Friday, April 9, 2010

Year One in Review

When I set out on this project, I knew it probably wouldn't ever really be finished.  Well, here I am at the end of the first year so I wanted to review the progress and plans to date.

Energy Storage:  The LiFePO4 batteries have lived up to their billing.  They are easy to take care of and hold a lot of energy per unit weight.  My original plan called for four independent batteries supplying four independent drive systems.  So far, I have only had two drive systems so I have been tying all the packs together in parallel but that is soon to change.

What I have been working on:  I purchased Volt Blochers for all my cells and they have worked exactly like they are supposed to.  I attached the circuits to the tops of the batteries but am now questioning that decision.  There have been a couple of instances of Lion batteries catching fire during charging and the culprit may have been a disappative balancing circuit overheating, causing the battery to catch fire.  The way the Volt Blochers work is to start shunting current to a large 10 W resistor when the battery gets up to charge.  This allows the rest of the cells in the pack to continue charging without overcharging the cell that is already full.  However, those resistors can get pretty hot (I have recorded up to 130 F) so I decided to move all the Volt Blochers off the top of the cells.  I'm going to mount them all on plexiglass and then use a fan to cool them while charging.
Along with this, I have been working on my BMS.  I'm not doing anything too fancy but I found the LTC6802-1 which is designed to monitor battery stacks of up to 1000V.  The IC is designed so that you can hook them up in series, with each one monitoring 12 batteries, and communicating to the next one up the stack using SPI.  This gets aroung the issue of how to measure the voltage of an individual cell without having to worry about high common mode voltages.  I am using a Netburner MOD5270 to communicate via SPI with the LTC6802-1. Then I'll upload the data to my TS-TPC-7390 to display it.  Again, not very fancy but fun to work on.  Here is the circuit board I've designed to mount my BMS on:

Drive system:  As I posted in my last update, I have built up a modified rear transmission that is going to give me some variability in drive ratios.  I found my direct drive set up, while efficient, was too limiting when it came to real world driving.  I have yet to install it in the car because it is still at Hardcore Customs having the front end modified to accept the hub motors I found.  Here are some pictures of what they have so far:

Control system:  It seems that the first thing anyone says about my project after describing it is "If you don't have a differential, how are you going to control the the speed of the wheels when you turn?"  My standard answer is, "You don't have to worry about it.  Each wheel is not tied to any other and therefore they will spin at whatever rate they need to."  This is correct as far as it goes; the car drives fine without any direct control of the speed of the motors.  However, when going around a turn, the wheel on the inside is supplying most if not all of the power.  This limits me to half power while turning so my control strategy:

Each of the motor controllers is controlled by the throttle pedal connected to a potentiometer.  Depending on how far the pedal is depressed, a resistance is sent to the controller and that determine how much PWM will be supplied to the motor.  I thought about using a digital potentiometer controlled by a micro computer, kind of a fly by wire system but I rejected it because I don't like having to rely on a computer to make the car go (Ironic, I know).  Also, there are all sorts of failure modes that I can think of with that set up and the subsequent control system would be more trouble than its worth.  So instead, I'm going to try something inspired by the helicopter (HH-65) I used to fly.  In the mechanical control runs for the cyclic (and collective) were servo motors that could vary the length of the push-pull tubes to the rotor head.  You could fly the helicopter without these servos turned on but with them, it was easier to fly.
So, I'm going to put servos in between the throttle pedal and the potentiometers.  Then by measuring the current (power) going to each motor, I will be able to change the throttle to an individual motor to get a little better performance.  I will still be able to drive the car without this system turned on but with it, the car will be easier to drive.

Instrument Cluster:  When I removed the transmission and engine, I removed the inputs to the speedo and tach.  I'm going to replace those inputs with another micro controller (number 6...  I told you there was going to be a lot of computers in this car).  I'm going to use a Netburner MOD5213 for this one.  It is going to get RPM inputs from the motors and wheels then use a PWM signal to drive the 2 pulse/rev input to the tach.  Now this is where it gets silly....  To drive the speedo, I'm going to use a 12V servo motor with encoder to spin the same input shaft that used to be driven by the transmission.  I'll control the RPM of the servo motor using the same MOD5213 to match the speed of the rear wheels, thus giving me back my speedo.

It has been an interesting year and I look forward to making more progress in the near future.


  1. Hi, what pitfalls did you find using the CVT cone drive as seen on the your u tube video, it looks like it works great giving the extra ratio required when replacing a transmisson
    Regards Gary.

  2. Gary,
    The biggest pitfall was that it didn't work :-) The problem was that I was trying to do direct drive so I needed the motor to wheel connection to spin in both directions. Unfortunately, the belt drive CVT only works in one direction.

    Basically, the CVT has two components: the driving pulley and the driven pulley. The driving pulley (the one connected to the motor) uses centrifugal force to press the sides of the pulley closer together as it spins faster. So, it actually doesn't care what direction it is spinning.

    Unfortunately, the driven pulley uses springs and a ramp to keep the sides of the pulley together. In the correct direction, the torsion of the spring keeps a plastic bearing pressed against a ramp which is sloped against the direction of rotation. So both the torsion of the spring and the ramp keep the face of the pulley from rotating which keeps the faces of the pulley close together.

    In the other direction, it is only the spring tension keeping the face from rotation. As soon as torque is applied, the face of the pulley rotates allowing the faces of the pulley to separate.

    Kinda hard to describe in words but that is basically the problem.

    The solution is to not use springs and ramps to control the amount of separation of the driven pulley. There is actually a youtube of a guy who built a electric actuator to control the width of the driven pulley but that was beyond what I was willing to do.

    Anyway, let me know if you want me to clarify anything.

  3. I wounder if a mechanical arm, maybe an electro-mechanical arm (because you wont be in the trunk often enough to activate the mechanical one :)would be an easy solution. A simple arm with a fork that pushes the the movable pulley side against the belt during reversing.