Paul & Sabrina's cheap DIY 144v motor controller

[Roger Heuckeroth]

Originally Posted by blwncrewchief

I I have the batteries (enough to run up to 156v), a 11.5" 250# series motor, and am planning to put it into a 5000# truck with an electronic automatic transmission. Don't flame me too bad So I think I can contribute to the extreme side of the beta testing.

Actually, you will be doing more good converting a gas guzzling pick-up than if you converted a Toyota Tercel, or some other small car. Essentially you are taking a lot more gas consumption off the road that way.

Over 10,000 miles a pick up truck averaging 12 MPG will consume 833 gallons of gas. A compact car getting 35 MPG will only consume 286 gallons. So, converting the truck reduces the amount of gas used by 547 gallons more than converting the compact.

On the flip side, I hope you are not expecting a lot of range.

[blwncrewchief]

Thanks guys. I don't want to clog up the controller thread with my project and it's issues. And yes I know it has many "issues" I'd be happy to start another thread if you'd like.

[Tango Charlie]

Well, Okay. But you have to promise to post lots of pictures and/or video.

[GrahamMc]

Hey-
Just Wanted to ask if we are modeling in Metric or English?

[MPaulHolmes]

The power board section is in English, since each full crank on the mill is 0.1", so maybe English for the components too??

[RedRod]

Hi all,

I'm following this tread for a long time but recently I'm having much work to be able to contribute.

Now is the time because as a programmer I can help in the programming area and code optimisation.

My 2 cents: as for the speed control your PI is too aggressive and CPU consuming you can make a simple
if(current > MAX_CURRENT || current > desired_current) PWM--;
because the program runs very fast and decrements PWM very fast.

Ex: current = 510A and MAX_CURRENT = 500A
in just 10 cycles the current is corrected and at 16MHZ it while take less then 1ms to do it even a 500A correction while be made in a few ms

as for optimisations you can put the optimisation level to 1 in the AVR studio as a level 1 optimisation makes the code allot smaller and makes no program change.

excuse for any miss spell as I'm not from a English spiking country.

RedRod

[MPaulHolmes]

Thank you for your suggestions, RedRod. Otmar, the guy who makes the Zilla controller, agrees with you about the current usually not changing that fast on an EV motor. He said that in the case of an EV motor, the ONLY time a PI loop is needed is with a LOCKED ROTOR (the car in gear, with the emergency brake on, and then press the gas pedal really hard from 0 rpm!) With hardware current limiting, you don't need to worry about that special case, because the hardware shuts off the current in about 3 us, which is plenty of time, because the mosfets together can handle like 5000 amps (or something ridiculous) for that amount of time.

[MazdaMatt]

I really stand by the suggestion to use a PI rather than fuzzy control algo. Considering the system, my greatest suggestion would be state-space, but we'll get into that later. It will do the same thing as the PI but with less varied response over the speed range.

If we added a speed sensor, we could have the controller automatically calculate the motors characteristics (V = I*R + x*rpm) and work the pwm output accordingly.

Current feedback pwm control using PI is really not that hard to implement or processor intensive... we're just using a really bottom-of-the-barrel processor. It also makes hardware limited completely moot except in disaster fail cases.

[MPaulHolmes]

Calling all people to improve power section layout so as to allow for through-hole soldering of mosfets/diodes!

Right now, with 2 sided PCB, if I were to drill through to the other side, and attach the legs of each mosfet and diode to it's own little island, it would be a bottleneck on the capacitor track side. I'd be eating up real-estate over there! At this point I plan on bending the legs so the expansion and contraction applies stress mostly to the legs rather than the solder joints. And if you want to get out the big guns, slather (I love that word! hahaha!) the solder joints, all the way up to (0.2 inches away) the bus bar! I'll bet you $10000000000000 it won't pop loose! I don't think cooling would be a huge issue, with a huge copper bus bar 0.2 inches away, and the solder connected to 4 or 6 ounce copper.

Note: The "Power Wheel" is patent pending, and attaching the legs with screws has its issues as well.

[MPaulHolmes]

Originally Posted by MazdaMatt

Current feedback pwm control using PI is really not that hard to implement or processor intensive... we're just using a really bottom-of-the-barrel processor. It also makes hardware limited completely moot except in disaster fail cases.

Hey! I'm upgrading the speed to 16.000 MHZ! Wow! I don't want to complain about this processor, but it would be sort of nice to have floating point for the PI loop rather than fixed point math. I only have 16 bit numbers available (simulated in software! Actually it's all 8 bit!), which makes the decimal resolution with fixed point math pretty small, so as to avoid overflow. We're talking +- 1/64 or maybe +- 1/128. Maybe that's OK, but I had a really hard time finding the right P and I with all the noise in the system last time!

It might be much better this time, because a lot of the noise issues are being eliminated! ya!