Details

Published: 07 December 2021

Battery Maintenance System (BMS)

Initially the Battery Maintenance System (BMS) will be provided by the Analog Devices DC2260A BMS Demo board which demonstrates the LTC6811-2 BMS integrated circuit.  This circuit has the ability to measure 12 battery cells simultaneously and communicate via either isolated SPI or standard SPI. The demo board is connected to a Linduino with a 14-pin ribbon cable. The Linduino is Analog Devices version an Arduino Uno and is required to communicate with any of the Analog Devices demo boards. Standard SPI is used to communicate between the Linduino and the demo board. If multiple BMS Demo boards are used then the isoSPI interface is used and the Linduino needs a DC2792B shield.  The D2260A demo boards are somewhat expensive ($150 +S&H) plus the cost of the Linduino ($125) and DC2292B ($75) shields (9 demo boards plus 2 Linduino and 2 DC2292B would be needed). Plus an overall controlling circuit needs to be designed to take the data on the from the Linduinos on the IsoSPI connection.  A new circuit based on the DC2260A has been designed and the initial build started. The new circuit design incorporates an Arduino Nano Every processor and an isolated CAN BUS interface.  The Nano communicates with the LTC6811-2 to read the cell voltages and then sends that data out on the CAN BUS.  A much cheaper implementation (almost one-third the price) and there already is a CAN BUS system in the car the controls the dashboard instruments.  Probably the first 4 battery modules will be used with the DC2260A  BMS demo boards and the other battery modules in the car will be monitored with my circuit design. The design of the new PCA can be seen (here).

The screen picture below shows the Multicell Monitor GUI during the discharge process.  In normal operation in the car, only cells that are at a higher voltage would be discharged. An algorithm will be written to control that process.

Top side of DC2259A  BMS Demo board showing active discharge channels.  When the MOSFETs are turned on to discharge a battery cell the LEDs light up. Again, 6 and 12 are not lit because they are not connected on the 10 cell battery module.

Bottom of DC2259A showing the discharge resistors.  The resistors are 33 ohm so on a battery cell that varies from 3.5V to 4.1V there can only be a little more than 100mA discharge current. On a cell that is 180AH that would take 1800 hrs of time to fully discharge (75 days). This board is not designed for that.  The design is to just trim the cell voltages so they are all the same value.

Image below shows the demonstration of the isolated SPI (isoSPI) that can be used to connect up to 5 BMS demo boards to one Linduino and a DC2792B isoSPI shield.  In the car there if that system was used there will need to be two Linduino/DC2792B controllers for the 9 Bolt Battery modules. The iosSPI is limited to 5 boards so that is why it would require two Linduinos and DC2792B shields.  Initially the IsoSPI will be used for the first 4 batteries and the remaining 5 will use the BMS board of my design with the isolated CAN Bus communication.

To see and Update on the BMS Click Here

Details

Published: 02 May 2021

Chevy Bolt Batteries

To get more driving range and because many of my original CALIB batteries had degraded, the batteries were upgraded in late 2020. The upgrade batteries that were acquired new are the Chevy Bolt 5.94kW-h module shown in the picture below. These modules are made by LG Electronics and are NCM (Lithium Nickel Cobalt Manganese Oxide) chemistry. Each module has a 180AH capacity and the 9 modules that will be used gives a 53kW-h battery capacity. For a 85% discharge that provides 45kW-h of driving capacity. At the measured power consumption of 292W/mile by the vehicle that capacity should provide just about 155 miles driving range. Although the new batteries are bigger and weigh more than the CALAB cells the new batteries will only increase the weight of the car 100 lbs but increase the driving range by 2 times!  A great feature of these modules is that they are already wired for a Battery Maintenance System (BMS).  Each module has 10 cells wired in series and there are connections to each cell on the module so the individual battery voltages can be measured.  The connections are a pair of multi-pin connectors at one end of the module. The Chevy Bolt uses a central processing unit for the battery maintenance so in the Bolt each battery module has a pair of cables that runs back to the central controller. For my build each battery module will have its own BMS controller. This is easier than running wires from each module up to the engine compartment. Although between the wires from one module the most voltage would be 41V (full charged battery) the wires will have the full pack voltage, with respect to ground. Local BMS control will be safer at this point in the 320e build. I was able to order the Chevy Bolt battery wiring harness on Chevy Parts Online. By taking the wiring harness apart I could build several cables with the corresponding connectors already attached and used them to connect to the local BMS controlling circuit. A BMS demo board, the DC2259B made by Analog Devices was initially tested to use for the module BMS.  The connections to the battery module and a BMS demo board are shown in the second and third photos below.  

BMS connection on battery module.  These connectors were obtain from a Chevy Bolt battery wiring harness that was purchased online. The connectors are keyed and lockable.

I did not have a wiring diagram of the BMS connections on the Bolt battery module.  I assembled a couple of barrier strips and broke out all the wires from each connector to determine which wires were the pairs for each battery cell in the module. I highlighted the connections for the first cell. Using this method allowed me to make a connector to connect the BMS demo board. However, the wiring order had to change for connection to the BMS demo board.  True that in the image M11 to M21 is the first cell.  For the BMS demo that M21 wire has to be next to the M11 wire (green and cyan).  By alternating the wires from each connector the battery voltage steps up for each connection, relative to the battery (-) connection because the cells are all wired in series.

A DB25 connector is used for the wires coming from the battery module to make connection the BMS demo board because the demo board is only supplied with screw connectors.  The DB25 was installed on all the modules to enable the test of all the batteries with one DC2259A Demo board.

Pictured below the DC2259A BMS demo board that demonstrates the LTC6811-1 BMS integrated circuit. It is the 48-pin integrated circuit at the center of the board. The other components on the board are a set of transistors and resistors for trimming the battery cell voltages and a group of circuits for data communication.

Below is a plot of the average of the ten cells in each battery module that was measured with one DC2259A BMS demo board on 12-21-2020. This measurement is of the as-received modules from the factory.  For effectively 90 cells the cell voltage variance is only +/- 2.5mV.  If the two modules that show the largest difference we adjusted to match the other modules the variance would drop to +/- 1mV!

UPDATE:  Below is the average cell voltage for 8 of the battery modules, measured recently. The batteries are at a different state of charge (SOC) but the cell voltage distribution is even tighter, less than +/- 1mV.  The 9th battery module was not included because it is being used as a bench test platform and the battery box to house the module has not been built yet, so it is not in the car connected to the other battery modules.  No adjustment of the individual battery modules was done.  This is after nearly three years of charge and discharge cycles.

 Read an update on the Bolt Batteries here.

Details

Published: 05 November 2020

Repaint, New All Weather Tires and Wheels

One thing I always wanted to do to the car since I started the conversion was to have it repainted. The paint was still in good condition for original paint.  But the roof paint was faded and there were several rust spots around the bottom of the windshield and several dents in the doors.  The problem is that the car could not be painted with the CALIB batteries installed.  To paint the car all the batteries would need to be removed.  That is because during many phases of the painting, the car is in a closed room with an explosive atmosphere - due to the volatile solvents used in painting.  One spark could cause a huge explosion. In fact, for a whole car painting like I am having done, the painters usually remove the 12V battery just to eliminate that as a problem. As more and more of the CALIB batteries began to have charging problems (due to too many over-discharges) I decided to have the car painted because I was going to replace the batteries. I removed all the CALAB batteries and had the car towed to the paintshop. The paint job took more than two months, mostly because it was painted during the height of the pandemic. I was however, able to visit the paint shop and take some pictures of the process. Shown below is an image after all the chrome was removed and windshield was removed. The rust that was around the windshield is more apparent.  The second image is after all the body work was done and the start of priming. The third image is after the car was fully painted and during the wet sanding process, also known a cutting.  That is done to smooth out any imperfections in the paint.  Once the wet sanding is done the paint is polished and once all the polishing is completed a clear coat is applied.

The paint color is not the original BMW color code.  I wanted to have more metal flake in the paint.  The painter said he could mix some metal flake into the BMW color, but then I would have a custom color that might be hard to replicate, if I ever needed some more paint.  So the color is actually for a 2010 Ford F150 pickup truck, a shade I picked out of the color coupons, that matched the BMW color closely.  In the sunlight the metal flake in the paint really pops. You can see that in the closeup of the paint in the lower image.

Once I got the painted car home and cleaned up, I wanted to drive the car, but the painting took until mid November to complete.  I usually have summer tires mounted on the 320i (Dunlop Direzza) so those cannot be used below 35F. To be able to drive the car in the cold weather I got a set of BF Goodrich T/A Radials, all weather tires and some Enkei ENKEI92 wheels, with a gold insert. The wheels look good with the paintjob and they were much cheaper than the Rial wheels I have with the summer tires. The image below is the car with the new winter tires and wheels

Details

Published: 01 October 2020

Scott Drive Installation

Although I was part of the group that originally developed the GEVCU to use with the DMOC so many years ago, I recently removed the GEVCU and DMOC and replaced them with a new inverter/controller called a Scott Drive. Even though I have been driving the 320e for nearly two years with the new tires and suspension I have had several problems with the GEVCU. The GEVCU would just suddenly fail to control. The symptom was all the calibration data was lost so the throttle and brake inputs would not register and thereby the car would not move. Also the contactors sequencing was lost so when the car was turned on so all the contactors would close causing one to fuse closed because of the high inrush current. I had to build another contactor system and placed that in the trunk. It ran independently of the GEVCU and used an Arduino Uno to control the sequencing and the timing of the contactors. I also had to jumper around my original contactor box. The GEVCU failure happened twice over the past two years and when it happened again recently I decided to replace the whole system. I emailed several other GEVCU users and some had seen similar failures. The consensus is that the EEPROM was failing – that is where the calibration and setup data is stored. That is very weird because the EEPROM is on the isolated side of the GEVCU board with the CortexM3 processor and no one saw a processor failure. It could be that I just got a bad group of EEPROM ICs. All the GEVCU boards I had were fabricated around the same time. But that does not seem like it could be the issue since others have experienced GEVCUs that were not built by me. All of the failures I experienced happened when the car was first turned on – I never experienced a failure while driving. Fortunately I was never stranded anywhere, the failures all occurred in my driveway or garage. The GEVCU failing was not the biggest issue, however. The fusing of the contactor was a serious problem because when the car was turned off it really was not off. The battery pack was still connected and that caused the battery pack to over discharge. This now made the third time the CALB batteries were discharged below 2.5V (see blog Driving soon???). Once I recovered the batteries I found several that would charge to the upper voltage limit faster than other batteries, even though they were all bottom balanced to the same voltage. That became a problem because the pack charging would have to stop and I would have less than 60AH charge in the pack. I tested several of the high voltage batteries on the battery cycling system I have and they only showed a small decrease in capacity, anywhere from 5 to 8 AH. But I was seeing batteries max out and limit the total pack charge to 45AH.

I had purchased the Scott Drive SD100 from EV West more than 5 years ago thinking at some point I would do the upgrade. Of course the Scott Drive upgrade was not without problems. The wiring harness had to be changed for the throttle and brake and the Scott Drive has a completely different connector that had to be assembled. When I first tried to turn the Scott Drive the motor would not turn smoothly. I discovered a bent pin in the Siemens motor connector that was one of the motor phase control pins. The pin must have been bent when I installed the new control cable for the Scott Drive. Once I fixed that I then found the motor would spin smoothly, but only in reverse. The Scott Drive has a great GUI for setup, calibration and control of the inverter. But what I discovered next was that the firmware in my Scott Drive was several revs old so the controlling GUI would not work, specifically to change the direction of the motor rotation. That required a firmware upgrade via an AnyDesk session with Scott Osborn, the maker of the Scott Drive, who lives in New Zealand. It took a couple of weeks to arrange the upgrade. I hope for any future upgrades I will be able to carry out the process because Scott charges for the upgrade and the time difference makes it challenging to communicate. One feature of the Scott Drive that I don’t remember setting a limit for in the GEVCU was the maximum output current. My original 60AH CALB batteries are capable of discharging at 10C. That would mean a max current rate of 600A. The Scott Driver 100 that I have is limited to 400A. The inverter is also limited to 150kW peak, which is more than the DMOC was rated for. With 400A and my pack voltage of 375V I would get to the 150kW peak, at least for a short time during acceleration. It is hard to say if the limit changed the acceleration capability of the car. I never really measured it by anything more precise than the seat of my pants. I had always planned to take the car to a dynamometer. Unfortunately the closest shop to me that had a dynamometer closed last year and all the others were far away. Driving a long distance to one is really not an option as the pack is discharged, the voltage drops, so the output power drops. But the real problem is just having the range to go there and back. With a pack charge of 45AH that means just under 50 total miles driving range. Probably the range would be way under 50 miles because some of the battery capacity would be used on the dynamometer.

The last upgrade I did was not really planned. During one of the times that the GEVCU failed I was moving the car in and out of the garage with an electric wench. To pull it out I would attach the wench to a tree at the end of the driveway and then to pull it back into the garage I have a bolt screwed into the concrete to attach the wench. Pulling out was no problem because the 320e has a tow hook on the rear of the car. But on the front there is none so I was pulling on the main cross-over. I think what happened when I was straightening the car the tow band slipped off the cross-over and slipped onto the steering rack, thereby pulling on the tie-rod connection. That must have pulled something out of alignment in the steering rack as the steering was locked, I could not turn the wheels. Because the car was not pointed straight I had the jack the front of the car with a floor jack and slowly pull it back into the garage. I ended up having to replace the steering rack. Fortunately remanufactured racks are still available for my car. The upgrade was the mounting bushings that mount the rack to the car. I upgraded to polyurethane bushings that should give the car more precise steering. Also fortunately there are several YouTube videos on how to install the steering rack on mine and similar model BMWs. But seeing how it was done and doing it turned out to be very far apart. I hope I will never have to do that again!

A video of all these upgrades can be seen here.