Paul & Sabrina's cheap DIY 144v motor controller

[MPaulHolmes]

This is the 600w test of the motor. 60v, 10amp current limit. The maximum continuous current is more like 40 or 50 amps, but I'm just trying to be gentle, and I'll probably end up burning up my ghetto 36v hub motor, which is NOT intended for lots of power.

It's "publishing" on my computer right now. It takes like 5 or 10 minutes. It's really good! I like the end the most, when it's over, because it was such a bad video.

I'm excited about trying it out in the car. It'll be awesome! I'll use those big yellow ring terminals to connected the controller to the motor and the batteries. I need to buy some more of those. I should set the current limit a bit higher, though, because 10 amps won't move a car very far I dont' think.

My wife said I could say more oprah jokes if it made me happy. I've been very appropriate lately. OK, just a couple. What did Oprah eat for dinner tonight? Everything except Ben's motor, on account of it's dangerous cougar growl scared her clean off his land.

The Dyson cyclone vacuum is based on the principle of how oprah eats. Has anyone seen stedman lately?

Video!

[The Atomic Ass]

Paul, your brand of humor is very similar to my own. Also, the end of that video had me in stitches.

[Intrigued]

LOL! I'll have to go back and listen to that last part again just to catch everything!

My Major is Business Administration. I'm attending Drury University in Springfield MO, which is where (Dr.!) Bob Barker of game show fame attended. I already have the Associate's degree, but I'll at least get the Bachelor's.

I'm hoping to go Entrepreneurial with some green ventures, including something with electric cars. I'm looking now at doing a DIY windmill setup with storage batteries, to dump charge whatever I might build for personal use. It isn't ALWAYS windy here, but enough to help some!

[Intrigued]

Hey, isn't that odd trace on decel an attempt by the wheel to regen???

[MPaulHolmes]

Hey! I think you are right, Intrigued, about the wheel generating electricity when you turn the throttle down. I wonder why it's so hard or impossible to re-capture some of the extra energy that the motor is generating when you are slowing down, and the motor's back EMF is greater than the voltage going in. People say it's really hard or impossible with brushed DC motors, but I don't understand why.

[MazdaMatt]

Paul, do you have any teacher buddies that get to work with a CNC mill? You could use your drill-press-sewing-machine method with a CNC mill for spectacular results.

Your project is making me consider selling my race car and doing something silly like build an electric car in place of it... i think it would be cheaper! Hmm... i wonder what sort of lap times i could put in at the sprints with a fully charged battery...

From what I have read the issue with brushed motors and regen comes from brush advancing messing with the efficiency of the regen... but then again, i'm just a forum shmuck that reads the same resources as everyone else.

I must have missed it... what are you doing for current sensing? Are you able to actually sense the current, or are you only able to tell when you've hit the limit. If you can actually sense the current, i would recommend making your pwm controller into a PID current control... if you didn't get that, here's the short version - pedal position would strictly determine current, not pwm duty.

If you don't have current sensing, i recommend it for rev2

[MazdaMatt]

Also... i checked your first post link to the MOSFETS... Ten IRFB4110 mosfets... the datasheet claims 180A continuous? wha?

Anyway, the reason i was looking them up was because i found these guys:

IRC640-ND

they have current sense leads coming off of them. Those micro controllers usually have lots of ADC inputs, so the current sense leads could be attached to those. They could be used for single-mosfet-failure sensing, too. Are these powerful enough? (again, i don't know much about mosfets as compared to other electronics).

So you push the pedal and the ADC reads 0.2v, it then ramps up the pwm until the mosfet feedback reads 0.2v (it is a current output, so use the appropriate resistor to match the throttle) and then it stays steady. When the throttle goes up to 2v, it ramps up the current to match it. This is totally RPM-independant.

[MPaulHolmes]

I don't know anyone with a cnc mill. That would be so awesome, though! Well, Ben Nelson said a friend of his had a modified thing that can etch stuff. I might end up mailing it over there, but I was planning on getting a basic mill next month. I'd have to turn the hand cranks, but at least everything would be straight.

Originally, I wasn't doing current sensing, because I thought I cooked my hall effect current sensor when I tried to solder it. However, when I caused 5 mosfets to fail (all at the same time!), I thought I should try it and see if it still works. It did! It can monitor current up to about 50 amps. It has been working great.

Your thinking on throttle controlling current instead of PWM duty (a truly awesome and original idea that I wouldn't have thought of!) was what Ian of zeva.com.au decided to do too just recently, because his low rpm current was really high, causing jerky starts. He would get full torque at like 10% of throttle. He has a burnout mark in his garage because of it.

He changed it so throttle would determine current. He threw in one extra detail, though. He has it set up so that at 10% throttle, you either get 10% PWM duty, or 10% of MAX_CURRENT, whichever happens first.

For example, let's say you are barely moving. Very low RPM. Maybe you only need 3% PWM duty to get 10% of MAX_CURRENT, but the throttle is at 10%. So, he sets the PWM duty to 3% (which limits current to 10%).

Let's say it's very high RPM now. The throttle is at 10% again. Let's say the current is only at 3% of MAX_CURRENT. Then he sets the PWM duty to 10%. He says his results with that setup make for very smooth acceleration. It feels just like a "petrol car" (crazy australian).

I changed my code so that throttle controls current and PWM duty as described above.

You bring up a very interesting topic! PID control (or PI control), and tuning to get the right constants so the unit step response is just how you want it! I've been reading up on that lately, because the experts on the EVTech list suggested the same as you. I have some questions as to why it would be better than what I'm doing:

[MPaulHolmes]

Sorry, that last message was getting sort of long, so I thought I should start the new topic here:

My thinking is this: (I know I'm wrong, but I want to understand why)

1. Why go to all the trouble of doing a properly tuned PID loop to modify PWM duty based on current. It was suggested to me that while I was tuning the loop using the unit step, I could get current overshoots above MAX_CURRENT that could wreck the mosfets, so that I would need to add a hardware based current control while tuning.

2. Is a good PID loop really faster than what I'm doing right now to find the target PWM duty that gives the correct current. Here's what I'm doing... During each loop, I read the temperature, throttle position, and current. If current is too high (like if at 10% throttle, it's above 10% of MAX_CURRENT), drop PWM duty by 1 (not throttle), read current, if too big, drop PWM duty by 1, etc...

I figure you shouldn't have to do that drop by one very many times, because during each main loop, I only allow the PWM duty to grow by at most 1, regardless of the actual throttle position. It's sort of hard to explain. I should upload my new code and explain each line in more detail.

[MazdaMatt]

Well the short answer is... an engineer would use a control loop, a technician would use your system. The results would be two controllers that "work" and one engineer that boasts about why his works better. The real difference is probably barely noticable. I think that a PI controller is really what you'd want and you would end up with an excellent control over the current and likely a very smooth "feel". The response time of the step response should be incredibly fast, so the user wouldn't notice that part, but you should always be getting as much "oomph" as you intend with the pedal.

I THINK that PI is overshoot free, but that may be PD... its been a year since i worked on a controller-type system and that was with stepper motor position tracking control... totally different.