Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[tesla500]

Hi JR,

With the fully isolated control-on-the-12V-side approach, I'm hoping noise won't be too bad there. The main area I'm worried about are the isolated supplies for the high side gate drives, those are slewing up and down fast, the traces from the power supply to the drivers could work as antennas radiating noise into the enclosure.

I think the base plate will be thick enough that cooling channels could be milled into it for water cooling. The thermal mass will be needed so handle the high dissipation during short high power bursts. Having the heatsink on the top would be very useful for convection cooling, but I don't believe it would be of too much help with forced air or water cooling. Heat doesn't rise in solids.

I'm using Altium Designer for the layout. I've uploaded a PDF of the (very incomplete) schematics to:

ww w3.telus.net/tesla/Incomplete_Inverter_Schematics.pdf
(remove the space in the 'w's)

Regards,
David

[sawickm]

Originally Posted by tesla500

www.telus.net/tesla/Incomplete%20Inverter%20Schematics.pdf ,David

Hi David,

Thanks for sharing your AC Controller project with this fourm.

The link that you supplied is not working (even with the www corrected) ???

I think that the % (spaces) used in your Schematic link might be the problem. If you replace them with an "_" it might work better.

- Mark

[JRoque]

Originally Posted by tesla500

Heat doesn't rise in solids.

Good point.

Originally Posted by tesla500

I'm using Altium Designer for the layout.

Excellent, so am I. I'm still on the Winter 09 version since upgrades either mess up my custom or the factory devices. I've been burned twice with wrong N2222 pinouts. I haven't paid for 2011 upgrades yet, hopefully they won't drop me when I renew.

ww w.telus.net/tesla/Incomplete%20Inverter%20Schematics.pdf
(remove the space in the 'w's)

I think you meant:
ww w3.telus.net/tesla/Incomplete%20Inverter%20Schematics.pdf

I'll wait until you're ready to annotate the parts before diving in there too much. You're still working on it but are you planing on a -15V rail to turn off the IGBTs? If you haven't decided the source analog devices , I'd like to recommend Linear Tech as they have excellent documentation and parts can be purchased in small quantities directly from their websites.

JR

[tesla500]

Thanks for catching the URL error, the correct URL is now:
ww w3.telus.net/tesla/Incomplete_Inverter_Schematics.pdf
(remove space)

I do have -15V, the gate drive page has an isolating DC-DC coverter that produces + and - voltage for each gate driver.

I've learned to be very leery of the devices included with Altium. They're probably made by some poor sap sitting there entering pinouts all day, so they're bound to have errors. I always check them before using them, and copy any that I use into my own library so they don't get changed by upgrades.

Regards,
David

[JRoque]

Hi. Yes, I see it now, thanks.

I also see you're taking no chances and truly building everything from the ground up. Please tell us that multitap transformer is off-the-shelf and you have no plans on winding that yourself :-)

I have no experience with DSPic and generally prefer Atmel stuff over Microchip's, though it's a necessary evil in this application. Are you also planing on releasing your project code open source?

Have you considered a step up converter to run the inverter at a higher voltage? I've seen some designs that claim as much as 97% efficiency in the up conversion. That would open up a world of possibilities by letting us run a smaller pack but use industrial motors designed for higher voltage. A >300V bus would make the machines happy. Hopefully the higher voltage bus increases efficiency enough to overcome the ~3% up-conversion loss.

Paul & others, sorry for hogging your thread, this is very exciting stuff!

JR

[Janis]

Hi all, some things from me:
1. system voltage; low voltage is synonimous to low efficiency, high voltage - high efficiency, and you can bouth safely die from 144VDC and 700VDC as either from gas evapours from your regular combustion engine, when lighted up in perfect air to fuel ratio; it all depends on how safely you make it. So please don't go low voltage, better higher as much as you can get. Higher voltage means also less copper, less weight, less amps to waste for hauling higher weight. IGBT's costs are AMPS not voltage, so if higher voltage it's less amps, less amps less costs. And deffinettly IGBT rating is 1/2 amps RMS and 1/2 volts RMS due to mistic spikes that doubles DC voltage so if DC volts are 600 IGBT must be 1200V (don't ask me why, I am not Tesla and don't understand electricity).
2. cooling/heating; controller and PM motor deffinettly should have liquid cooling. when rettrofitting car with electric engine you have from combustion engine radiator and cabin heating liquid type, so all the waste heat from IGBT's and PM motor you can get in cabin for heating it. Also working with industrial PM motors for machines dramatically falls torque momentum, even more than 2/3ds from nominal, if temperature gets over 50°C on body of motor and when cooled to 20°C it get's back full torque.
3. speed feedback. don't go low cost optical, I have seen what happens to motors (PM and AC closed loop) when they lose signals from optical encoders, they just rev up to full speed at full power or start to jog fwd rew at full power and this is no good for EV...... the best solution what I see for EV is resolver (other way relaiable optical are out of costs for diy they come in pricing 500usd and up) we should look up for some ic's like this: Synchro/Resolver to Digital Converters | Analog to Digital Converters | Analog Devices they are costly but weakest link for security is speed feedback, if you lose IGBT's controller just dies, if you lose controller evrything stops, but if you lose speed feedback controller can make up very a lot different fancy things which can lead to fatal endings to the story......
4. please have fun before you do something and while you are doing something, because otherway evrything will be crap what you are doing and pushed out of second end!!!

http://www.youtube.com/watch?v=baaX4-1_df0

And come to Latvia to have some beers for fun. And think only good thoughts, because they make you good and celebrate life for full speed and rews etc. feel free

[MPaulHolmes]

What a beautiful song!!! Very helpful post too!

I agree that the resolver is a much more rugged solution. High voltage is also better, yes! But brand new IGBTs can be expensive. Ebay has a bunch, but new they are expensive.

Oh, I just tested the mosfet based 3 phase controller yesterday and it worked! hurray! I thought it would but it was nice to see. I believe it can do around 40kW peak with an AC motor, and closer to 50kW peak with a PMSM motor, but that needs to be tested. I'll be re-doing the driver section to make it MUCH cheaper and overall better. ya! The voltage will only be 144v because of the mosfets.

The cost of it also just dropped by almost $100 because of Fran's suggestion about the DC-DC. Because of that, I really want to do a good IGBT version now! Janis is right! Higher voltage is better.

[Janis]

If we want to see EV on fire Higher Voltage is perfect He He...

[Janis]

Yes some frends from Ogre town are going EV to Dackar: OSC Race Car Engineering — OSCar, Rally Raid & Racing

[JRoque]

Hi.

Yes, higher voltage is better... for the controller and that's about it. It's not so good for when you need to attach your peripherals to the pack like your DC-DC since there are no cheap ones for higher voltages. Charging your pack can only be done with a boost converter and it's complexity, expense and losses. Your pack will be considerably heavier than a low voltage pack when using prismatic cells like we do. Do a quick check on this and you'll see. This means that any efficiency gains for higher voltage will be lost in extra battery pack weight. There are more cells to check or BMS and more interconnects to buy and maintain. I'd say ~350V bus is a good compromise.

EV controllers are not servo/position drivers. The loss of the "encoder" will not default the controller to spin up the motor out of control. Or at least it's not... hmmm... "best-practice" for the software to spin the motor in all directions if it can't see shaft feedback. That used to happen in the early days when servo controllers were all hardware. Ever seen an IBM mainframe printer feed a whole box of paper in 45 seconds? But yes, LED feedback on this application is not that good.

That's good progress, Paul! Hopefully David hasn't given up on this plans either.

JR