The next curves show, for the four configurations described previously: Maximum driving torque as a function of the vehicle speed (top); energetic yield (mechanical driving energy, vs. electrical energy consumed in the battery, including energy wasted in the control circuitry) (middle); and the mechanical power (bottom).
These curves show that:
A “green” drive (maximum energetic yield) requires minimum currents in the stacks, therefore small motor torque (low accelerations).
The described sizing is not sufficient to provide energetic yields larger than 75% over all the speed range. There is a "dead zone" close to 50 km/h (31 mph). A finer granularity (more stacks) would be required, or other command schemes (for exemple, one group of nine stacks, leaving three stacks unused).
In the zones of “reasonnable” yield (>75%), driving power at high speed is in the 20 kW (30 hp) range. This sounds correct as compared to existing vehicle performances (this value has not to be compared to the power of the engines, that is in the 100 hp range, but to the “real” driving power of the vehicles (i.e. the power of the engine multiplied by the energy yield of the vehicle that is smaller than 20%)).
When the driving torque is 200 Nm at 140 km/h (87 mph) (four groups of three stacks), the current in each stack is 20A. The current in each wheel is 80A. The battery has to provide 320A (four wheels). This is almost the maximum current that should determine the selection of the battery configuration.
Losses are due to joule effect in the switches. In these simulations, we used a series resistance of 0.2 Ω per switch (very conservative approach including parasitics), that would generate a 4V voltage drop for 20A current, and a power dissipation of 80W per switch.
Incidentally, in reference to earlier comments related to unsprung mass, see the viewpoints from Protean Electric:
Refer to patent number: 7595574
When I investigated this in the past, the technology appeared to be quite sound and a great way to eliminate mechanical gear boxes. There are, of course, concerns over the cost of copper and magnets.
Also see the LaunchPoint motor (which uses Litz wire instead of PCB, but I think PCB is a good approach and there are others employing PCBs for stators and even rotors).
The subtlty is the rdson will be bigger for the worst case (parallel) requirement but will be an extra efficiency loss when in series times # in series.
Butt this idea has a bank account of efficiency and cost savings to draw down from in removing weight and mechanical transmission.
Just can't give up more than ~75% of these advsntages to be viable.
I think it has legs.
With brushless motors, you have to reverse the current as the rotor turns. The switch from series to parallel is done at the same time as this current inversion.
But you're right. As this design requires many transistors in series, each Rdson has to be as small as possible.
Because when you switch in and out transistors from series to // the voltage rating and the current ratings must be able to handle the worst case. So when in // voltage worst case...this sets the Rdson (higher voltage higher rdson).
Whne these are put in series then you will have excesive voltage drop compared to a series only design.When in sries the current is max (per xstr)and the die area must be larger.
But it may still be better considering all the benifits. Lots of semiconductor cost, again may be offset.
I am not convinced that this requires more wires. The torque and the power of the motor are proportional to the length of the wires. So, for the same power, you need same length. The difference is that you have to split the same wire length into different pieces, in order to connect them in series or in parallel.
Even if this design did require a greater bandwidth, the frequency is low enough that almost any transistor of adequate ratings can handle the job. But the big advantage is that it is still cheaper than actual gears, and the switching function is both cheaper than gears and not requiring mechanical shifting. So really it is quite elegant. It needs a few more wires, but that is certainly cheaper than gears and mechanical shifting, which would still require wires and power transistors, plus, the powered shifting mechanism would use much more power.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.