[black-cat]

Lockdown has given us a chance to do some long overdue work on our Insight. We were very lucky as about a week before the lockdown the IMA pack finally said 'no more!'. It had been complaining about low capacity for a couple of months...

Anyway, today we had first IMA start and brief drive on our BimSight lithium ion conversion project so I am rather happy. I have been working on this with another forum member who has an equally dead IMA pack.

You will have to excuse the very sketchy install as this was a proof of concept/initial test fit so it is all coming out again and getting a custom made bracket to hold everything in place. And yes, those are 2 magazines shoved in there as temporary insulation.

After an aborted plan to use some second hand but lightly used (and as it turns out uterly dead) 15Ah pouch cells I saw the BMW PHEV modules on the open inverter forum. It turns out they are a very nice fit in the Insight case! And someone had reverse engineered the CAN protocol!!

We bought a 2018 vintage 530e pack from a front end smashed car which yielded 6 modules each with 16 26Ah Samsung SDI cells. On the end of each module is a CAN connected supervisor which measures cell voltages and temps. It even has balancing build into the supervisor.

The stock BCM has been replaced with a custom board running a Teensy 3.6 module and a LEM 200A current sensor. The Teensy intercepts the METSCI (and can re-write the dash output) and generates a BATSCI to keep the MCM happy.

The test drive was all of 50ft as nothing is secure and the car is missing its front seats as part of a very long overdue deep clean. But it did show 4 bars of assist under the very brief acceleration. And no dash lights or codes showing on the OBD2C&C.

Next job is to weld up a support frame for the modules and the fuse/contactor assembly as well as get the seats back in ready for a longer drive. Still a lot of coding to do to get it to measure SOC and ensure it goes into low power mode when not running so it doesn't flatten the 12v battery

Not sure on cooling at the moment. In the BMW the modules are liquid cooled BUT the motor is 83kW! So hopefully in the Insight it will just need some forced air blowing over them.

 

[retepsnikrep]

This is a very significant development for the Insight ZE1 community.

Please keep detailed updates coming. Thanks

 

[Jonny69]

Very neat install.

The 530e pack is 354V and 12kWh, so those modules are just under 34Ah. Amazingly compact for that much capacity.

Edit: actually, those 2018 ones might be 9.2kWh, so 26Ah assuming the voltage is the same. Looks like the capacity was increased in 2019.

Edit 2: or what I could actually do, is read your post properly because you actually told us what the capacity is. Duh!

 

[black-cat]

Yup, these are the previous generation BMW PHEV modules at 26Ah. The 2020 spec went to 34Ah.

I will give more detail after more testing. Running for about 2 mins and driving up and down the driveway does not make for a conclusive test

 

[jime]

On the end of each module is a CAN connected supervisor which measures cell voltages and temps. It even has balancing build into the supervisor.

Do you see a way to make the 3 modules talk to each other so that the entire back balances across the 3 modules?

The stock BCM has been replaced with a custom board running a Teensy 3.6 module and a LEM 200A current sensor. The Teensy intercepts the METSCI (and can re-write the dash output) and generates a BATSCI to keep the MCM happy.

Innovative. Anxiously awaiting more development and test results. It certainly is an easier package to install that the low specific energy LTO. What is the chemistry?

[/QUOTE]

 

[black-cat]

I believe they are old school LMO-NMC cells. So not as forgiving or long lived as the newer chemistry cells but they are from a good supplier. We'd been looking at LiFePo4 and using an Orion BMS but when I found these modules and we discovered they'd fit in the OEM enclosure it was an easy decision.

To balance the cells you send a target voltage to the supervisor and it will discharge any cell that is (I think) more than 5mV over the target. The balancing is not automatic so the BCM replacement board will have to look after that.

The cells are dated march 2018, the car had 44k on the clock and the scrapyard said the car had been with them for a year. After all that time the spread across all 96 cells is 9mV.

My current concern with the BCM replacement is the current sensor. We will have to see how fussy the MCM is about the current reported by the BCM

 

[Puggie]

Do you see a way to make the 3 modules talk to each other so that the entire back balances across the 3 modules?

They should do this with the BMS, the balancing is CAN controlled so your main 'battery brain' (Accumulator Monitor System is a common name) that talks to each module should manage this.

Innovative. Anxiously awaiting more development and test results. It certainly is an easier package to install that the low specific energy LTO. What is the chemistry?

They are NMC 6:2:2,

 

[retepsnikrep]

My current concern with the BCM replacement is the current sensor. We will have to see how fussy the MCM is about the current reported by the BCM

It's not super fussy. See attached..

But you will want the BCM Current fairly accurate if you are going to count current.
I don't know what ADC resolution the Teeny has for your current sensor output 10/12 bit?

 

[black-cat]

Ta, that is a useful bit of info.

The teensy has 16 bit ADCs but I'm using 14 bits to aid in noise reduction. Our resolution is in the order of 32mA which is better than the stock BCM's 48mA resolution.

 

[jime]

Is it possible to reconnect the liquid cooling arrangement again. That would be the most effective. Water cooling is a lot more effective than poorly engineered air cooling

 

[Puggie]

Its aircon based, hooked into the evaporator circuit. Its hellishly over engineered. Its not like we are going to stress these cells with the insight, the IMA motor will give up long before these show much stress.

 

[black-cat]

Hellishly over-engieered is an understatement! We will likely end up putting a good finned heatsink on the base of each module and then a fan or fans to pull air over them.

 

[Puggie]

The cells do not want to be over cooled either, After some test miles we can get an idea of their temperature rise and see what kind of cooling setup they will need. Even with current hacks we will not be pulling more than 20kW, I reckon some basic heatsinks and simple air cooling will be more than adequate for these (we pull 80kw from a A123 20Ah pack with simple convection cooling, and they are a much older cell). No point in making it over complex, when you start pulling hundreds of kw from battery pack then cooling becomes an issue, but we are really not asking much of these.

 

[Harborman]

The cells do not want to be over cooled either, After some test miles we can get an idea of their temperature rise and see what kind of cooling setup they will need. Even with current hacks we will not be pulling more than 20kW, I reckon some basic heatsinks and simple air cooling will be more than adequate for these (we pull 80kw from a A123 20Ah pack with simple convection cooling, and they are a much older cell). No point in making it over complex, when you start pulling hundreds of kw from battery pack then cooling becomes an issue, but we are really not asking much of these.

I’m not knowledgeable in battery chemistry. Would you expect more heat to develop in this new battery than the original Nickel Metal Hydride? Are these new cells more or less sensitive to heat? I own a Volt, and I’ve read that GM over-engineered the liquid cooling. I suppose they didn’t want any chance of battery failure like Honda experiences with our G1 Insights.

 

[Puggie]

Massively less heat for the same power output. These cells have a much lower impedance, so they are more efficient at converting chemistry to electricity. This means less heat produced for equivalent power output.

Lithium cells should last longer if managed properly. I think the biggest issue we are likely to see in the insight is lack of a decent load to run them hard. They dont like the cold and lack of use. These have not been engineered for the insight, more chosen out of convenience, so there specs are good if not ideal.

I dont think the original honda failiures are due to heat, just a combination of age an poor (as in none whatsoever) battery management. Good state of charge and temperature management is key for longevity of large battery packs.

 

[black-cat]

Good day today, got the roof trim clips replaces on both mine and Puggie's insights. Both have had windscreen replacements at some point so there are some broken bits to contend with.

And got some time with the welder to make a frame to hold the BMW battery modules. It is a lot less hacky than the first try. No magazines for insulation or shoe boxes but there is some duck tape

We also managed about 10 miles of road running with the BMW modules with a laptop watching values and Puggie has driven home. We found a nice big hill and managed about 60 seconds of full throttle and full bars of assist. It peaked at about 12kW and dropped to 6.5kW (ish) for the rest of the run up the hill. We dropped 2 volts! We stopped at the top of the hill, batteries were cold. We saw a 1degC rise during the 10 miles of running.

I still have a goodly amount of software to write. The BCM replacer has a boot up issue as it takes a while to get all the CAN data so the MCM throws a hissy fit at being told 'no volts!' over BATSCI. I now have a second board so I can do some work off the car and have a permanent on car test rig.

 

[Puggie]

Hopefully I should put about 90 miles on it probably 4 times a week. It will be interesting to see how this affects my daily mpg as my commute only usually varies by about 2 minutes and 1.5mpg from day to day. Its certainly nicer to drive with a bit of assistance again!

I dont think even finned heatsinks will be required for thermal management. We need to keep tabs on the cell temps, but I expect the motor thermal capacity will be a much bigger hindrance to performance. I suspect just a bit of airflow over the cell modules will be adequate. I joked earlier but in cold weather getting the cells up to working temperature could be a bigger issue than keeping then cool.

 

[black-cat]

It would be interesting to set no regen one day and see how much battery you use and if the MPG gets any better.

 

[Puggie]

First run to work.
50 minutes and 44 miles, fractionally quicker than normal by a couple of minutes. Started with 164v, finished with 151v. I did drop to 118v under load at one point which cut the assist off. Regen will suck uoto 40A briefly and all seems to be behaving. My mpg has risen, in the last 200miles 70.5 (that is pretty bang on for my commute), end of this journey 71.9 cumulative on the previous 200, so I'd think that is nicely over 75mpg for this journey. I've reset the trip counter to see what the journey home brings.

Cells are still stone cold!

 

[black-cat]

Nice! Time to get the SOC working so it can adjust assist and regen properly.