- Details
BMS Update
This is my new BMS design, based on the reference design of the Analog Devices DC2260A BMS demo board that I have been using to characterize the Bolt battery modules since I got them in 2021. (See the Blog below Battery Maintenance System (BMS) Link). The BMS demo board was great for understanding how the BMS integrated circuit, the LTC6811-2 worked. The BMS demo boards also helped me to understand more about the Bolt batteries. After using the BMS demo board I decided that I would like to have a BMS system in my car, especially since the Bolt battery module is wired for BMS connections. But the system would be more like what Tesla does where all the BMS electronics are at the battery and they communicate with a central processor. In the Chevy Bolt all 100 battery cell connections were brought to a central processing system. That required a very long and complex battery wiring harness, which I really did not want to deploy in my car. Even though the individual cells in the Bolt battery would only have a max voltage of 4.1V, the whole pack would be at 370V, which means some part of that battery wiring harness would have that high voltage on it. Running a 90 wire cable to connect my 9 battery modules together though my car is just not feasible.
The DC2260A is a demo of a battery maintenance system. I detailed in the blog below of how the demo board has circuitry to trim the individual cell voltages. I think the greatest information I got using the DC2260A to characterize my Bolt batteries is that after nearly three years of using the batteries, with many charge and discharge cycles and no maintenance of the battery cells, the average cell voltage of 90 cells only varies +/- 1mV! (See Blog below Bolt Batteries Link)). With that kind of tight distribution and no change with many cycles it appears that the batteries do not need a battery maintenance system but rather a battery monitoring system. The DC2260A is still a great design and would work great as a battery monitoring systems so I adapted it for my car. The printed circuit board CAD files were available from Analog Devices for the BMS demo board so I used the same PCB design as the DC2260A but without the Isolated SPI (IsoSPI) circuitry. I am not populating the MOSFETs and support circuitry for the battery cell trimming. That is why there are blank pads on the PCA images below. I kept that circuit design just to future-proof my design in case I find I need battery maintenance. I detailed in the Battery Maintenance System (BMS) blog below about how the IsoSPI circuit of the DC2260A works. To use the IsoSPI requires another small PCA in addition to the DC2206A and the Linduino PCA that has to be used with the BMS demo board to measure the battery cell voltages. As I describe in that blog all those boards add up to an expensive system for the 9 battery modules in my car. That was the motivation for making my own design and building the boards. To communicate with the boards, once they are deployed on the batteries my circuit design incorporates an isolated CAN BUS circuit, instead of the IsoSPI. I already have a CAN BUS processor in my dashboard for reading the CAN BUS messages from the inverter and controlling the instrument cluster. Adding more CAN BUS frames to process is much easier than adding an IsoSPI circuit to my instrument cluster controller. The CAN BUS will only require two wires to the BMS board and they all can be connected by the same two wires because all CAN BUS communication is addressable. I also incorporated an Arduino Nano Every as the main processor. It controls and reads the LTC6811-2 IC using SPI for the cell voltages and then formats the frames and sends the data, also using SPI, to the isolated CAN BUS.
By not installing all the circuits for the battery cell voltage trimming saved money on the build. The CAN BUS integrated circuits are very inexpensive because they are used in nearly all cars now. The Arduino Nano Every is much lower cost than other micro controllers because they are very popular. Also the program that ran on the Linduino to communicate to the demo BMS is an Arduino sketch that was very easy to adapt for the different processor on the Nano Every. Most of the programming was already done to access every feature of the LTC6811-2. I was able to get samples of the LTC6811-2 IC so those were essentially free, but that will be the most expensive component if I have to buy them at $22 each. The Isolated DC/DC converter ended up the most expensive component for the first builds (at $9 each) - all total the assembled board is less than $70 each, including the cost of the LTC6811-2 IC. I built up a couple of boards and they match the commercial BMS demo board for cell measurement with only a 400 microvolt offset, that is 0.0004 V. The measurement error specified for the BMS integrated circuit is 1.2mV which is 3 times larger than what I am measuring so I am happy with my design.
The BMS board will be housed in a small CINCH enclosure shown below. The CINCH is a water-tight plastic enclosure with an IP67 rating. The enclosures were designed for automotive systems. It has structures inside the enclosure to support the PCA securely. I have blank front plates that I will machine for the DB25 connector. The DB25 connector was another improvement that I incorporated in my design. The original DC2260A has a screw terminal for connecting the battery cell connections. (See the Blog Bolt Batteries below). I put DB25 connectors on all the batteries because I had attached a DB25 to the DC2260A to make it easy to connect the demo BMS to all my battery modules. Those battery connections will now be used with my new BMS design.
Just so there is no confusion because I talk about both, I used the DC2260A and the DC2259A BMS demo boards from Analog Devices to test my batteries and to understand how the BMS IC works. The only difference in the boards is the DC2260A has the LTC6811-2 and the DC2259A has the LTC6811-1. And the only difference in these two integrated circuit is the IsoSPI, the LTC6811-2 is addressable, the LTC6811-1 is daisy chained. But since I am not using the IsoSPI it does not matter. The same code is used to control either and they are identical circuits on the battery input side which is all I am using the integrated circuit for. My CAN BUS circuit, just by the nature of CAN BUS makes my BMS boards addressable. The only advantage I can see in that is being able to load different battery trimming parameters in different battery modules. But again, I am not using that function of the IC. Just like keeping all the trimming circuitry I can use that if needed in the future. Since my BMS design is working so well I now plan to only use my BMS for the 9 battery modules. It will be easier to integrate into my car and I will not have to add any additional circuitry to my instrument cluster controller.
Add a comment
- Details
Tesla Drive Unit Measurement
The photoshop image in the blog post below looks great, but that is only from one view. The Tesla drive unit fits laterally without any issue. The more important measurement is how the unit fits in the Z direction, under the car, at the level of the output shafts. Below are two images of the measurement of the differential position from each side of the car and an image of a Test drive unit with a scale. The Tesla drive unit, in the area where the output shafts connect, is about 12” tall, with the output shafts centered in that 12”. The differential images shows that there is just clearance towards the bottom of the trunk, but when the Tesla drive output is centered on the current output shaft axis, the Tesla unit will hang nearly two more inches lower that the current differential. You can see in the lower image how far down that would hang. That could be a ground clearance issue as the car is already lowered. The only way to have the bottom of the Tesla unit at the same level as the bottom of the differential is to cut a hole in the trunk so that at least two inches of the unit would be sticking out into the trunk and the spare tire well. But there should still be enough room for two batteries and several other components like the DC/DC converter and the battery pack charger. The drive center will be above the current output shaft axis so the output shafts from the Tesla will be angled down to the wheel hubs, but the CV joints should be able to handle that small angle just like they do as for how the independent wheel suspension moves.
- Details
Tesla Drive Unit Photo Fitting
The biggest unknowns about using a Tesla drive unit is how it will fit and how it will attach to the car. Since I am a photographer and I know how to use photoshop I made a photographic fitting of the Tesla small rear drive unit to my car. First, I took some pictures of the bottom of my car directly below the differential. I then made some measurements so I could scale the photographs. The distance from the end of the axle carrier to the side of the spare tire well is 21”. The Tesla small rear drive unit is 22” between the mounting points. So, it seems there is room to fit the Tesla drive unit. But the most important alignment is that the center of the drive section must be at the center of the output shafts. I grabbed a screen capture of a Tesla small rear drive unit and brought that image into photoshop and embedded it in the image of the bottom of car. I adjusted the size of the Tesla unit so it matched the scale of the photo. From this exercise the Tesla drive unit will fit with the drive section aligned to the output shafts. Toward the front of the car a simple mount can be used to bolt the Tesla drive unit to the axle carrier once the differential is removed. But the image shown below indicates that the other end of the Tesla drive unit mounting will have to be inside the spare tire wheel well. That will actually work out well as there are frame rails on each side of the spare tire well to tie a support structure for the Tesla drive unit. The mount that is on the left side of the motor in the photo can be changed to also tie into the frame rails. My Brusa battery pack charger is the only thing in the spare tire well and that can be located to somewhere else in the trunk because some of the batteries in the trunk will most likely be moved to the engine compartment. Other components that might have to move to the trunk are the DC/DC converter and the primary bus connection. The design of the car currently is engine compartment centric. That will all have to move to the trunk because the drive unit will be in rear of the car. The water cooling required for the Tesla unit will be another complication. The reservoirs/ pumps and cooling fan/radiator are in the front of the car. I will have to determine if pumping coolant from the front of the car to the back makes sense. A quick calculation of weight indicates one or two batteries will need to reside in the trunk. The other three will go in the engine compartment in a new battery box that will have to be designed to be weatherproof. The placement of the batteries will depend on the weight distribution once the Siemens motor, 5-speed transmission, driveshaft and rear differential are removed and the Tesla drive unit installed.
- Details
Rinehart No-Go, Tesla Drive Unit, Yes-Yes
I have decided not to return the Rinehart for evaluation and repair. I am worried that the repair could be as much as a new unit. Even if the repair was a fraction of new unit and they test the system before returning it to me, there is no guarantee that it will work in my car. It is odd that I have had two other inverters in the car and they both worked, after initial startup issues. But I worked continuously on setting up the Rinehart for over two months and could only get it to enable, but not turn the wheels. I would rather have a new unit than one that was repaired because they will have only repaired the failed components and their tests might not reveal what the issue was that caused it to fail in my car. I have said to people that if I was doing a conversion today that I would put a Tesla drive unit in it. Used Tesla drive units have been steadily coming down in price as the number of Tesla cars are on the road and consequently the number of wrecked Teslas. The Tesla drive unit is complete package, motor, inverter and output drive. Just a controller and appropriate accelerator and brake pedals are needed. The biggest unknown for putting a Tesla drive unit in the 320e is rear drive structure strong enough for the doubling in torque and nearly tripling in horsepower a small Tesla drive unit would be capable of. The Siemens/DMOC was never tested in the 320e on a dynamometer but the test vehicle for the Siemens/DMOC showed 200 ft-lb and 160HP, so the rear differential may have already seen 200 ft-lb of torque. The Tesla Drive unit would not be connected through the differential, however. In fact, the whole drive train would be removed, Siemens motor, Getrag 5-speed transmission, drive shaft and rear differential. New output shafts would have to be fabricated to fit my car. Some kind of mounting frame would also need to be fabricated to hold the Tesla. The rear axle carrier and trailing arms would have to be looked at for strength. I might see if someone could model that for me.
Add a comment- Details
Rinehart Inverter Failure
This inverter has been very much a problem from the beginning. Mechanically it was very easy to install, to get it mounted on the same mount as the Scott Drive, connected to coolant and connected to all the low and high voltage lines without any issues. But electrically and programmatically it has been problematic. First, I had to create a completely new wiring harness because the Rinehart uses Ampseal connectors and the inverter I replaced used Epic. The motor encoder control signals were also on an Ampseal connector so I had to build interface to the encoder cable that has Sealcon M23 round connectors. Even the cross-over box that I used to break out the wires had to be significantly changed and new wires had to be run for more control. I found that the inverter requires a very specific set of conditions of connections and firmware settings to be met so that it will enable the output. What makes the Rinehart great is that I had access to every firmware value that affects the opertion of the car. But that is also what makes it difficult. Even after working over the past month all the conditions still were not satisfied to enable torque output. The inverter shows that is enabled and in Run mode, but nothing happens when the accelerator is pressed. I have been getting support from Cascadia Motion and working through the issues. They discovered reviewing the output from the firmware listing is that the Hardware (HW) version had changed. That is not possible outside the factory so some transient in the system corrupted the EEPROM memory. They thought that updating the firmware may fix the inverter issues. Updating firmware in an inverter is not a new thing for me, I had to do that on both the Scott Drive and DMOC. For the Rinehart it just required the serial interface and the connection to the Software upload enable pin. I was preparing to do the firmware update, I had gotten all the connections I needed, but when I turned the car off to make the connections, I hear a loud bang and I could see the Rinehart visibly shake (I could feel it in the car too). There are no mechanical switches in the Rinehart so that noise could only be a large electrical discharge inside the unit. I tried to turn the car on, but the main contactors would not close. Those are now controlled by the Rinehart. I discovered that the 12V 5-amp fuse to the Rinehart was blown. I replaced the fuse, but when I turned the car on it blew it again. I measured the input impedance for the 12V into the Rinehart and it is several mega ohms so this is a semiconductor short – when the circuit is energized it draws too much current. I checked with the support from Cascadia Motion and they said a 10-amp fuse would be okay, but I found after a few minutes the inverter would blow that one too. It was after the 10-amp fuse blew that I could smell the familiar smell of burnt electronics. So now the Rinehart is probably bricked. Cascadia Motion does have an RMA and repair service. They charge $650 to evaluate the unit and give you a quote for repair. The problem will be what actually failed. Was it just the controller board or did one or more of the IGBT transistors blow too? Something had a large electrical breakdown to make the unit move and create that loud bang. I requested and received an RMA, but I will have to think about if I even want to know what went wrong. If it is something major the repair could be as much as a new unit. That unit won’t even turn on with is not promising and the $650 could be just a waste of money. I removed the inverter from the car, it had to be cleaned of cooling fluid and the inlets capped, in case I decide to send it back to Cascadia Motion. While I had It out, I put it on my test bench. With a DC power supply that was only 40W I was able to turn the inverter on. It drew more current when it started up and pulled the voltage down on the supply but that only lasted a second or two. When it settled it was only drawing 650mA. I was able to connect to it with the RS-232 and using their GUI I found that it now showed multiple hardware failures. And the HW version had changed again. I took a screen shot of that and sent it off to Cascadia Motion support. From those faults they said it’s most likely the power module (IGBT), potentially the gate drive board too, but not the control card and not likely the IO board. The repair could be somewhere between about $1500-$2500, but I would know without sending the unit back. The timing is also an issue. The RMA request said that process of evaluation my take 4 - 6 weeks and the repair could be twice that, even longer if they do not have the spare parts. Since the Rinehart is now removed from the car I went ahead and put the Scott Drive back in because I would like to drive the car now the seasons have changed. I had retained the wiring harness and mechanically did not alter the engine bay so it dropped right in. I did not even change the cooling lines. I had bought all new hardware because Rinehart uses AN fittings and the Scott Drive is barb hose fittings, but I never got a chance to change the fittings. The Scott Drive uses many fewer inputs than the Rinehart so it only took a day to rewire the cross-over box. I started up the Scott Drive and run the Scott Drive GUI. The throttle and brake seemed to still be calibrated with the offset and gain that was in firmware and the motor spun right up when the accelerator was pressed. The regen even slowed the motor spin down with the application of the brakes. I just have to clean up the connections inside the car and then I will be able to take the 320e for a test drive. I will post another blog after I decide what I am going to do with the Rinehart.
Add a comment