Bill541 wroteThunderhead, since we are getting off the thread subject a bit, perhaps we should start another thread. Fun to discuss for sure.
Indeed it is. We're a little off-topic, but I hope the bystanders don't mind. :-) Anybody that drives is welcome to an opinion, since a lot of this is concerned with the "man machine interface" - that is, how to produce an EV that would pass the "mother test" (one that my mother could drive without needing to know it was an EV).
One idea I would like to share would be to use the vehicles brake system in the feedback loop for regen. One thought was to use the normal brake switch and a pressure sensor in the brake lines. The switch would trigger an A-D conversion of the pressure sensor and when the circuit noticed an increase in pressure, it would apply more regen, if the pressure drops, the regen would also drop.
I've driven cars which used the brake light sensor as a trigger for "engine braking": the old Daf 33 CVT used to use the brake light switch to "change gear" and provide additional braking. There was no analogue sensing, just on or off. My dad used to have one: it drove OK, even without analogue sensors.
Because the brushes will be off-centre anyway, the motors can only generate maybe 1/2 the torque in reverse that they can going forward - which is pretty small potatoes as braking forces go. But even if the motors were completely reversible, it wouldn't be much. Most cars can brake more than twice as hard as they can accelerate: 60-0 times are normally less than 4 seconds, but a car that can do 0-60 in less than 8 is quick, especially for an EV. For this reason, just applying regeneration when the accelerator is released may be enough.
Another possibility is to reproduce the conventional automatic transmission, with the "P-R-N-D-L" type selection. When 'D' is selected, only a small amount of engine braking would be applied when the accelerator was released, but when 'L' is selected, much more would be applied. 'R' is 'L' but with the motors reversed. 'N' has the motors disconnected. The only part that cannot be reproduced electronically is the 'P', which is the 'N' but with the motor gearbox mechanically wedged in some way.
I suspect there's going to be a lot of fooling around with different algorithms for accelerator response, and so I'm planning to use a microprocessor to control motor current. The same sort of argument applies to charging and fuel metering, which are big problems as far as EVs are concerned.
If the batteries are already full, then the regen circuit lays dormant until needed. The vehicles brakes still operate normally but during regen, they would seem to work even better. The goal of the regen circuit would be to keep the brake line pressure constant.
Most battery technologies will not allow rapid charging when the batteries are at anything more than 80% capacity: so for more than 20% of the journey, regeneration is not going to be possible. What is worse, if you have 75% capacity as you crest the summit, it will cease to be available halfway down the hill, which could be ... exciting.
I can't help thinking of that as "the vehicle's brakes operate normally but when the batteries are full they don't work properly", which is what it is going to seem like to my mother. "I was driving down the hill outside the house, and when I tried to brake for the roundabout at the bottom of the hill, I couldn't stop in time. Sorry about the car." It doesn't pass the "mother test". Hence the desire for a dump resistor to take the regeneration current when the batteries are full.
By the way, my work is in electronics engineering as well. Mostly sensor interfacing with microcontrollers and automotive networks. Not much in the high power realm, for the most part under 1000 Watts.
My work is mainly in telecoms and embedded control, but a project spent designing battery charging systems for mass production has taught me a lot about the chemistries and algorithms. It is a racing certainty that someone reading this has a mobile phone whose battery management software was written by me.
Most of my electronics has been small stuff, too, which is another reason to find the idea of using a modular design attractive. I can't buy the phat semiconductors I need to control 600A anyway, certainly not at a reasonable price, so building a +100A-30A traction control unit and then stacking four to six in parallel seems like the way to go. How many to use would depend on the design.
I do want a charger that runs on AC or DC, though: if I ever do end up offline, converting DC to AC just to convert it back again seems silly, especially when it's several kW of power being converted.
My old electrical engineering lecturer at college used to tell me that anything less than 1kW was noise. I think he'd approve of this project...