Precharge

After learning that we do in fact need capacitors, I bought a nice big 230 uF capacitor made especially for this purpose:



The problem of limiting inrush current still remains- the main contactor wouldn't last very long under the 1kA or so that cap would draw as it's connected across the battery. I ruled out having the cap upstream of the contactor, because that would mean the contactor would need to be in the fron of the car, and the main high voltage cables running underneath the car would have no means of disconnection, other than manually pulling the pack splitting connections in the back. So, I set off to design a precharge circuit. The principle of this is that when the key is turned on, it applies 12 volts to a small relay that charges the main cap through a 1k or so resistor. This takes about 1 second. Meanwhile, another RC timing circuit charges up and turns on the main contactor. I used the free circuit modeling tool LTSpice to model the timing, and I think it will work nicely. Here's a screen grab of the circuit tool:


Note that for modeling purposes, the 12v ground (V1) and the negative terminal of the battery (V2) are connected, while in the car, they are isolated. The main contactor is modeled by the winding resistance, R1, combined with a voltage-controlled switch, S1. As C1 charges through R2, Q1 turns on and draws current through R1 and turns the main contactor on. The main capacitor C2 has been charging through R4 and the small relay S2. The Schottky diode D1 is to make contactor turnoff instantaneous, and D2 is to subtract D1's bias voltage from the gate of Q1. R3 is to bleed off C1's voltage while the system is off.

Feel free to copy this design at your own risk!

Snubbed!

Well, it turns out that my assumption that the car failed from overheating was wrong. I dug into the inverter, and discovered that two of the IGBTs had died. The only thing that could really cause this, I reckon, is voltage spikes caused by switching. The problem was that when I removed all the DC bus capacitance (except for some small snubbers near the IGBTs) I didn't count on all the inductance I had added to the system in the form of cabling and the battery pack itself (essentially a big loop of wire). When the IGBTs try to turn off, this inductance causes the bus voltage to rise until something happens, in this case catastrophic failure of the IGBT itself. Pictures of the carnage soon!
Luckily, I was able to source off-the-shelf replacements, which will be arriving along with some bigger snubber caps later this week.

Still learning

I have learned quite a bit about motors since setting out to rewind mine. First, I was wrong about my old motor being a 2 pole machine. It turns out that most 4 pole motors use something called consequent, or phantom poles, where there are 6 coil groups, 2 for each phase, so identical windings to a 2 pole motor. The difference is, the coil groups opposite each other (in the same phase) are wired to produce the same magnetic polarity, say North, in the air gap. This produces two consequent South poles at 90 degrees to the two North poles, making four total poles for each phase.

As far as the rewinding, I have put that off. I have bought a nicer motor-- aluminum frame, inverter duty, and will probably rewind eventually. For now, though, I will convert it from star to delta topology, which is as simple as bringing the central star point out in three leads. This has the effect of converting a 208 volt motor to 120 volts, meaning I can run at 3033 rpm instead of 1750 at full (240) volts. This means I can run the motor at 10 Hp continuous instead of the 5.5 rated. Since peak torque is roughly 2.5 times rated torque for a motor of this type, I should be able to get 25 Hp peak. How long I can keep that up will depend on my cooling scheme, which I haven't really settled.

Winding update

It turns out that I made an error in assuming that there must be an even number of slots per pole. There is a method of winding called lap winding (as opposed to concentric winding) that allows almost any number of slots per pole by winding coils with constant pitch. I will post more as soon as I learn how to do that in this case.

Nameplate mystery

I decided to tackle removing the motor and beginning the process of rewinding by having a look at the old windings. It turned out that the nameplate was wrong--- It's not a 4-pole motor at all. It is a 2-pole motor, meaning the full-load speed was around 3500 rpm, not 1750 at the nameplate advertised. There are 36 slots in the stator, meaning that 6 slots are used for each pole, times 2 poles, times 3 phases. I would like to increase the number of poles to increase the torque. Rewinding for 4 poles is impossible, as 36 slots divided by 3 phases gives 12. 12 divided by 4 poles gives 3 slots per pole. The number of slots per pole have to be even, though. The number of poles also has to be even, and there are only two even numbers that multiply together to get 12 -- 6 and 2, so this motor has to have to have 6 poles with 2 slots per pole or 2 poles with 6 slots per pole. Two slots per pole isn't enough because it results in a poor approximation of a sinusoidally varying magnetic field around the stator, so I guess I'm stuck with 2 poles. This means that at 180 Hz, the motor is spinning about 10000 rpm, which isn't too good for the transmission, I'm sure.

Heat death

Since the last post, I drove the car back and forth to work (2.5 mi each way) a few times, and enjoyed getting the control settings to be a little more driveable. The car accelerates well, but has only a 35mph top speed. The problem seems to be that the torque falls off fairly rapidly as the motor is driven above 60Hz because of field weakening. The way induction motors work, they need a voltage proportional to the frequency to maintain constant torque. As my drive system peaks out at 240 volts or so, the torque falls off, because I am not able to force enough current though the windings. The solution to this is to rewind the motor to run at a lower voltage, say 80 volts. Then, we don't run out of voltage in a 240 volt drive system until we hit 180 Hz. The result is that the motor will have usable torque over a much wider speed range.

Luckily, I was able to speed up the process of rewinding the motor by overheating it to the point the windings shorted. I thought it would be a good idea to drive down to Stanford and back (10 miles) on a hot day, and as I was about to pull into my place, the inverter shut down because it had detected a fault. I knew perfectly well what had happened. I had been meaning to install some sort of temperature instrumentation, as well as a blower for cooling air, but had been putting it off. I figured I would test my luck with my $150 ebay motor...

I've never rewound a motor before, so I am practicing with a smaller motor from my blower, which I also got on ebay, and which I will be installing in the car when I have the motor off. I bought the blower with a 575 volt motor, which I will rewind as a 220 volt motor and run off a smaller inverter. Here's the stator of the small motor:


It's a 2-pole, 1/2 hp motor, with 4 slots per pole, star wound. The windings are 25 AWG, which I am replacing with 20 AWG to do the voltage conversion. There's about 2.5 lbs of copper in this, which works out to about 1000 feet of 20 gauge.

Here's the rotor, along with the end bells and the impeller from the blower. Such elegant machinery.

Kermit comes home

Kermit left his home of the past 15 months at TechShop, and made the 10 mile trip back to my house in Mountain View this evening. It was a little bit nervewracking setting out on the open road for the first time, but we made it without assistance from AAA. I set the maximum current to the motor at 30 amps, and limited the maximum RPM to about 3600, so I definitely didn't set any speed records. It was a trick accelerating from a stop through certain traffic lights before they turned yellow again, and I'm sure I irritated more than one other driver. Accelerating from 0-5 mph was very difficult, with torque not picking up until the car got going a bit. Once rolling, though, it felt peppy, at least compared with the stodgy original acceleration.

I need to fiddle with the voltage, current and frequency parameters a bit until I get acceptable performance. Mostly, I wanted to be assured of getting home without overheating my motor, which was no problem. I could hold my hand on the motor once I got home.

It feels really good to have come this far. Even though the car is by no means done, it is a usable means of transport at this point, and I will be driving it to work and dragging groceries home.

I'll keep updating this space with further news as Kermit rolls on.