Paul & Sabrina's cheap DIY 144v motor controller

[MazdaMatt]

Paul,

generally i try to design PI control in excel... but you may have better luck with a simple circuit to test it. Doing it "quickly" in excel is best for linear systems - yours is not.

Make a circuit that has a button and does a sudden switch from 0v to 1v. Have this output go to your throttle input. Tie your scope "trigger" lead to the button and the other lead to the current sensor. Push the button, see what happens.

If you wanted to get fancy about it, make your code examine itself. Instead of reading the throttle line, read a variable. On startup, make a 1 second timer, when it expires, set your variable to the equivalent of half throttle or full throttle and set a flag. When the PI loop sees the flag is set, have it start storing its output and input values in an array. Copy that array to the computer and you can examine your system response. With that, you could even tune your P and I values in excel without trial an error.

too much? Sorry, just brainstorming solutions...

[order99]

A little something to go with all these Open Source Controller and Charger projects popping up:

Desulfator for 12V Car Batteries, in an Altoids Tin

Not mine (sadly, i'm just not that bright) but the creator has declared his Altoids-tin BMS as Open Source. What next, an Open Source DIY EV-capable motor? Not that I wouldn't give it a whirl...now where's that fire extinguisher?

[motor_control]

Originally Posted by order99

What next, an Open Source DIY EV-capable motor?

I'm working around it - see page 18.

[MPaulHolmes]

Originally Posted by MazdaMatt

Paul,

generally i try to design PI control in excel... but you may have better luck with a simple circuit to test it. Doing it "quickly" in excel is best for linear systems - yours is not.

Make a circuit that has a button and does a sudden switch from 0v to 1v. Have this output go to your throttle input. Tie your scope "trigger" lead to the button and the other lead to the current sensor. Push the button, see what happens.

If you wanted to get fancy about it, make your code examine itself. Instead of reading the throttle line, read a variable. On startup, make a 1 second timer, when it expires, set your variable to the equivalent of half throttle or full throttle and set a flag. When the PI loop sees the flag is set, have it start storing its output and input values in an array. Copy that array to the computer and you can examine your system response. With that, you could even tune your P and I values in excel without trial an error.

too much? Sorry, just brainstorming solutions...

These are wonderful ideas! Not too much! Thank you! Hmm... Store results in an array, and copy it to the computer. I love that idea. I've never tried that. Is that what the EEProm is for? If I stored the array in standard memory, is it lost when the power is cut? I bet it is. I need to figure out how to use the PROM.

-Paul

[MazdaMatt]

Paul, you don't really need to do that if you can set up some serial communication to the board. It is generally very easy to do that. At work we generally use a USB to TTL converter from FTDI Chip Home Page. The TTL end of it connects directly to the cpu.

So here's the plan:
1 - set up a 1 second timer, a large array (as large as you can with that chip) and a fake_throttle variable set to 0.
2 - when the 1 second timer expires, set the fake_throttle = whatever 1V input would be read by the ADC.
3 - when the fake_throttle value != 0, put the control system output into the next spot in the array.
4 - when you have reached the end of the array, shut down your pwm and begin a serial communication routine that will spit out the whole array from the serial port.
5 - on your PC use some kind of port monitor, these are a good way of getting a stream of data in a copy/paste-able form, then drop it into excel for analysis.

For this data to be worthwhile, you need to be sure that your PI control is happening at a known interval, not just in the main loop. It is completely impossible to tune a PI controller that executes in the main loop because timing will always be different.

I hate to say that i'm too busy, but i am... i am really looking forward to punching some code on my own motor control - there is a set of "scrap" boards here at work that are basically just a coldfire processor with a bunch of IO. (really, i want to use them to make an AC motor control, but the PI could be transferable)

[Tango Charlie]

Originally Posted by MPaulHolmes

It's like 625 pounds! Why the crap is a dang glorified drill press 625 pounds? Whatever. It's all good. My engine hoist can hoist 2000 pounds. So, if Oprah... Nevermind.

D'oh!
Mosfets, PI loops, triggers, eeprom...I dunno what the heck you guys are talking about, but Oprah jokes I GET!
NEVER pull up short on an Oprah joke!

[MPaulHolmes]

Originally Posted by Tango Charlie

D'oh!
Mosfets, PI loops, triggers, eeprom...I dunno what the heck you guys are talking about, but Oprah jokes I GET!
NEVER pull up short on an Oprah joke!

I place full blame on my wife! haha! It's true though. She told me I had to stop picking on Oprah's... um... "black hole like" gravitational field.

MazdaMatt: I can do that serial interface. I need a dang laptop though. Due to living conditions, the computer and the motor controller test area are separated by a rather significant distance. I'll figure out something.

As for constant time for PI loop tuning, since I have the hardware current limiting built in (well, once the dang mail gets here I will), I won't need the current sampling to be very fast. I think I'll execute the PI loop from inside a timer interrupt that will be run about every 1 ms. I think that will work out very nicely. It will simplify the code too. I won't need a slowChangingThrottle, and destinationThrottle anymore.

[MazdaMatt]

Never underestimate the power of your processor. Make generous use of the timer interrupts, and feel free to run them FAST as long as you aren't doing long arduous calculations. Main loop doesn't make things run faster than if they were in a timer interrupt.

If you have an easy way (pre-written driver?) to burn values into eprom, you could do the exact same method as above, but instead of pipe the array out the serial port, you could just copy it to eprom, bring it to the computer and carry on from there.

[order99]

Originally Posted by motor_control

I'm working around it - see page 18.

Please keep us informed...I can't wait to see your results!

Lessee-Open Source Controller, Charger, Desulfator, Motor, DIY Headlights, hmmm...now we just need a few Open Source DIY Chassis and we could give away plans for an Open Source EV on the Net!

So, can suspension, wiring bundles and turn signals be copyrighted? Or can anyone make them if those parts became unavailable?

[motor_control]

AVR: Very Hi Voltage Programming
http://www.youtube.com/watch?gl=RU&hl=ru&v=EHOxc-FvaUs

Originally Posted by order99

Please keep us informed...I can't wait to see your results!

ÎÊ !