Paul & Sabrina's cheap DIY 144v motor controller

[Roger Heuckeroth]

The current Dewalt router is like 15,000 + RPM, so we are looking at a different machine to do 1,500 RPM. Maybe just a drill?

The cool thing about using this machine is that once you have the boad fixtured properly, its just a mater of hitting run. It will just take a few minutes to do a board, so it could do it very inexpensively. No chance for the "mother-in-law effect" :^) Precision on this machine it 0.0005", so the boards will come out perfectly.

[nubie]

Originally Posted by MPaulHolmes

Probably the easiest way is to get a pcb that is maybe an inch bigger in the x and y-axis direction, "lock it down" flush to the table somehow or other, and etch the designs and drill the holes, but not going all the way to the edges (since there's an extra inch in x and y direction), flip it over "precisely", and etch the other side, and then do a final rectangular cutout to the correct size. Otherwise it causes problems, and is SUPER annoying to cut all the way to the edge. Maybe if there existed a table that has a suction on it, you wouldn't need the extra length and width.

Don't forget that you are going to need holes in the board anyway, just drill 2-4 of the holes and bolt it to the table with them

Then when you flip it over you just need to center it on the holes, and you can swap in new boards in seconds.

You might also be able to drill a short stack of boards all at once with a press.

(Ever read that book "cheaper by the dozen" where their dad had a job as an efficiency consultant? He is my hero.)

Originally Posted by Roger Heuckeroth

The current Dewalt router is like 15,000 + RPM, so we are looking at a different machine to do 1,500 RPM. Maybe just a drill?

The cool thing about using this machine is that once you have the boad fixtured properly, its just a mater of hitting run. It will just take a few minutes to do a board, so it could do it very inexpensively. No chance for the "mother-in-law effect" :^) Precision on this machine it 0.0005", so the boards will come out perfectly.

See above, fixturing a board that is supposed to have holes in it is no problem.

Say, how big is your CNC table anyway? You could set it up to do a dozen boards at a time you know.

What is the surface of the table? If it is MDF you can use a screw and a nut (maybe a wing nut so you don't need tools), and pencil marks to align the boards.

This is a "Tee nut", if you could put some in your MDF table to accept the PCB holding screws or bolts.

http://www.yourautotrim.com/5t5.html

[Roger Heuckeroth]

The table is 48.5" x 48.5". If you could get large sheets of PC board then it would be easy to do multiple boards in one shot.

Right now the surface of the table is a grating, as it is was orignially just for plasma cutting.

See: PlasmaCAM Cutting Systems

It would be simple to put a 1/2-3/4" piece of MDF on top with fixturing pins.

For now Paul can make the beta units with his mill, but when we move beyond the beta stage this is the way to go. You could do a thousand boards a day with this machine.

Also, this machine could be used to machine the end plates for the enclosures, as was resently mentioned. The Zilla controllers have a metal enclosure and plastic end caps for the busbar and interface connections to poke through.

I'm going to start looking for a lower speed drill type tool to attach to the tool holder.

[MPaulHolmes]

Hey Roger! 48"x48" hehe, ya that'll do. The only reason I do the milling at 1500 rpm is because my machine only goes up to 1500 rpm. It's not a limitation of the drill bits. 1500 isn't the recommended speed. I'm sure 15000 rpm wouldn't hurt them at all. It would probably be better! And allow for faster cutting rates too. When I was using a dremel, it worked fine for etching, and that was like 30000 rpm or something ridiculous.

I got the crystal working. We are officially at 16 MHz! Wow! What is this, 1985? There were lots of weird debugging issues. Man, I love debugging. I never had such weird problems when regular programming. When you start interfacing with the real world, weird stuff happens. For example:

I messed with the fuse bits some like DCB was suggesting. I think that helped. The "optimizations" made the program not work at all. Having too high a sampling rate at 16 MHz made each analog to digital conversion stop half way to it's actual value. It does a successive approximation, and it was stopping too soon, so it wasn't getting all the way down to the correct value. That took a while to figure out. Then the slower, more accurate analog to digital conversion was making me almost run out of time in my interrupt before the next interrupt was going to happen! Makes me mad! But now, each current sample will be when the circuit isn't switching. It will be at the same point of the PWM wave each time. That should allow for much more accurate current readings. However, since I'm re-using the board from the blown up one for my own private version 1.1 to get to the alternative energy fair, things are pretty ugly. I had to remove a few components and add a few others to that junky Radio Shack pcb! Man that was annoying! I didn't have the right resistor values, so I had to solder in random ones that when paralleled with the one that's already there, they are pretty close to what they need to be.

It is tested and working, though. Well, only the low power section. The high power section should be fine.

I sort of want to do Ben's on the 3 ounce pcb, since I have 23 of them, and they are 8"x12", and it's pretty easy to re-enforce the capacitor side by laying down a bit of extra solder and a bit of bare copper wire. My 250 watt soldering gun can melt that crap on there really easily!

[NiHaoMike]

A tip for soldering is to put the board and components on a hot plate just below the maximum storage temperature of the components. It takes less time to heat up the connection when it's already part of the way there.

[MPaulHolmes]

That's a really good idea, NiHao! I'm going to try that eventually.

[scengineer]

I have been a sideline lurker for quite some time on this post and would like to offer my assistance. I work for a capacitor manufacturing company as a mechanical and electrical designer and can help with these components if needed. I have begun work towards my own controller and have a high interest in following along with your design further. My vehicle is a 98 cavalier, a 144 volt system planned.

[MPaulHolmes]

Hey scengineer! You know, I think the most expensive part of the controller is the ripple capacitors. I've been using P11613-ND .

As you can see, they aren't real cheap, but they "have the highest ripple current capability for demanding inverter applications" according to the datasheet. We've been using 24 per controller, which works out to about $50 in capacitors for a single person (rather than from a bulk buy). Getting that down at all would be awesome! Do you guys have "high ripple current capability" capacitors? I'm definitely not married to those capacitors (I'm divorced from them, since they cheated on me with a resistor). Could you check to see if we could get a bulk buy that would beat the digikey ripple capacitors, with the same quality? Thanks for your help! Maybe you could even help us to get the right types of the other capacitors we are using. I'll be posting the updated schematic very soon on our new website (because the dang PDF is too big), and I'll link it to that from here.

[SVOboy]

Paul,

Just so you know, if you ever need free webspace we could gladly rope off a corner of the EcoModder domain for you to put project stuff on outside of the forum framework. Send me a PM if you're interested.



Keep up the good work

[GrahamMc]

Hey where is this website?