Paul & Sabrina's cheap DIY 144v motor controller

[Intrigued]

100 hp per wheel! Wow!!! Time to take on PlasmaBoy's White Zombie!!!

Forget Ben, let ME drive it!!!

...well, don't actually forget Ben. He's too good with the videos and the instructions. Let him drive the wimpy 144 Volt 500 amp version. (Did I just type that??? ) Then, Ben, I'll buy that 5-point harness off of you, LOL!

When you get to the REAL horsepower, Paul, let me know. I'm gonna start saving up for that like yesterday!!!

[MPaulHolmes]

Don't worry, it will be parked, and the e-brake will be on, and a "fuse" will be in place, and by a fuse, I mean a couple strands of 18 gauge wire.

Originally Posted by MazdaMatt

To be clear, the system says "i have 0 current and pedal is at 100, hocus pocus, output is 10", "i have 5 current and pedal is at 100, hocus pocus, output is at 200 because it saw that current went up 5 from output change of 10, so 2:1 meaning we need to shoot for 200 to get 100", "I have 100 current and pedal is at 100, hocus pocus, maintain output of 200"

Obviously we wouldn't be that agressive.

Yes.

[MPaulHolmes]

Originally Posted by Intrigued

...well, don't actually forget Ben. He's too good with the videos and the instructions. Let him drive the wimpy 144 Volt 500 amp version.
When you get to the REAL horsepower, Paul, let me know. I'm gonna start saving up for that like yesterday!!!

hahaha!

Well, maybe higher current at 144 or 156v is doable, but much higher voltage means departing from mosfets and using IGBTs. Maybe with a bigger, longer copper block, like say 12 or 14 inches long, and inch wide, with say 14 or 16 mosfets/diodes bolted to it, and the block has a U shape drilled through it, with water being pumped through it all the time, I bet we could get those mosfets/ diodes up to maybe 70-80 amps each instead of 50 amps each?? It sounds sort of naughty, though. 16*80 is what??? 1280 amps??? We would definitely want to prototype something like that at a smaller scale, like 1 or 2 in parallel, with a little baby liquid cooled cold plate. That sounds awesome! Then we could crank up the current until something exploded or caught on fire, and then we would know how much power we could get, cheaply! ya!

[Intrigued]

Oh, man! I can already tell I live too far away from Seattle WA... I'd take 144V and whatever a mosfet could handle any day, then just have a contactor for the "voltage doubler", LOL!

[MPaulHolmes]

I'm just about ready to test the PI interrupt driven software, with fixed point math! ya! Fake floating point accurate to +/- 0.015!!! Wow! is that crappy or what! But it's enough to get the job done. If the current is at 'x', and throttle is at 'x+1', the PI loop drives it to x+1, and that's all I wanted.

In a PI loop, there are 2 gains (2 constants), called P and I. I'm searching for the perfect 2 numbers between like 0 and 2 or something that will help the current chase the throttle and stay as close to it as possible, without going over much. Here's what I'll do:

I'm going to get a video camera to watch the oscilloscope. Then I'll start with P = 1/16, and I = 0. I'll pulse the motor with a throttle of 0%, 10%, 0%, 10%,... It will do 0% 10% 0% 10% every second. This will give a signal that my $100 crappy oscilloscope can lock on to. It better!!! Or I will whup it like a red-headed foster child (wait, you can't spank them, I mean I'll give my oscilloscope a time-out).

On the oscilloscope I will be watching the output of the current sensor, which is a voltage in the range 2.5v to 3.125v. Oh dear... That means I need to take the box apart again. That's not good, because I glued the ends on it, sealing it once and for all! Oh well, It's just that rubbery caulking stuff. I'll scrape it off and put it on again later.

Anyway, I'm glad I just figured that out. hehe. I didn't know that before writing this. After about 5 seconds of doing the pulses with P = 1/16, I'll bump P up to 2/16, and do the pulses again, watching the current sensor response. Then I'll do it for 3/16, 4/16, etc... Getting it all on tape! Then once I find the highest value of P that doesn't create too much overshoot (or any overshoot???), I'll make P a bit smaller, and that will be my P for ever and ever for this motor! Ben's motor is a little bigger, so I'll have to make my P even smaller so it will run well on his too!

Once I have P, I'll start incrementing I. I'll start at 1/16, keeping P fixed on the special number that I found earlier. Now I'll watch the response on the oscilloscope again. Once it damps really fast with still very small overshoot, I found the beautiful I!!!!!!!! HUrray! I'm so happy! Wait. I haven't done it yet.

Time to go rip that stupid box apart. This seems like a lot of work just to make the throttle work basically like it works right now.
Oh well, at least all the professionals will be able to sleep at night, MAZDA MATT!!!!!! HAHAHAHA!

[MazdaMatt]

Good plan, but just a question... have you ever tested your smallest possible PWM duty to see how much current is generated on a locked motor? As with anything there are two ways to go about it... you can "just try it" or you can gather some system properties and get a design starting point... Its up to you... if you nail it quickly with your tuning method, great, otherwise we'll talk about some systematic approach to quantifying how the system works, then designing a control method from there.

[MetroMPG]

Some of us are here for the open source, affordable wonderfulness of watching this project evolve.

Originally Posted by MPaulHolmes

Or I will whup it like a red-headed foster child (wait, you can't spank them, I mean I'll give my oscilloscope a time-out).

I'm just here for the wise cracks.

[MPaulHolmes]

I've never tested a pwm duty of 1 at locked rotor. That's a good idea. I think this will be OK, because Fran suggested this approach, and Otmar suggested something similar (but his way required that I have a Digital Storage Oscilloscope). Otmar's idea is to charge the capacitors, and then pulse the throttle just with the energy in the capacitors. You can capture about 10ms of data from that. Now *all I need* is like $5000 for the DSO. I can make that much money in like 5 seconds!

[MPaulHolmes]

Originally Posted by MetroMPG

I'm just here for the wise cracks.

So you are here to see King Solomon's Bottom!? hahaha!

[MazdaMatt]

I don't like the idea of just using the charged caps. You instantly start losing system voltage so you can't really tune it properly. For the sake of safety, i think you should really just write a program that outputs PWM Duty 1 for a quarter of a second, then just stops. Measure it all on the scope and see what you've got. If it turns out that the current at a duty of 1 is very high... we'll cross that bridge.