2004 BMW 330 with SMG/SSG Gearbox

[mdrobnak]

Hmm.

Well, I didn't get to do what I wanted, but I did learn something.

Logging on the "party" bus with the EVTV thing doesn't upset the car because it's the backup.

Messing with the charge port controller..Well...yeah. Uh. Don't.

1. It's a tight fit.

2. I understand this part now:

INSTALL

Installation procedure is the reverse of removal, except for the following:

Update the firmware.

I plugged in the power, and after a second or two, the battery contactors OPENED. I thought that was odd. So I opened the driver door to update the charge limit from 70 to 80% (I wanted some headroom to test with this ECU), and that's when I saw it was really, really pissed off. It said car was not driveable and to contact Tesla service. Yeah, oops. Hmm. Let's open the charge port door. Oh, that works. Look at the door. Tesla logo in RED. RED = Error. Hmm. Plug in the cable, still red. Ok then.

Swapped the connector back to the one that came with the car, contactor closes, coolant pump starts. Plug in cable, and it is charging currently.

-Matt

[Isaac96]

I've been reading up on Tesla things and swapping parts. I think the charge port ECU is one of the things that has hundreds of firmware revisions (or at least number changes) and you need to redeploy firmware to the whole car if you swap it. Maybe after redeploying you could swap them back and forth (they should have the same firmware then).

Model 3... No idea how to handle redeploying. On the S you can telnet into the CID (one of the computers) using a special Ethernet cable and run commands.
https://teslamotorsclub.com/tmc/threads ... rt.142122/

INSTALL

Installation procedure is the reverse of removal, except for the following:

Update the firmware.

Right - updating firmware does the same thing as a redeploy. But if you update with the extra ECU in the car, the original won't work (unless you redeploy).

-Isaac

[mdrobnak]

Yeah I don't think that option is available to Model 3 owners. AFAIK the newer MCUs are much less hackable.

On the upside, I did get a list of the ARBIDs with the ECU disconnected, so I know what is available to it.

-Matt

[mdrobnak]

So, this has been the case for the last 30+ minutes:

no_bus_errors.png (780 Bytes) Viewed 3348 times

So a simple replay keeps it blinking merrily away, with no bus resets. So that's a data point.

[mdrobnak]

There is an alive message that comes from the battery pack (which is what the controller is connected to, if I read the schematic right.

More data points - the car is more upset by a module with 1/2 of a CAN connection than it missing the module. One of my CAN wires came apart (I had to shove thse individual pins (like on arduino / breadboard type projects) onto the controller that was mounted in the car. Not an easy feat I'll tell you that.

But, I did get a log of a charge off AC (Tesla SWCAN style).

I did find the connector: https://www.te.com/usa-en/product-1452349-1.html - But unobtanium it seems. Not sure which harness it would be from Telsa, so that's gonna be fun to figure out.

-Matt

[mdrobnak]

I asked about paralleling the LG (Apparently LG A7) modules at EV West, and I didn't get a very encouraging response back.

On top of that, I stumbled upon a thread on DIY Electric Car where someone had two wires short..and lets say their entire pack went up in flames.

Therefore, single string, and range is what it is. Let's see how we do to being with before messing with it.

One more piece of information from that email - apparently LG sent out cease and desist letters to the sellers. Not sure based upon what, but..that explains why it all dried up nearly instantly.

-Matt

[Isaac96]

I asked about paralleling the LG (Apparently LG A7) modules at EV West, and I didn't get a very encouraging response back.

On top of that, I stumbled upon a thread on DIY Electric Car where someone had two wires short..and lets say their entire pack went up in flames.

Therefore, single string, and range is what it is. Let's see how we do to being with before messing with it.

One more piece of information from that email - apparently LG sent out cease and desist letters to the sellers. Not sure based upon what, but..that explains why it all dried up nearly instantly.

-Matt

Oh dear... I was considering getting another string if I moved to a Tesla DU...

Got a link to that thread? I keep hearing horrific legends about the deaths of batteries, and haven't seen any direct data myself (besides the lithium cells I've stabbed for fun).

Range - I'm getting up to 90 with careful driving (having to keep my motor cool, it's full of air still) with a double string.
Also my cells are staying at exactly the same voltage difference (13mv, the modules were at slightly different voltages when I installed them). Which seems promising for longevity.

-Isaac

[arber333]

Oh dear... I was considering getting another string if I moved to a Tesla DU...

Got a link to that thread? I keep hearing horrific legends about the deaths of batteries, and haven't seen any direct data myself (besides the lithium cells I've stabbed for fun).

Range - I'm getting up to 90 with careful driving (having to keep my motor cool, it's full of air still) with a double string.
Also my cells are staying at exactly the same voltage difference (13mv, the modules were at slightly different voltages when I installed them). Which seems promising for longevity.

-Isaac

I had the pack of 28kW LiFe cells die on me one sunday morning years back! This was because i paralelled them with LiPo cells. I calculated their compatibility at 3.55V LiFe vs 4.0V LiPo. I also put a contactor and BMS charger control in, but that didnt save them. I suspect BMS contactor was jumping in and out and causing curent surges from lower Ri to Higher Ri cells.
The LiFe released magic smoke and died gloriously in a blanket of fire smoke (fire is too dramatic word, it was quite high temperature though, enough to melt snow around them!). LiPos didnt care... they remained at 4.0V per cell the whole time.

Well now i am experimenting with one VW Passat 96S pack in paralel with my Ampera pack and it works. Same chemistry... Though sometimes car BMS just throws an error out of spite. I will install a DC contactor with a diode in reverse so rear pack can be charged from original charger.
When you arrive at destination and turn off the car batteries should not be connected still! There is a lot of current fluctuations after a drive because of minute differences in cell Ri. Current flows back and forth and makes BMS crazy.

If you need to paralell cells it is best to do it on the cell level. Otherwise you need a contactor and BMS enable control. And NEVER try to charge both pack to the cell voltage limit. I set my limit at 4.05V for voltage per cell and it works good enough.

[mdrobnak]

Oh dear... I was considering getting another string if I moved to a Tesla DU...

Got a link to that thread? I keep hearing horrific legends about the deaths of batteries, and haven't seen any direct data myself (besides the lithium cells I've stabbed for fun).

Range - I'm getting up to 90 with careful driving (having to keep my motor cool, it's full of air still) with a double string.
Also my cells are staying at exactly the same voltage difference (13mv, the modules were at slightly different voltages when I installed them). Which seems promising for longevity.

-Isaac

Yeah that seems in line with my estimate of 45-47 miles.
Then arber's post...yikes! Yeah I did know not to mix chemistry that's for sure.
Here's the thread: https://www.diyelectriccar.com/threads/ ... st-1048165
Which was the result of: https://www.diyelectriccar.com/threads/ ... 161/page-2

So definitely some poor design choices there, but definitely a good warning reminder...

-Matt

[Isaac96]

Yeah that seems in line with my estimate of 45-47 miles.
Then arber's post...yikes! Yeah I did know not to mix chemistry that's for sure.
Here's the thread: https://www.diyelectriccar.com/threads/ ... st-1048165
Which was the result of: https://www.diyelectriccar.com/threads/ ... 161/page-2

So definitely some poor design choices there, but definitely a good warning reminder...

-Matt

I would blame that fire on the poor decision of using only electrical tape to isolate bus bars... Didn't seem to be a problem of paralleling.
That whole thread is pretty insane
"Plus electric cars are a lot safer than Gasoline cars. I have already rolled my jeep twice and am lucky that thing hasn't caught on fire yet from oil leaking into the cylinders. I doubt an electric car would catch on fire upside down with a proper roll cage protecting my batteries. yeah I messed up. By putting my positive wire so close to the negative busbar. But I highly doubt I will ever make that mistake again."

Wow.

-Isaac

[mdrobnak]

Yeah. Exactly.

In other news:

I am starting to attempt things with my port and the Teensy. Just replaying traffic makes it do some interesting / weird stuff, do I'll try and get it doing more reasonable things soon enough.

-Matt

[arber333]

Range - I'm getting up to 90 with careful driving (having to keep my motor cool, it's full of air still) with a double string.
Also my cells are staying at exactly the same voltage difference (13mv, the modules were at slightly different voltages when I installed them). Which seems promising for longevity.

-Isaac

Well i recommend you use additional hose length and a bucket and force the coolant into the motor with an additional pump on top. Car needs to be on incline hood down so that motor is the lowest part of system. Then you set the pump to pull the coolant from the bucket and on the other end coolant returns to the bucket. While system is working in that position you grab the return hose and you limit/squeeze it to increase dP in the system. This will cause some cavitation on the bends, but faster liquid (dv) will take air bubbles further along the lines.
When you are certain you dont have any bubbles there you take apart the additional pump and hoses.

I found out 12V Bosch Audi pump core needs to be immersed in coolant to work properly. So i pour coolant into the pump first and then run it.

[mdrobnak]

Interesting / weird stuff == "Everything is in trouble, SHUT DOWN".

The replay had a lack of the keepalive counter, so the "bus reset" behavior I was seeing was preceded by a shutdown request in this case.

At this point it is going into a AC_CHARGER_FAULT state. A a few possibilities:
* Missing more messages
* Latch enable needs to be in the correct state (unclear if high or low)
* Motor needs to be hooked up

The fault pin is ~4V. Goes into that state immediately on boot-up, so not related to the latch not enabling. That's not to say that the motor not being there isn't causing this...
Proximity line does the expected thing, 12V idle, 0V when plugged in.
Just unclear what to do with the latch enable.

On the note of power supplies... Probably need a 5V and a 12V supply. Any suggestions for a reasonably inexpensive bench supply?

-Matt

[mdrobnak]

4V is normal. Thanks to there being a lot of debug info, I looked at the working charge log, and saw that when things are good it goes high.

We'll see.

[mdrobnak]

Ugh.

Finally pulled a log from the 330 yesterday, opened it, and realized the channels I moved around to clean things up also meant..I forgot to log them. So, useless data there.

On the charge port front there's so much data to go through its kinda nuts.

[mdrobnak]

M3 is instrumented to try and figure out what I'm doing wrong with my full-throttle shifts
330 is instrumented and I'm doing something silly as the stock ECU torque request vs mine seem to be about 2:1.
I have a Charge Port Door on the way from EBay to see if this moves me further along. Until then, I've found a 1.3A current-limiting power supply:

For $10 from Best Buy, I bought a supply which came with multiple plugs. A Deutsch connector is more fitting, so there you go. Lots of capacitance in there, as that was a good 30+ seconds after unplugging and there was still enough voltage to keep that LED going.

-Matt

[mdrobnak]

To those trying to get the CP ECU to work: good luck.

I can't even get it to open the damn door.

-Matt

[mdrobnak]

I'm going to take a more methodical approach here today.
I am going to try and open the charge port door from the touch screen, that way there's no timing involved regarding pushing the button on the charge cable.
I'm going to attempt to put the scope on the 3 pins - prox, fault, and latch, and see what they do. The latch pin is..odd.

I'm then going to take the EVTV box and turn it into a CAN Gateway and try and filter out only what's needed to make the port happy in an idle state. Because as of right now it's quite a frustrating experience, so definitely need to step back.

-Matt

[mdrobnak]

Ah... Ahhhhh, a-ha!

They are NOT replay-attack resistant; rather - they support multiple units!

I replayed the traffic capture from my car on the purchased ECU. Nothing.

I pulled the connection off the HV ECU connector and replayed from my laptop (filtered out for the original messages from the CP ECU, of course)...and the door opened.

Now to just figure out how to do address the thing.

-Matt

[Isaac96]

Right. What you're seeing is the firmware difference -- this is why they want you to have a software upgrade ready when you replace parts.
It's possible to telnet into a Model S to force a firmware redeploy, but the 3 may or may not be more locked down. The Toolbox software can certainly do what you need, if you can either get it or find a helpful service center.
Or just walk around looking for 3's and steal all their CAN logs
-Isaac

[mdrobnak]

Maybe. That is yet to be determined.

However, I have the minimum required to make the door open, the minimum to have a non-red LED, and the minimum to set Number of Alerts to 0.

Next steps is to see what happens manipulating the proximity, latch, and fault lines, as well as finding the serial number. After that, try and replicate on the eBay unit.

-Matt

[mdrobnak]

I'm unsure what I saw on one of the IDs, but it was not an (obvious) serial number comparing to the label on the outside of the CP ECU on my model 3. So that wasn't it.
Manipulating the proximity, latch, and fault lines did not change a value which I was expecting a change on my Model 3.

I did however have trouble reproducing my results from last night when I went back today. Until I switched CAN devices. It seems SavvyCAN / the CANDue board is a bit processor hungry. It was not able to smoothly transmit the info required. When I put the same playback through my ValueCAN device with Vehicle Spy, voila, the door opened on my command.

Ok, so now lets try and replicate this. Lesson: Don't ignore your discrete signals!

Still no go, even when brought down to the bare minimum required to make the door open. Ok, what's the difference here. The latch. When I set the latch signal to what it should be...Voila! The door opens and closes on my command, on the bench. Now we're getting somewhere.

There's some slight variance, as even though I'm replaying the 'no error' filter, I still have 1 alert, and the LED is red, not white. Very close though. Definitely the most progress I've made in some time.

-Matt

[mdrobnak]

Later I'll try and plug in the charger and see what happens now that I know I really need to manipulate the latch.

-Matt

[Isaac96]

Awesome!
Is this your ebay part?

[mdrobnak]

Awesome!
Is this your ebay part?

Yep. Thanks for reminding me. Updated first post details and total.

-Matt