It is supplied by 5V but it should be held in reset by one pin of the precharge relay. I left myself some leeway by not connecting directly to relay. Also relay itself can connect GND on both actuating pins.
This would be an option to use third power section which is hidden under the main cap and can drive AC compressor.

And you will want to rebuild both power layers, i just cleared them so you are able to see details on both sides... on second thought i think i will just correct the file and post it back.

Ah, this makes much more sense now.
So to sum up:
Board is set up to drive one motor with either a single output, or if jumpered, both outputs from the same logic.
Parallel the phase wires if driving from both IGBT sections.
Populate current sensor resistors accordingly.
Lebowski brain can be used to control an AC compressor, but timing coordination of the HV bus and the startup of that needs to be given care, if I'm understanding that correctly.

I'm not sure of the difference between resolver / encoder. I assume most are using resolver? (ie Remy, Siemens, Leaf)

I'll wait for you to update the power layers and I'll look at it again.

[arber333]

Bas just made a new code for Lebowski controler which uses 45deg start pulse so motor can take off easier in sensorless. I tried it and it works really nice. Now you may have two ways of controling AC drive. With another chip that would simulate throttle signal and reset function or a CAN bus throttle from another chip. I am still working on CAN bus control here... Dont know yet how to use the CAN throttle...

Idea is to use very lean acceleration limiter and ramp up throttle for the AC drive. When throttle stabilizes drive should be kept there until AC clutch signal wants to disconnect ac compressor. Of course this would not happen. Instead it would get brain to reset and keep it there untill AC clutch signal gets back online. Basicaly ON/OFF AC.

[mdrobnak]

For reference here is the vehicle in its natural habitat, bundled with laptop.

And two complete Volt BMSes:

-Matt

[mdrobnak]

After talking to the current owner of the motor, likely 600A is probably going to be the max for heat dissipation for the motor. Approx torque at 600A is around 427Nm extrapolating from a graph Borg Warner has. Constant Tq -> Power ~4000 RPMs. The upside of this is that it means I only need to use one half of the inverter. Which means simpler wiring.

Attempted to order a SimpBMS, unfortunately back-ordered currently. Inverter will arrive in 2 days.

[mdrobnak]

Just ordered this:

$215 + $20 shipping = $235. Usually they're like $260 a piece. It makes the inverter less of a deal, but still decent!
Added price to first post.

[arber333]

Just ordered this:
s-l1600.jpg

$215 + $20 shipping = $235. Usually they're like $260 a piece. It makes the inverter less of a deal, but still decent!
Added price to first post.

Yeah, but any DMV inspector seeing this will just look for other things. This is totaly foolproof as phase wiring is considered. I remember now i have seen measly 16mm2 also on Remy motor. Your contacts may be beefier, but still you only need some 30kW to drive on level road. Anything else is for acceleration and it is over fast .

Also if you intend to use that BMW transmission you may build Alu casing on the bell to put motor inside with lubricating and bearing seals and everything. This could be a nice compact unit.

[mdrobnak]

Yeah, but any DMV inspector seeing this will just look for other things. This is totaly foolproof as phase wiring is considered. I remember now i have seen measly 16mm2 also on Remy motor. Your contacts may be beefier, but still you only need some 30kW to drive on level road. Anything else is for acceleration and it is over fast .

Also if you intend to use that BMW transmission you may build Alu casing on the bell to put motor inside with lubricating and bearing seals and everything. This could be a nice compact unit.

Yep, it's why I'll pay a little more to get it to OEM as possible. More than OEM performance is always welcome though.

Yes, I'm using the 6-speed that comes with the car. I'm not sure of how it will end up aside from I know I'm going to use the EVWest coupler:
https://www.evwest.com/catalog/product_ ... cts_id=326

-Matt

PS:

And you will want to rebuild both power layers, i just cleared them so you are able to see details on both sides... on second thought i think i will just correct the file and post it back.

Did you get a chance to look at this?

[mdrobnak]

Inverter arrived today. Metal was more pitted than I expected, but.. $81..sooo..

Next step is to get it opened up.

-Matt

[mdrobnak]

Cables arrived, but haven't gotten to it yet. Trying to get the torque commanded to equal torque delivered on the 330.

Also paid for the SimpBMS today. Will add to running total cost when I get it, and will post pictures.

[mdrobnak]

Pre-req progress:
Commanded torque is torque delivered, for the most part. I had a hack that I forgot about that was throwing off my calculations. Now as long as airflow calculated = airflow delivered, then torque is the same as well.

EV progress:
Received the HV cables a few days ago:

Packaged well:

Two cables, with thank you card:

And...they're a little rougher than in the pictures (I'ms seeing a trend here.):

I used some Simple Green on them, and that cleaned it up quite a bit. I haven't attempted to take them apart yet:

There are some tears in the orange sheathing:

Looks like it was just in the shielding:

The connector themselves are indeed weird as others have noted:

Questions at this point:
1. Best way to ensure the second set of cables is ok and not cut into? The other set looks to be fine.
2. It looks like the pins sit against some sort of ball with a hole in the middle? Are they supposed to be separated like that, or do I have to completely disassemble these connectors? (I'm probably going to anyway...)

I hope to get the inverter cover off tonight or tomorrow.

-Matt

[arber333]

Questions at this point:
1. Best way to ensure the second set of cables is ok and not cut into? The other set looks to be fine.
2. It looks like the pins sit against some sort of ball with a hole in the middle? Are they supposed to be separated like that, or do I have to completely disassemble these connectors? (I'm probably going to anyway...)

-Matt

1. Go and measure continuity from phase contactsd to its shield. There should not be a beep or any ohm measurement.

2. Why would you want to dissassemble them? Pins sit against a spring because they need a good contact in high vibration enviroment. Still i dont trust just friction contacts. I wont them bolted down. But those 16mm2 measly cables are good if you want to go assembly like OEM. I will just go for larger contacts and cables....

You can just use the casing part that bolts to inverter and cut cables downstream and splice them to your choice motor cables.

[mdrobnak]

Questions at this point:
1. Best way to ensure the second set of cables is ok and not cut into? The other set looks to be fine.
2. It looks like the pins sit against some sort of ball with a hole in the middle? Are they supposed to be separated like that, or do I have to completely disassemble these connectors? (I'm probably going to anyway...)

-Matt

1. Go and measure continuity from phase contactsd to its shield. There should not be a beep or any ohm measurement.

2. Why would you want to dissassemble them? Pins sit against a spring because they need a good contact in high vibration enviroment. Still i dont trust just friction contacts. I wont them bolted down. But those 16mm2 measly cables are good if you want to go assembly like OEM. I will just go for larger contacts and cables....

You can just use the casing part that bolts to inverter and cut cables downstream and splice them to your choice motor cables.

1. Duh. Thanks for that. Sometimes I think too hard.

2. Ah, didn't think of it like that. Ok, good to know it's not an issue. I was going to disassemble to ensure they were as clean as possible. For now, stock diameter is fine I think.

-Matt

[mdrobnak]

Stopped by EVDrive today, who had some leftover goodies which I picked up:
2 main contactors
1 precharge contactor
1 fuse

All for $210 for pretty much brand new stuff. I also got to talk to Bob Simpson, the owner / founder of the company, and Paul Lackey who has been helpful with some general guidance.
I'll post actual parts / pics tomorrow, but wanted to share before I went to bed. Running parts list and prices updated on first post.

-Matt

[mdrobnak]

It's free!

Finally got the Volt inverter opened. Wow, what a pain in the butt.

Needed to use my Wagner heat gun at about 600 degrees F, the scraper that comes with it, a flathead screw driver, and a mallet. I tried using a knife to cut the seal initially but that failed miserably, don't bother.

You can see the two main contactors and the minitactor in the top right of the picture, too. Not sure where I put the fuse.

On the upside, looks definitely intact. So that's a score for $81.

-Matt

[Kevin Sharpe]

You can see the two main contactors and the minitactor in the top right of the picture, too.

If they're Tesla parts then don't forget to add an economizer.

Here's the Model 3/Y versions but the issue is the same for Model S/X;

https://openinverter.org/wiki/Tesla_Model_3_Contactors

[mdrobnak]

You can see the two main contactors and the minitactor in the top right of the picture, too.

If they're Tesla parts then don't forget to add an economizer.

Here's the Model 3/Y versions but the issue is the same for Model S/X;

https://openinverter.org/wiki/Tesla_Model_3_Contactors

Not from a Tesla, as far as I know.

GV200MA-1 for negative and main contactor. That has an economizer built in.

As far as the MiniTACTOR, doesn't look like it. But on the flip side, I should be able to kill that current path once the cap is charged, no? If that's the case, do I need external circuitry on that?

-Matt

[Kevin Sharpe]

Not from a Tesla, as far as I know.

GV200MA-1 for negative and main contactor. That has an economizer built in.

The Tesla GIGAVAC part is a custom GV200 and the ones shown in your photos are the stock part with "PWM" coils which I think are ok. Maybe test on the bench at 12V and check the temperature rise?

As far as the MiniTACTOR, doesn't look like it. But on the flip side, I should be able to kill that current path once the cap is charged, no? If that's the case, do I need external circuitry on that?

Not sure about the MiniTACTOR specs (Maybe test on the bench at 12V and check the temperature rise? ) but worth remembering that Damien's designs typically wire the precharge contactor in parallel with the Negative HV contactor so are on whenever the battery is powered up. Check your inverter controller for details of the precharge/main contactor control scheme.

[mdrobnak]

Not sure about the MiniTACTOR specs (Maybe test on the bench at 12V and check the temperature rise? ) but worth remembering that Damien's designs typically wire the precharge contactor in parallel with the Negative HV contactor so are on whenever the battery is powered up. Check your inverter controller for details of the precharge/main contactor control scheme.

Oof, I may have goofed on buying that then. I'm holding off on retrofitting a board to the Volt inverter until Damien is finished up with his design. Although for that inverter its basically a collaboration between Damien and Arber, so I'd have to look more closely at the schematics again.

In other developments, I did the continuity test, nothing between the HV and the shield, so that wasn't compromised. Yay! I did try and take apart that cable a bit to see if I had shrink which would fit over it - I don't. I have to get larger heat shrink to fix up the damaged areas. Another thing I noticed is that there is U V W markings on the connectors, which is nice, as well as them having unique notches in them...GM (Hitachi?) made it pretty foolproof.

I'll probably order my batteries next.

Still working on control code. I swapped out the idle control algorithm and need to test that further. Have not had the time to do that.

I also haven't had the time to get the EV Components thread put into the wiki.

But at least I am making slow forward progress.

[mdrobnak]

Based on a reply from Damien, the current code turns off the contactor after precharge. So I'm OK there.

I finally unpacked the charger today. Here's what we have:

EVCC controller, charger with the 3 phase AC input receptacle needing some work, a weird 1/8" jack to USB cable, and AC / DC pigtails.

Also found out that it's a CODA charger:

I love that the part number didn't even exist.

-Matt

[tom91]

Luckily the SimpBMS should directly work with a CODA charger, canbus is slightly different then the LEAR variants.

Good thing these chargers open up easy and make changing the plugs on them quite straight forward so you can fix the AC side to a pigtail or another panel mount connector.

[mdrobnak]

Luckily the SimpBMS should directly work with a CODA charger, canbus is slightly different then the LEAR variants.

Good thing these chargers open up easy and make changing the plugs on them quite straight forward so you can fix the AC side to a pigtail or another panel mount connector.

Man, CODA seemed like they wanted to change stuff just to obfuscate things. Glad to hear the integration will be straightforward at least.

Jon Volk ran the charger like that for some time, so the connector is OK enough to work if checked often, but yes I definitely want to replace it.