Designing an electric car
By Don Wires, chief engineer, AMP Holding, Inc.
AMP’s all-electric AMP’d Saturn Sky is a finalist in the Progressive Automotive X-Prize competition
With the advances in electric drive, battery, and motor technology the practicality of the electric car for everyday use may be at our doorstep. Some of us think it is already here. Most of the barriers are associated with cost and acceptance centered in the concern about range. There are technical issues to be sure, as with any new technology, but these are being solved at an ever increasing rate as more companies move into this space.
Cars that we put our families into and drive everyday cannot be science projects; they must be carefully engineered and tested, validated and revalidated.
The approach that we have taken at AMP Electric Vehicles is to build on existing vehicle technology; that is, to electrify a select number of existing ICE platform vehicles that lend themselves to the process. This process results in everyday vehicles that have antilock brakes, airbags, stereos, etc., have been through rigorous testing and validation by the OEM, and have a range of up 150 miles. We used one of our production Saturn Skys as our X-Prize entry, but also deliver the Pontiac Solstice and Chevy Equinox as additional 100% electric-powered offerings.
Our strategy is to work with major tier one automotive suppliers such as Remy, Delphi, etc., and integrate their products with our architecture into the vehicle existing infrastructure. Our IP is the architecture and software.
The technical areas that are of most interest in delivering total electric vehicles are batteries, motor controls, motors, supervisory controls, and ancillary systems. I’ll touch briefly on each of these based on our experience.
Recognizing that fuel cells may play a significant role in the future of electric cars, at the moment most offerings are focused on battery technology. There are any number of battery technologies out there with varying amounts of energy density, volume density, amp hour rating, volatility, etc. In our view at AMP today, LiFePo4 appears to be the safest. It’s not the most energy dense, but it’s sufficient for the vehicles we produce.
The biggest issue with battery life and cell reliability is the BMS (Battery Management System). This whole area needs to be greatly improved. Few systems out there are truly noise immune and protect the cells while charging. The connections are usually so fragile that longevity in a mobile platform over a long term is highly suspect, in my opinion. At any rate, with these systems you have to pay particular attention to grounding, shielding, and robust termination. In our applications we rely on CAN communications to other systems such as the battery charger and supervisory controller to monitor the state of charge and cell health, and to control the charger. Robust network communications is mandatory and requires care in the network layout.
The battery pack should be laid out in a modular fashion such that if cell replacement is required, dismantling of the total pack will not be necessary. Sealing the battery pack from environmental contamination while keeping it within temperature limits will usually require some form of thermal management. The safer the cells inherently, the less you have to design complex safety systems. There is a tradeoff of course, which is less energy density.
Control of variable speed motors has been around for a long time in industrial applications. The trick in automotive applications is to get high power and reliability in a small, robust, cost-effective package. In most industrial applications, a 100KW drive/motor combination costs around $10K. Most are too heavy, too big and too costly. There are emerging systems out there that fit the bill, but with these systems also, getting the control algorithms right for a mobile application has proved to be a challenge. The control strategy varies as a function of the motor technology used. If you take the low speed motor approach with a high gear ratio, acceleration of the vehicle becomes an issue. If you do dual winding, that has its own problems. If you take a low gear ratio and high speed motor operation in a field weakening mode, that presents its own set of problems, especially with low inductance motors.
In most of the applications above, paying close attention to PWM noise and its isolation presents a challenge. You have to ensure that noise does not get on the CAN network, as most modern cars are drive by wire. Also, PWM noise can interfere with the radio and other communication devices in the vehicle. The drives have to be able to support regeneration and be fault tolerant. Integration of the regeneration with the brake involves reasonably complex software to get just the right feel.
The trick for motor technology in electric vehicles is to deliver high power in a small package. That usually means high speed, or if low speed, that usually means heavy. At any rate, high power/small size means heat, the enemy of electric drives. The challenge is to match the motor characteristics with the drive and have enough temperature monitoring and control to not damage the motor. If using a permanent magnet motor, excessive heat can demagnetize the motor, with a drastic reduction in performance. I personally prefer a lower speed motor for electric vehicles as I think they will last longer and are easier on the bearings and generally do not require field weakening control. There are all kinds of motors out there from high speed AC induction, permanent magnet AC, etc. In our applications so far, high torque permanent magnet AC at moderate high speed seems to fit pretty well. Motor cooling, and pressure/flow control must be carefully controlled.
Integration of the existing vehicle controls such as the dash, accelerator pedal, shift control, etc., with the motor controls, battery management, DC/DC converter, and ancillary controls added to convert an ICE vehicle to full electric is not a trivial matter. An incredible amount of real time software managing multiple devices on multiple networks with redundancy and fault tolerance is required to insure safe operation of a vehicle. There are a couple of platforms out there that are up to the task but in order to have any amount of success you have to work closely with these suppliers to meet your requirements. This is an area of great opportunity for the electric vehicle community.
The big challenge is noise immunity and redundancy/fault tolerance. Great care must be taken to keep these systems pure. In most of our applications we use two motors driven through speed reducers to the axles to provide the motive power. This requires an electronic differential algorithm. While not outlined here, suffice it to say this requires high-speed computation of rather sophisticated algorithms in our supervisory controller.
In most of our conversions we have to add air conditioning, power steering and power brakes. Some have electric power steering inherently, but most do not. There are aftermarket items out there but you have to choose wisely as most are not documented well and if you can use the OEM equipment you are better off. Integration of these systems over CAN with the supervisory controller requires careful layout and coordination. The goal is to make the car look and feel just like any other car you may have in the driveway. When you turn the wheel, power steering feels the same, the brakes feel the same, pushing the accelerator feels the same (actually faster in most cases than the ICE version).
Other posts you may be interested in reading: