A Prototype Electric Car: How I'm Doing

The Plan

The plan is to build a prototype electric car and drive it to work every day. First this will save me over £1000 per year on transport costs. Second, it will give me a real test car to do my electronic development on, with a view to producing a proper design for a practical EV.

The batteries are not practical. The intention is to produce something cheap, and the batteries are secondhand units from eBay. When everything's working, I'll buy better batteries, but I don't have the time, the skill, or the budget to start developing battery technology. So since the batteries are bought off-the-shelf, I don't much care about them, as long as my electronics will handle it and they'll get me to work.

Things to do to get back on the road

  • Fit batteries and electronics, test
  • Build battery boxes
  • Strip all petrol components from the car
  • Develop discharge electronics
  • Acquire motors (buy or build)
  • Acquire a tailshaft alternator/vacuum pump for the brake servo
  • Develop charger electronics
  • Acquire prototype batteries
  • Acquire car to prototype it

    Things that can be done later

  • Develop regeneration dump resistor
  • Develop instrumentation electronics
  • Develop a microcontroller that'll use GCC
  • Develop a battery unit sense module for the chosen unit
  • Develop 12v electric heater
  • Develop auxiliary 12v charger

    Things that aren't planned anymore

  • Fit motors and reduction units to half-shafts & weatherproof
  • Build mechanical reduction units for motors
  • Build the garage to do the work in

    The diary

    Tuesday 25 October 2005: Research

    I've spent most of my spare time ferreting around looking for things that I need for this. Finding transistors that'll deliver the current at the voltage has been one problem, and trying to work out how to test them is another. (Possible solution: rapid-charge a battery from a bank of batteries.)

    I've also been enjoying typically-British rainy weekends, which are a problem as I have no garage.

    One particularly constructive ferret-around at the local vehicle spares recycling centre showed that the way the vacuum servo is powered on diesel Rovers is by a pump attached to the alternator. So a quick-and-dirty solution would be to put a tailshaft on my motor with a pulley on it, and use a belt to power one of these. That gives the auxiliary power for lighting - albeit not very efficiently - and powers the servo to boot.

    Tuesday 4 October 2005: DiY done

    The house just needs decorating (which my wife can do) and some roofing work (which I'm not doing because I don't like heights). And the engine on the Quantum is beyond economic repair. So I'm going to buy motors and fit them, along with 10 of the batteries I have lying around. I'm probably going to get a pair of the Lynch motors, since they are far lighter than the Advanced DC alternative.

    It'll not be a beautiful solution, and it'll still have a manual gearbox, but it'll work, and it'll give me a platform for developing future motors and instrumentation, and some experience at this sort of power electronics.

    Sunday 10 April 2005: quiet?

    There is a reason why things have been so quiet. Certain recent economic changes in this area mean we have a new priority - sell the house and rent one.

    For now, that is taking all my time. But when that is over, believe me, the electric Quantum will happen. With predictions of oil to exceed $100 a barrel, I don't see how we have a choice.

    Sunday 19th September 2004: Motors Again

    The more I think about homebrew motors, the more I'm inclined to have a go. The motor design I am thinking of should make a pretty good generator for charging a 60V battery bank off a VAWT (I'm thinking of 60V three phase at about 333RPM) so even if I build three and they're not up to being car motors, they'll be perfectly up to being 1kW VAWT generators. They'll also be so cheap, and so easy to scale if I want to do juvenile things like multiple rotor designs. And even if they don't work, then I've lost very little money, and I've learned a lot, and I still have some good generators for one day going offline.

    Friday 17th September 2004: Motors

    Somebody on the OtherPower website suggested that I should consider building my own motors. So here is a dream spec for motors, to give me a starting point for feasability.

    Stall torque: 500Nm (1500N@0.33m)
    max RPM: > 1000 (78mph)
    cruise power: 10kW@1000RPM (also 105 rad/sec, 96Nm)
    max voltage: 120v@1000RPM

    I'd also like it to work when fully submerged!

    If we used 12 of those wedge magnets we could build a 150mm radius motor on the hub of our 330mm wheel. (Or rather, at the other end of the half-shaft.) It would need to pull each magnet with 280N (about 30kg of force) and that means 23A.turns per coil. At cruise power we want 96Nm, which is 53N per magnet, or 4.4A.turns. Unfortunately, the motor must draw at least 84A to meet the voltage requirements. Which means that even if the 12 coils in each phase were in parallel, each would need to be less than 2/3 of a turn each (giving a stall current of 414A).

    It'd also need a rotor which could secure magnets doing 1000RPM and pulled by 300N each, and a stator which could have 30 coils each capable of handling 1/2 of 414A, and also of standing a torsion force of 300N. It's not impossible, but it could be interesting, especially as both rotor and stator should be made of non-conducting materials (to minimise eddy current losses). Perspex, maybe?

    Tricky but not impossible. More thought required, I think.

    Tuesday 7th September 2004: Mains Connectors

    I think I can settle on two connectors: the CEEform 32A three-phase connector and the IEC 320 EN 60-320 "kettle" connector. Then we can use three leads: a 13A to "kettle", which will draw 2.2kW; a CeeForm 32A 1p to 3p lead, which will draw 7kW; and a CEEform 32A 3p lead, which will draw about 25kW. The only non-standard lead in this collection is the single-phase to three-phase lead, which has the neutral and earth wired to the normal place, and the live wired to (any) one of the three phases.

    The connector on the garage wall will be a 32A CEEform single phase, connected to a dedicated RCBed spur. That should be able to use the "special" lead to get that 7kW.

    Getting power away from home will use the IEC 10A lead, which plugs in anywhere.

    The three-phase lead is for the future: one day, maybe, there will be charging points available in places like motorway service stations. Stopping for an hour and plugging in while you get something to eat would gain an additional hundred miles range, which would allow almost all journeys in Britain to be attempted. A 40kWh car (70 200Ah lithium batteries - probably a good target) would recharge to 80% in less than 90 minutes.

    Of course, nobody knows what carpark charging points will look like, but given that CEEform is the standard for three-phase receptacles in the UK, it's the best bet. If not, I guess we can offer an alternative lead. It's certainly going to be the right lead for a "fleet" situation, where industrial power is available in the depot.

    Tuesday 31st August 2004: Peak Currents ... and Battery Capacity Again

    Another battery discharge over the weekend showed only 97Ah or so ... but it's turned cold. This may be a foreboding of times to come, when it's winter and the batteries are cold.

    A quick calculation based on the Lynch armature resistance (looks like 64mΩ) says that to keep within a 36kW power limit for each motor, we cannot drive a stall current of more than 750A. As soon as the motors start moving, the maximum current drops to take account of the increase in voltage caused by back EMF. Here are some maximum currents based on 100A modules:-

    Current resistive drop max voltage Max mph
    700A 44.8v 51.4v 4mph
    600A 38.4v 60v 13mph
    500A 32v 72v 25mph
    400A 25.6v 90v 40mph
    300A 19.2v 120v 63mph

    It looks to me that the benefit of more than four or five modules per motor is mainly in terms of the amount of rubber left at the intersection, rather than any significant improvement in acceleration.

    The thought about pseudo-random sequences still looks good: it should also be possible to use it to smooth out the regenerative current, so limiting regenerative current to a power rate (say 1C = 100A = 18kW) rather than a current rate. Of course current control is important to prevent the regenerative braking putting the passengers through the windscreen at slow speeds. But the more constant charging current should improve the regeneration efficiency.

    Friday 27th August 2004: Battery Capacity

    Three things. The first is concerned with pseudo-random sequences for the clocking of the control modules. The advantage of a PRS is that the car hisses when you toe it, rather than whistling. But there is another advantage: if the PRS is generated by a shift-XOR block, different "bits" of the shift can be passed to different modules, which will have the effect of averaging out the battery current. I'd been wondering about ultracapacitors, but this I like better.

    The second thing is that recharging a flat battery using my charger is *slow*. The battery I discharged on Monday is still being recharged - it got to less than 1A in about 36 hours, but it takes a long time to drop to anything like zero. Since it's unlikely to be charged again for quite a while (as I'll be testing others) I want to make sure it's properly charged. I feel a charger is a highish priority.

    The last thing is that I want to look into building a "Z80 ICE", probably something with the Centronics port. A control, two addresses and a data register, read in sequence, should enable a (slow) bus i/o cycle to be performed. Software could decode and emulate the instructions. It would be slow, sure (although an 800MHz PC might well be able to emulate a 6MHz Z80, but the Centronics port won't go at more than about 500k/sec, which at 4 read/writes per bus cycle gives only 125,000 cycles per second. But it would work, and could also be used for things like programming the flash ROM, which would save a lot of work.

    Wednesday 25th August 2004: Battery Capacity

    Last night I connected one fully charged battery to my invertor, which was powering a 60W lightbulb and a timeswitch. The invertor allegedly has a quartz timer inside, so the timeswitch should record the time that the invertor runs for before it cuts off. Fortunately I was up when it cut off this morning: so I could measure the starting and ending currents, and check the accuracy of the timeswitch. The battery actually delivered 100.5Ah in a bit less than 12 hours, to an endpoint of 1.56v/cell. That's pretty close to the manufacturer's figures, especially as the (small) timer inaccuracies and the drop of the current during the discharge probably make this result about 10% uncertain: and also as the wires I'm using are long and not especially fat.

    I'll measure the others over the course of the next few weeks. No quicker than that, of course, because the battery will take at least 24 hours to bulk-charge and probably another 12 to finish. They should be recharged after such a deep discharge, as leaving them discharged would be very bad for them. So I'm unlikely to do more than one every two days.

    But even at this early stage, it looks like the batteries are in good condition.

    Sunday 15th August 2004: Transformers and Wound Components

    Thinking further about the chargers, the following statistics come to mind for standard mains plugs and my batteries:-

    Charge required 12.5v/unit average - 19.5kWh
    Max power 25.5A secondary - 4.8kW (will take 4 hours)
    Mains power @6A - 1.4kW (will take 14 hours)
    Mains power @10A - 2.3kW (will take 8.5 hours)
    Mains power @30A - 6.9kW (will take 3 hours)
    Mains power @32A 3phase - 22kW (will take less than an hour)
    Total cost to recharge = £1.17

    It's worth bearing in mind that I'm only expecting to use between 50% and 70% of this, so these figures could be multiplied by a factor of 2/3. But that's still nearly 10 hours to charge from a 6A supply.

    That 25.5A charger will be one-per-bank, too, which means that an improved battery which has two banks might well end up drawing that full 30A supply - probably the most that can be taken from a domestic environment. The 22kW three-phase supply sounds a bit mad, but a larger 4-bank vehicle will draw most of it, and with a higher voltage than 180v/bank it'll certainly draw more.

    The 6A supply is mainly interesting for small 1.5kVA generators, and the 10A for charging "away from home". The right way to go when we're wiring the garage is to set up a 30A socket on a dedicated spur. Even then, charging still takes four hours plus the battery conditioning time.

    Saturday 14th August 2004: Some Ideas

    Not been well over the last week. :-(

    The design I have in mind is based around current sources, each controlled by the microcontroller. A current source will be used to charge each battery bank, and another current source to provide control for each motor.

    Controller Energy Source Peak Current Current Step Output Voltage Range Peak Power
    Charger Mains Electricity 25.5A 100mA 150v-260v 5.6kW@14.6v/unit
    Motor Control Battery Bank +95.5A
    -32A
    500mA 0-120v 11.5kW
    Accessory Charger Battery Bank 102A 400mA 10.6v-14.6v 1.5kW

    There will be a charger for each battery bank, with the currents individually controlled. The controller will have to regulate the currents to each battery bank, but also limit the total current to keep the mains current within limits.

    Each motor will have a bank of controllers, with the control currents in series and the outputs in parallel. Three will be the minimum to drive the AVT motor successfully, but up to six might be used if we want more acceleration and think the motors can stand the current.

    There will be an accessory charger for each battery bank, with the control currents in parallel through nominal resistors.

    In all cases, the control current will be in the range 0 to 1.02A, in steps of 4mA, In the case of the motor controller, the first 0.764A will control positive current, and the last 0.256A will control regeneration current, using separate sense circuits.

    The control circuits will be current regulated supplies relative to 12v, and will be earthed on the other side via a 10Ω resistor to limit current in event of a fault. In the case of the motor controller, the resistor will be 12Ω for the positive current control, and 39Ω for the regeneration current.

    Sunday 8th August 2004:How to Test Motor Electronics

    Some static tests of motors could be done by dumping the current into another battery or batteries. One discharged battery ought to stand having 100A dumped into it for a few minutes. So 0 to 5 batteries could be used as motors at fixed speeds of 0, 8, 16, 24, 32 or 40mph. That should prove drive and regeneration currents, at least.

    This is another reason to develop 100A/30A modules and use these in parallel.

    Sunday 1st August 2004:Prototype Batteries

    Today I bought 20 secondhand Sonnenschein 104Ah AGM batteries. They weigh 37.5kg each, and the plan is to put 15 of them into the car. That will only get me in on the back roads, but they will have the advantage of being cheap: I paid £12.50 each for them. Since going and returning on the back roads will use about 70% fuel, they should get between 400 and 800 cycles, depending on how the ZDV works out. If they get me to work and back, requiring 14kWh per cycle, then the total cost per day will be 84p for electricity and between 32p and 63p for battery wear. That comes to a total of between 3.3p and 4.2p per mile.

    The plan is to put eight under the bonnet: six across standing on their ends, with two across the back between the brake servo and the control electronics. Then the remaining seven can go in the boot, because if my estimates of weight for engine/gearbox/fueltank/ancilliaries is right, there's only going to be about 220kg of payload available - which means children or luggage. It'll do for prototyping, but later we'll get better batteries with the money we save and turn it into a more practical car.

    Saturday 20th June 2004:Car Acquired

    I've bought a blue Quantum Coupe. I drove it back, and it seems to be significantly slipperier than a Fiesta. I haven't measured it yet, because I'm concentrating on getting it reliable: it has fuel issues and another 31 faults to resolve (and counting).

    On balance, I doubt I'll join the Owner's Club, because I'm not likely to have the range to visit their rallies, and anyway they'd probably get offended if they saw the Monster Garage treatment the poor thing will get.

    Measuring the space under the bonnet is another step - it's smaller than a Fiesta, it seems.


    This page is part of an Open Source Electric Car Project, and is written and maintained by Simon. At this stage these pages are constantly under revision. Thoughts and comments are welcome.