Paul & Sabrina's cheap DIY 144v motor controller - Page 137 - Fuel Economy, Hypermiling, EcoModding News and Forum

eco_generator · 05-23-2009, 01:12 AM

I think we are in the equivalent of... say closed-loop vs. open loop in a gasser. Or maybe between hitting the "power" button on an automatic.

Maybe there is a way to trigger direct (or less laggy) control when needed, I think that's one of the ideas you were having Paul. Or maybe make it user switchable, choose one or the other. Nice smooth everything or jerky movements from jerky pedal. That way they can't complain and just live with whatever choice they make... until they switch it again.

MPaulHolmes · 05-23-2009, 01:29 AM

Hey, Ecogenerator! I think you are right! I'm not 100% convinced that with a DC series motor at low RPM up a hill that it's going to be perfectly smooth without a hair of mushiness (man I like that word). And people also won't be completely happy unless they get immediate response from the throttle. Hmm...

ON a side note, have you tried to super gently go up a hill with a gas car, at maybe 0.1 miles per hour? It's really jerky! I tried it a couple days ago! How come it has to be so dang fancy with an electric car! Man, electric cars are held to a higher standard. I think it's because you feel every wiggle, since it's so dang quiet. In a gas car, you are vibrating, and there's a background white noise, and takeoff is sort of jerky up hill! Try it!

Once that stupid PI loop gets tuned real good, we'll see where that puts us. It would be nice to look at what's happening with an oscilloscope, but this current controller has the old control section, which is a piece of crap, and very noisy. We might have to wait until the new control boards get here to see how the tuning changes when there isn't nearly as much noise in the control part of the circuit.

If we had RPM sensing, then this would be easy! Makes me mad.

eco_generator · 05-23-2009, 01:41 AM

Honestly I would lean towards a more damped control (only slight mushiness). Most have not driven electric cars extensively and would not be able to say whether the pi loop was miscalibrated (barring major hiccups) while just driving the car.

But if it's just a tad too aggressive, then the weenies come out saying how the open source controller is too jerky. Why do you think every single OEM car comes out with built in mush... in the sway bar mounts, engine mounts, transmission mounts, sometimes the body mounts - to appease the weenies that whine about little things they don't like. Most cars would handle much better out of the box with better mounting, but you might feel the engine idle or the transmission shift, horrors! *runs in terror to bed* goodnight! (it's late here with the east coast time zoning)

dcb · 05-23-2009, 09:27 AM

Quote:

Originally Posted by MPaulHolmes

If we had RPM sensing, then this would be easy! Makes me mad.

Can RPM be extrapolated (or a relative indication anyway) from motor voltage and current? So that if you add a motor voltage feedback to an analog pin, you can tell that the rpms are going up as the voltage goes up, for a given duty cycle (or current sense I mean).

MPaulHolmes · 05-23-2009, 11:03 AM

Oh that's good! I looked up something that said EMF_back = c*angular_velocity, where angular velocity is in radians and c is a constant, so I guess just solve for angular frequency, and measure back emf? I asked the EVTech list about it.

MPaulHolmes · 05-23-2009, 11:22 AM

The 32 bit math is HOPELESSLY slow. We're talking like 150 uS to do a single MOD (%), and it's all needing to happen 16000 times per second! what a joke! I'm going to make due with 16 bit.

esoneson · 05-23-2009, 12:13 PM

Quote:

Originally Posted by MPaulHolmes

ON a side note, have you tried to super gently go up a hill with a gas car, at maybe 0.1 miles per hour? It's really jerky! I tried it a couple days ago! How come it has to be so dang fancy with an electric car!

The answer is: because it CAN. ICE can't because of extremely LOW torque at LOW RPMs. While an EV has ALL the torque at LOW RPMs, so it can creep stealthily up the steepest hills smooooooothly.

On another note, you guys are finally coming around to selectable acceleration response modes. Remember Granny to Grand Prix?. Just add another pin read to an external pot and you have a dynamic selectable sliding scale for accelerator response. The knob on the pot can go from 1 to 10 (or 1 to eleven if you are a fan of Spinal Tap), or Granny to Grand Prix.

Way to go Paul!!!!!!!!

Eric

Christ · 05-23-2009, 12:16 PM

Quote:

Originally Posted by esoneson

The answer is: because it CAN. ICE can't because of extremely LOW torque at LOW RPMs. While an EV has ALL the torque at LOW RPMs, so it can creep stealthily up the steepest hills smooooooothly.

On another note, you guys are finally coming around to selectable acceleration response modes. Remember Granny to Grand Prix?. Just add another pin read to an external pot and you have a dynamic selectable sliding scale for accelerator response. The knob on the pot can go from 1 to 10 (or 1 to eleven if you are a fan of Spinal Tap), or Granny to Grand Prix.

Way to go Paul!!!!!!!!

Eric

There are 1 0 kinds of people in this world... those who understand binary, and those who don't.

MPaulHolmes · 05-23-2009, 12:17 PM

or 1 to 16 if you are a fan of the "shifters"...

charlie_fd · 05-23-2009, 02:29 PM

Quote:

Originally Posted by MPaulHolmes

The 32 bit math is HOPELESSLY slow. We're talking like 150 uS to do a single MOD (%), and it's all needing to happen 16000 times per second! what a joke! I'm going to make due with 16 bit.

ATmega168 (4.11$ at digikey) is pin compatible to the atmega8 but you can go to 20MHz and it has a hardware division/multiplier unit (2 cycles). Can this be the key for 32bit math?