Paul & Sabrina's cheap DIY 144v motor controller

[MPaulHolmes]

10:
If Oprah dressed in a Bikini.
That failure mode would be impossible to happen, since there does not exist a bikini that Oprah could fit into. Thankfully, we don't need to test that... If it were to happen, it would represent a catastrophic failure of my eyesight.

1.
Connects B+ and B- in reverse.
I've done this! It's bad! hahaha! It makes a basically short circuit because the mosfets act as a diode from drain to source (which is now + to -! aahhh!!!)

If you were to close the main contactor with B+ and B- reversed, almost instantly the mosfets' legs would blow off! Don't do that! Hmm... Would there be an easy way to prevent that? Maybe a big dang diode somehow? I don't know. Maybe NiHaoMike is right in that case?

[NiHaoMike]

The easy fix is to integrate the precharge circuit into the controller. Add a relay to control the precharge resistor (maybe use a ceramic heater and have it serve dual purpose as a cabin heater by letting the relay reconnect the heater to a small MOSFET PWM control?) and an optoisolator to monitor precharge progress. If the input polarity is reversed, the precharge will never complete and the current will be low enough that damage is unlikely.

Or a simple solution is to install an alarm that sounds if the input polarity is reversed.

[MPaulHolmes]

Those are really good ideas Nihaomike!!! I think we should do one of them! Thank you! I like the idea of adding the precharge resistor to the control section.

[slurryguy]

Originally Posted by MPaulHolmes

1.
Connects B+ and B- in reverse.
I've done this! It's bad! hahaha! It makes a basically short circuit because the mosfets act as a diode from drain to source (which is now + to -! aahhh!!!)

If you were to close the main contactor with B+ and B- reversed, almost instantly the mosfets' legs would blow off! Don't do that! Hmm... Would there be an easy way to prevent that? Maybe a big dang diode somehow? I don't know. Maybe NiHaoMike is right in that case?

Actually... YES.

There is a way to prevent that.

When the first guinea pigs start assembling and installing their own controllers we document step by step what we are doing and essentially write the DIY how-to step by step instructions for the wiki. We include a bunch of tests after important steps to double check.

This kind of catastrophic error is the kind of thing that I was thinking about. Everyone can help prevent this problem by documenting careful steps and double checking testing procedures to make absolutely, positively, rock solid, extreme confident sure that it doesn't happen.

This is one item that should get checked at least a half a dozen times as someone is installing. The cost of checking and testing extra times is very small. The cost of this mistake is extremely large.

Visually inspecting the wires to be "positive that postitive going to controller positive" is the obvious first check.

I can imagine going so far as to hook up every connection to the controller EXCEPT the traction pack. Then have someone turn on the car "ignition" so the controller will close the main contacter. Then put a volt meter on the cables and confirm that the traction pack voltage polarity is exactly what it is supposed to be. Then turn the car back off before hooking up the final connections.

Anal retentive? Perhaps. In this case it's probably well worth the extra testing steps considering the cost of a screwup.

Hey... I admit that I make plenty of screwups. That's why I work so hard to avoid the biggest and most expensive screwups by asking silly namby pamby questions.

Are you feeling my positive vibes?

[dcb]

As I fancy myself and most actuators as non-idiots, I have a hard time appreciating making everyone pay for the added complexity of speculative idiot-proofing.

Get it working, hook it up right, see where it breaks, no need to speculate about how someone could possibly screw it up too much, they will tell you in short order

[edown]

To test my wiring, I just temporarily put in a 10 amp fuse in and manually close the contactor.. If nothing happens, then all's well..

Evan

[NiHaoMike]

Another useful tip is to connect a light bulb across the contactor right after wiring. If it does not light up at all, then there's a bad connection, the bulb is burned out, or the controller capacitors are already charged. If it lights up and then fades off, then remove the bulb and proceed to additional testing. If it stays on, there's a short circuit somewhere.

[squiggles]

"The Wombat has turned into a POSSUM!!!!"

Why such an Australian flavour? I know this is a multinational effort but calling anything a wombat is a bit rough. Actually I knew a girl we called Wombat but that is a story for a different forum...

[ev59rag]

Another useful way of doing things is make a comprehensive checklist. 1) do this, 2) check this, 3) do this... etc etc...... Germans have 40 point checklist just to install wheel/tires for their bmw's.... This way, the user dont have to think what to do next when installing/servicing these expensive devices. They are all laid out for him/her.

Make it stupid/fool proof checklist...

2 cents here.

[squiggles]

Dear team, just like to play devils advocate for a moment.
Been having a scan of the designs and have a few simple questions/comments.
1. I note that there are opto isolators on the pcb with circuit tracks passing between the row of pins. This effectively reduces the isolation level. It is not necessarily a problem if the isolation level required is low. It should be considered though.

2. I am a little nervous about the NAND gate circuit, can somebody please describe its purpose. Again it may be absolutely fine and could be me causing problems for no good reason.

3. In the software posted on the WIKI I see several while(); statements that look like potential hang points to me. Look at each one and ask is it possible that the exit condition does not occur. For instance in ReadCurrent() are we 100% positive the conversion will complete 100% of the time. What happens if it does not, will the WDT rescue us? Is that the best outcome?

4. Purely for my benefit, where is the TIMER1 initialised?

5. For future versions, maybe it would be better to build a Delay_uSec(int) routine rather than the simple Delay(int), only because tells us something and if it uses #defines as the count numbers is is easily adjusted for different speed processors.