Zach’s Electric Diablo

I spent tonight writing software for the battery monitors.

Warning! Warning! Super technical discussion ahead. If you are put off by such talk, turn back now!!!

For the most part everything seems to be working well. There is only one outstanding issue. I designed the circuit with a Schottky diode on the positive battery rail. This diode serves a couple purposes. First, it protects the circuit if its polarity is reversed. Second, it helps with noise immunity by allowing the 100uf filter capacitor to power the circuit during brief power glitches on the battery rails (I am not sure this will be needed, only driving will tell). The problem with the diode, is the fact that it is a diode. And like all diodes, it has a voltage drop. In this case .4v. The batteries are basically dead at 2.5v and the processor is good to 2v. So, it is not a problem for the processor. However, the output pins on the micro seem to output at a voltage roughly 1.1v less then the supply. The problem shows itself when I send data out of the processor through the opto-isolator (connected to the UART for communication). The opto wants a minimum of 1.5v. This means that with .4v+1.1v worth of voltage drop, my comms start dropping out if the battery voltage is below 3v. I am trying to think of options. I could ditch the diode, but that might come back to haunt me. What to do?

The mail man dropped off my BMS printed circuit boards yesterday.

I did a quick test fit to make sure they fit on my batteries:

I first built one, and tested it:

It worked perfectly. Here are the features it has:

1. Measures battery voltage
2. Measures temperature
3. Has a 3W 10 ohm resistor that can be programmed to shunt excess current during a charge cycle (balances the cells)
4. Talks over an opto-isolated serial bus.
5. Has a programmable RGB LED (used as a diagnostic tool and for adding some “coolness” to an otherwise boring battery pack)

Then I built 10. (This took about two hours)

Only 40 more to go…

Today, I sent off my Battery Monitoring System (BMS) circuit to the fabricators. If all goes well, I should have 50 of these boards in my hands by Friday.

I also mounted the battery box (which is now powder coated) to the chassis and bolted on the power electronics (the charger, controller, main fuse, and both contactors):

The next step is to install the batteries and start wiring everything up.  I want to make everything really neat and clean, so I will be taking my time.

Today is battery box day!

I used the waterjet to cut the pieces of the battery box out of 1/8″ 6061 aluminum.

Next, I TIG welded all the corners (don’t look too close, it’s been a long time since I have welded aluminum).

A quick test fit in the chassis

Now to make sure the batteries will fit.

Perfect! I’ll drop it off at the powder coaters tomorrow.

I am a little ashamed to admit it, but I decided to use a couple pieces of 80/20 to support my battery box. I thought about welding up a fancy aluminum support system, but this seemed so much easier. Plus, if I ever have to take it out, it is very simple to unbolt.

A sheet of aluminum will go over these supports to form the bottom of the battery box.

I can’t wait to install the new batteries and take it out for a REAL test drive, but I should probably install the seatbelts first…

Yesterday a box arrived on my doorstep. It was the Manzanita Micro PFC-30 charger I ordered 6 weeks ago. Yeah!

Here is the design spec for my BMS:

1. Monitor the voltage and temperature of each of the 50 cells.

2. Report all information over an opto-isolated shared serial bus.

3. Balance the cells during charging

4. Alert the driver if any cell drops below a certain cutoff voltage. (This may get upgraded to actually adjusting the throttle on the motor controller)

5. Provide visual state information via LEDs. I want to be able to look at the cells and determine if they are being charged and/or are balancing. Be sure to use frosted LEDs for high viewing angle.

6. Interface to the car PC for data collection and user interface.

7. Provide an electro-mechanical means of setting the “address” of each board on it’s network (either jumpers or a set of solder-bridgeable pads)

8. Use a communication protocol with error checking. Checksums with retrys will probably be fine.

9. Choose a connector that is standard so we can use manufactured cables (I hate making cables).

10. Keep part counts to a minimal. Remember, each part we add is multiplied by 50.

Here’s the plan so far: build fifty printed circuit boards, one for each cell (and one special board that interfaces to the PC and to the charger) Each board will have a PIC16F616 processor connected to an opto-isolated bus. Mechanically the bus will consist of two RJ45 connectors per board to allow for daisy chained wiring with twisted-pair wiring. There will be two LEDs per board that are software controllable.

I ordered parts to breadboard up a couple of these for testing. My only real unknown is how much current I will need to dissipate during the balancing portion of the charge. We’ll see…

I used the forklift to bring the body into the shop. I kind of hope the car never does this on its own.

I bent up most of the pieces for the roll cage and started welding them together.

I flipped my body over to start laying out the roll cage:

Well it’s less of a roll cage and more of a body mounting system. Either way, it’s going to be made from 1.5″ .120 wall steel tube. I bought an el-cheapo bender from Harbor Freight with the hopes that it would make decent enough bends for my needs. I even tried packing the tube with sand.

Unfortunately, it didn’t work so well.

Which gave me all the more reason to learn how to use the three wheel bender. It can’t do a real tight radius, but after looking at all the bends I actually need, it should do the trick. Here it is bending the piece that goes right above the windshield.

Q: What are the detailed car specs?

Q: How fast will it go?
A: On paper it should get up to 100mph. However, because I still don’t know the final weight, rolling resistance, etc. it is a little tough to predict.

Q: How long will it take to go 0-60mph?
A: Another tough question, but I would guess it will be under 6 seconds considering it has an 11 inch motor, a 2000 amp controller and batteries that can put out 1600amps (if only for a few seconds).

Q: What range will it get?
A: If I can get 80 miles under normal driving conditions, I will be quite pleased.

Q: How much did X cost?
A: Check out the table above.

Q: Are you interested in selling it or building another?
A: You might want to see how it turns out before considering this.

Q: Can I come and help?
A: Sorry, but no.

All the worrying is over. They cleared customs by 2pm today and I had them in my hands by 4pm. Fifty Thundersky LFP160’s fresh off the airplane. They should give me 160v nominal (and float a bit higher). Now I just need that charger I ordered last month.