[DRIVING] VW Touran powered by Nissan Leaf [FINISHED]

[JSJFIN]

Here is some info:
https://www.diyelectriccar.com/forums/s ... html?amp=1

[johu]

So today I was gonna fit the Nissan charger on top of the inverter. Not that I'm very keen on actually using it but it makes a nice, sealed DC junction box and Nissans main battery and ChaDeMo cable would fit in seamlessly. Certainly a plus for passing the road worthiness test. Also the DC/DC converter would have come in handy.

But: it doesn't fit! How can you make a stupid 3kW charger THAT BIG?

I removed the firewall cover to fit the charger but of course then the firewall cover doesn't go back on.

So I took it off again and wired the battery cable directly to the inverter. Will have to cover that with a box. Also need to think of some switch gear for the ChaDeMo port now.

On the plus side I don't have to go down the rabbit hole of reverse engineering the charger and DC/DC.

[arber333]

You have to be careful with this cover. It is intended to steer drain water out from under the hood. In case you would cut it somehow you can change the flow of drain water and have a flood in the wiring...

[johu]

Yes I thought so too. Thus the decision has been made for me (by the cover): I will not use Nissans bigass charger.

[arber333]

Yes I thought so too. Thus the decision has been made for me (by the cover): I will not use Nissans bigass charger.

Could you make a sheet metal case for the charger that would be smaller? You could still use it then...

[johu]

I probably could, but the main reason for trying to use it was TÜV. It looks good, the cables are well routed etc. So it was more about the case, less about what is inside.

Todays news:
DSC light is off! I replicated the ECU CAN messages as far as possible and it stayed off. Will document it later in the wiki. Bad news: it doesn't stay off. more on that below.

On other news I did another test drive. So I can enjoy "full" torque till about 4500rpm, then it drops quite suddenly and I can see Vd (field weakening voltage) maxing out. I can drive till about 7000rpm then torque becomes so low that it doesn't accelerate anymore. Good changes: no more trips and no more hopping around when rolling in the "wrong" direction (like forward when in reverse gear). The latter transitions really smooth into regen, so say you roll backwards the car regens backwards and you can then catch it with the throttle and go forward again.

Now, back to CAN: There is some quirk that results in no messages being sent for about a second. Timeout -> DSC light on again. It happens quite regular, every 75s so it must be systematic. Weirdly it doesn't happen simultaneously for all messages (this rules out a scheduler problem). So it stops sending message A, some time later message B etc. I've tried all sorts of things in software, also deleting all but one message. No improvement.

The TX error bit is not set but I get quite a few "arbitration lost" messages. At the instance when no message is send, the arbitration lost does not increase more. I also plotted the TX queue length and you can see it building up when all 10 messages are enqueued and then go down again within about 1ms back to 0. Also no correlation with the lost messages.

Any ideas?

[Jack Bauer]

I have seen similar using a Sam3 communicating with the E65 7 series. I believe it is a CAN arbitration problem. Are you using message filters on your end? Suggest setting a receive none filter and see if the problem persists. Another possible cause is the ack bit not being set. Are there messages destined for the ecu that you are filtering out? if so they won't set the ack bit and this can annoy other modules and of course if you are receiving all messages you are automatically acking all messages which will annoy the other modules who want to ack the message but find someone else has already done so:) I managed to solve it on the E65 by filtering everything except the messages bound for the ECU.

[johu]

Yes filters are set. If I clear the CAN config (which I did) it will set filters to receive none. Well, except 0x601, maybe I can try removing that. It's not on the bus though.

Has anyone ever observed lacking messages from the inverter?

[johu]

Just did an isolated test with the same software. Messages pour out as expected. So yes it must have something to do with the bus topology. So while the controller in question is connected to it at the original ECUs tap, the inverters tap forks off. Could that be causing issues? I'm also not sure about termination. The bus should read 60Ohm when the car is shut down, right?

[Jack Bauer]

Yes should be 60 ohms when powered down. Should be no problem having the inverter tapped off the bus.

[johu]

Found it!

May I cite from the STM32 manual

NART : No automatic retransmission
0: The CAN hardware will automatically retransmit the message until it has been
successfully transmitted according to the CAN standard.
1: A message will be transmitted only once, independently of the transmission result
(successful, error or arbitration lost).

Guess what? I'd set it to 1.
So setting it to 0 solves that problem but makes debugging impossible, at least with the Analog Discovery. With only that connected the STM32 never stops re-transmitting the very first message. Maybe I should have it as a parameter. On a busy CAN bus I'd say it's a must.

DSC light off, forever, yay

[johu]

Video on fitting the battery

[Jack Bauer]

Might be able to use the analog discovery with a can transciever. I'm guessing it's retransmitting the first message due to no ack.

[johu]

Measured the cell taps today by attaching the multimeter to B- and equip the other test lead with a cut off resistor leg that fits into the plug.
So it looks better than initially thought. The front brick has the cell taps arranged exactly like the leaf, only it is mid-pack to B+ instead of B- to midpack. But if I understand correctly the two halfs are galvanically isolated and can be swapped either way.
The other plug needs rearranging as it used to be connected to the left and right half in a convoluted arrangement. Yes it will be sort of painful to remove 49 crimp leads from the plug while under the car but I think it's worth it. Any hints on how the crimp parts can be pulled from the plug?

[Kevin Sharpe]

Any hints on how the crimp parts can be pulled from the plug?

Think we had some posts on DIY electric regarding that... look up Wolf's posts and my Leaf battery work... sorry, can't remember the details now

[Tremelune]

I believe a safety pin achieved success for Wolf...

https://www.diyelectriccar.com/forums/s ... 81297.html

I don't think they ever figured out the connector type. I don't see great value in keeping them in lieu of a more common/standard connector...

[johu]

Thanks guys, that was the thread I was looking for. Tried the technique and it works well.

[johu]

Rigged up my planned "pot simulator" in Touran today and it works nicely. What it is is Two BC547 transistors with emitters grounded and base connected to an 18kHz PWM signal via a 100k/1µ low pass filter.
So at 50/50 duty cycle you can see the fuel gauge is exactly at half-full. At 35/65 you fill get a low fuel warning and at 70/30 a full tank of gas, half a pack of cigarettes and you are wearing sun glasses.

I found that the wiper of the original fuel tank sensor is just grounded and the two pot taps probably get some constant current injected. So that makes it easy to simulate a pot without one of those digital pot simulators.

[arber333]

Do you have an idea how to simulate Tacho reluctance sensor. I need sine signal when each line carries its own part of sine signal. It seems BSM and BSI needs one to see motor is alive, but it needs both sides to show RPM .

Can anyone propose schematic on how to simulate this kind of signal?

[johu]

@Arber: maybe you need +-5V, connect emitter to -5V connect base to DC-decoupled PWM + low pass filter.

Sorry don't have time to think about it more right now, as I have my own issue: burned the Nissan BMS. I plugged in one of the connectors that I thought had the same arrangement as the original but immediatly heard frying sounds from inside with according smell. Not sure whether I traced out the signals mirrored or something, anyway some diodes and a least one voltage sense chip have an extra hole in them now

So I might end up using my own BMS instead.

Damn, this shit can be frustrating

EDIT: just double-checked: yes I had called the negative-most cell "cell 1" and the positive-most "cell 48" while Wolf named it the other way around. So as initially traced out, the wires are reversed.

[purplespark]

That's really annoying when that happens !
Don't know if it will help , but there's a 2013 bms on ebay.de at the moment -item no. 223507867094 - doesn't have the grey and black plugs though,
so I don know if that would be the right one .

[johu]

Thanks for that.Meanwhile a gentleman from UK offered to send me his left over BMS for free - In community we trust

Took reception of the counterparts today. This will allow me to extend the cables a bit and undo the wrong pin mapping. Will test with the partly-fried BMS first then connect the new one.
EDIT: mini BOM:
16-pole male 1473796-1
16-pole female 1318386-1
24-pole male 1473413-1 1473799-1
24-pole female 1318917-1
32-pole male 1473799-1
32-pole female 1318747-1
40-pole male 1376113-2
40-pole female 1318389-1
Pin male 1376109-1
Pin female 1123343-1

The 40-pole male isn't actually a crimp part. For some reason thats not available for 40 poles, even if you look directly at TE.

[johu]

Adapter done. Will connect it to the battery and verify once more that the voltages are now in order before doing any more damage.
It seems the male pins are different for different housings. They all fit except the 24-pin housing. It also fits with some force though

EDIT: I stole the populated 16-pin connector from the LV BMS harness as I don't have anything connecting to it. Anyone knows what it is?

[johu]

Alright, adapting successful.
These are the voltages.

Code: Select all

3586	3785	3776	
3776	3778	3776	
3771	3779	3772	3781
3776	3926	3611	
3780	3771	3771	3776
3772	3780	3779	
3772	3776	3776	3780
3781	3780	3776	
3780	3776	3776	3772
3771	3773	3776	
3776	3776	3781	3776
3776	3775	1225	//oh no, fried chip
4095	3927	3662	962 //another one
3894	3759	3996	//and another
3605	3762	3718	3928
3618	3754	3727	
3637	3776	3780	3780
3779	3769	3776	
3665	3930	3680	3776
3771	3920	3631	
3780	3700	3902	3645
3776	3772	3771	
3776	3772	3776	3780
3772	3776	3701	
3771	3776	3727	3754
3771	3771	3772	
3776	3771	3764	3776
3776	3767	35769	

I read them inspired by this: https://github.com/lincomatic/LeafCAN/b ... anData.cpp
First send message 79B#02 21 02 FF FF FF FF FF to get the first frame then 79B#30 01 00 FF FF FF FF FF to get the subsequent ones. Its 28 frames total with 96 cell voltages. The last value seems to be the total voltage (357.69V).
I'm guessing the weird voltages result from the sense chips that I fried. Will see when the replacement BMS arrives.

I also found I'd made a mistake when ordering the 24-pin male part. Part number 1473413-1 does NOT fit the female part and the pins don't go in all the way. There are 5 other possible parts, from the numbering scheme I'm guessing 1473799-1 is correct. Updated the post above.

[doobedoobedo]

Looks like you've got a fair amount of variation there. For the really low ones you could try bringing them up to around nominal voltage slowly and see if they manage to hold the charge.