Comment by phkahler | Hacker News Reader

Comment by phkahler | Hacker News Reader

Parent Full thread

That's a very good question. I have an EV motor control and calibration background so lots to say here. First off, peak efficiency tends to occur at mid to high speed and mid to high torque, in other words high power. Most EVs spend very little time operating anywhere near their peak power. It may take as little as 4kW to drive down the highway at constant speed, but a motor may be capable of 100kW or much more.

Efficiency at 500 to 1000 RPM is important, and reluctance machines tend to be very poor at those speeds and low torques. So maybe he increased efficiency from 30 percent to 61 percent, which would be really good. Or maybe he increased efficiency from 50 percent to 65.5 percent (65.5/50 = 1.31 but I hate looking at percent change of a percentage). What he did not do is increase the max efficiency significantly at an already efficient operating point where most motors are already over 90 percent, and that's OK because like I said we don't operate up there very often.

BTW, IMHO the best way to characterize electric machines like this is not to look at their efficiency, but to look at their losses. There's a really nice way to plot losses but it's a bit hard to explain in a comment. He also doesn't seem to have a dyno which makes taking data and testing a lot easier.

Anyway, he did a great job and will undoubtedly continue doing so!

> Efficiency at 500 to 1000 RPM is important

Why is this? Thinking out loud: can't you gear your motor so that it's most efficient at cruising speed for the vehicle? But the problem is that the car still needs to be able to accelerate from zero and some gearings will make that impossible, so having an efficient band at 500—1000 RPM gives you the best chance of being able to accelerate from 0–500 RPM?

I mentioned that efficiency isn't a good measure. We need low losses at all operating point, and the "cruise" points in particular. The problem with measuring efficiency is that at zero speed (starting from a stop) or at zero torque (coasting) your output power is zero (torque times speed) and since there will be at least some input energy that means output/input = 0. Because of this, all motor efficiency maps show poor performance along both axis and a big region of "good" at higher power. There is a steep gradient of efficiency between those extremes and small changes in losses can really shift the "efficiency" a lot.

I like to plot Torque on the X axis and losses (in watts or kW) on the Y axis. Make a series of curves at different speeds (each curve is at a constant speed). These curves intersect the Y axis at zero torque, so friction can easily be seen on the Y axis. The curves have a parabolic shape (I^2 copper loss) forming a sort of butterfly. If the wings spread out at the tips (curves separate more at high torque) there may be magnetic issues. These maps are really nice, are easy to compare, and never mention efficiency at all ;-)

If the design is sufficiently low cost, and since it it wont have drag when free wheeling, multiple motors could be ganged to provide different peak/torque efficiencies. High torque to get off the line and a cruising motor to travel at speed. Given the power to weight densities of modern motors, having multiple compact motors will provide zero weight burden in a car.

>He also doesn't seem to have a dyno which makes taking data and testing a lot easier.

how do you test without a dyno? and isn't it easy to set up a electrical generator tied to a high wattage variable resistor to do it?

We often want the a dyno to maintain a given speed and do torque control with or motor under test. These roles can be reversed too.

When working on water pumps I didn't need a dyno, as the fluid provided sufficient load. It was a bit tricky to make before/after comparisons though be cause getting the exact same operating conditions (torque and speed) was not simple.

Reply on news.ycombinator.com