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

[Piotrsko]

big pieces of copper require large quantities of watts which is only restriction. I have a Dillon oxy acet torch that will dial down to a 1/64 x 1/4 flame tip at 6500 degrees for making jewelry. a resolder oven should work with 60/40 leaded solder and a good paint on flux paste and pre tinned everything. heck a harbor freight hot air gun should work with through holes. EASY PEASY.

trying now to convince chief finance officer that I REALLY need one of these. hurry up

[e*clipse]

Originally Posted by Piotrsko

big pieces of copper require large quantities of watts which is only restriction. I have a Dillon oxy acet torch that will dial down to a 1/64 x 1/4 flame tip at 6500 degrees for making jewelry. a resolder oven should work with 60/40 leaded solder and a good paint on flux paste and pre tinned everything. heck a harbor freight hot air gun should work with through holes. EASY PEASY.

trying now to convince chief finance officer that I REALLY need one of these. hurry up

LOL about the CFO...

Thanks a bunch for the encouragement! Guess I need to play with this a bit. I have used both the toaster-oven and hot air reflow soldering methods with tiny stuff.

My main worry is overheating the parts. The reflow soldering methods have pretty specific ramp up ramp down times. I wonder if the relatively huge heat sinks would hurt the IC's. Maybe a person could put the finished part on a big chunk of cold aluminum or copper to speed up the cool-down part?

- E*clipse

[e*clipse]

I've been looking into some advanced cooling methods, mostly for the motor side of this problem. Here's some of the stuff I've found:

Direct spray cooling one of the most effective ways to cool stuff. In fact the military is actualy working on direct spray cooling. Here are a couple of interesting links about it:

https://www.highpowermedia.com/blog/...oling-concepts

An Overview of Liquid Coolants for Electronics Cooling « Electronics Cooling Magazine – Focused on Thermal Management, TIMs, Fans, Heat Sinks, CFD Software, LEDs/Lighting

Spray Cooling Electrical and Electronic Equipment - COTS Journal

In the first link, it mentions that a typical passively cooled motor can handle 5-8A/mm^2. A cooling jacket will allow about 10-15A/mm^2. The highest ratings were for direct spray cooling: 30A/mm^2.

In the second link, a bunch of sooper-dooper coolants are mentioned.

In the third link, the military's interest in the topic is illustrated:

Quote:

Testing by Southern California Edison on a GE 85HP traction motor was significant and consistent with NAVSEA’s FIAL efficiency data. The GE 85HP traction motor was fully loaded and tested with air-cooling and spray cooling (Figure 2).

In the mid- to 3/4-loads, a 25% to 30% improvement in efficiency was reported with spray cooling. Also, a 10% improvement was reported at 85 HP load and a 15% improvement in sustained overload at 100 HP (total power, in direct contact with the liquid, exceeded 73,000 watts). The test rig and shaft were not designed to operate at higher loads; however, the designer indicated that sustained overloads of at least 200% of rated load were possible. The overload results show that the use of less costly, lighter and more efficient generators and motors in ship, air and spaceborne applications is possible with spray cooling.

Wow! a 25% to 30% increase in efficiency! Also a 200% increase in the rated load! I'm interested! For the inverter, between FOC, leading edge film capacitance, Silicon Carbide FET's, a planer bus and direct component cooling, we've got a killer app!

So, I think I figured out how to circulate coolant between the two plates. The top plate will have good seals to separate the coolant from the control board. The bottom plate won't and it will have a port to all0w the coolant to drain down to the capacitor. It can then pool around the capacitor before pulling it out and cooling it.

- E*clipse

[Piotrsko]

Take a lesson from the gamers. If it dont conduct electricity who cares where it leaks? Mineral oil isnt maybe so efficient thermally BUT.....A lot can be said for dumping your electronics in a pool of coolant be it tranny fluid or baby oil. Thermal mass sinks are rather nice.

[e*clipse]

Originally Posted by Piotrsko

Take a lesson from the gamers. If it dont conduct electricity who cares where it leaks? Mineral oil isnt maybe so efficient thermally BUT.....A lot can be said for dumping your electronics in a pool of coolant be it tranny fluid or baby oil. Thermal mass sinks are rather nice.

Funny - that's kind of the conclusion I came to last night. I did have a bad experience using water on one of my previous computers. Ruined the MB with a tiny leak. So, no conductive coolants - like you say they may perform better, but who cares.

I'm going to attempt to keep the stuff away from the control board, but let it flow all over the power electronics/power bus and capacitor. They all need cooling anyway.

With that statement, there is a lot of simplification - less than 1/2 the leak points.

It would probably be good to make a weep hole or something to indicate leakage is occuring.

On some other notes - Has anyone found a very good **awesome** single pin connector that could be used on the gate pin? It would be nice to allow some disassembly.

Also, I think I've resolved the current sensor issue - it can be one that lies flat on the main circuit board with the power pin sticking straight up through the circuit board. This means there will be no extra wires - woo woo!

Oh yea - thoughts on one big sealed connector, like the OE ECU connectors for all the other I/O? Yes, it would require a harness, but I'm pretty comfortable with that issue.

- E*clipse

[e*clipse]

Regarding sizing those SBE ring capacitors, I found this:

*** edit ***
Yes, this paper is too good to be true. Best to avoid it - I'll keep looking.
***********
http://www.ecicaps.com/wp-content/up...inal_Rev_4.pdf

It is not by SBE. Again, I'll keep looking.

Anyway, the numbers I come up with are quite small for a 50kW inverter. In fact, all that matters, according to the paper is ripple current and ripple voltage.

After running through their numbers, the very worst I could get, using 650VDC bus and the 1.4mH inductance of the motor @ 5kHz switching frequency is 23A peak for the ripple current. That's nothing for these capacitors. They can handle 100A>300A all day long.

For the ripple voltage calculation, I spec'd 1% regulation. Again, the worst case was 5kHz switching frequency, leading to 178uF for the bus capacitor.

If the switching frequency is increased to 20kHz for the SiC FET's, the numbers would be tiny and I would need less than 20uF.

This seems too good to be true. I mean, for the design I showed, I would need the 8.5" diameter capacitor just to physically fit things. The 8.5" 1200V/220uF capacitor would be overkill in about every way one could look at it.

Am I missing something? ?

- E*clipse

[MPaulHolmes]

That's the paper that the EVTech list was saying had mistakes I think. I don't remember the specific mistakes though.

[e*clipse]

That's what I was afraid of.

I'll post an edit so folks won't be mislead by that...

I've E-mailed SBE about this - "we are not in the business of designing inverters..."

Yea, but a tip toward a good reference would be nice...

The main issue I see comes down to this: The motor has an inductance and resistance that probably is important for a good design. The bus characteristics are significantly smaller (both inductance and impedance) yet SBE makes a BIG deal on their website about reducing inductance by using a planer bus:
http://www.sbelectronics.com/wp-cont..._paper_web.pdf
Sooo, perhaps there are two affects happening here - needing to smooth out the bus ripple with a certain amount of capacitance AND needing to deal with overshoot and various switching issues by reducing inductance and impedance within the inverter.

The way that paper dealt with stuff, the motor's inductance completely dominated. This reduced any affects from bus inductance, etc. to "not important at all."

How did you spec the one you're using?

- E*clipse

[MPaulHolmes]

I use Simetrix Simplis, which is free. It's a simulator that has proven to be very accurate when I actually test stuff, so I've come to trust the results. I think it's basically a spice simulator, but I don't know anything about spice. I just use this. To simplify things, I just add an ideal switch, ideal diode, and then a capacitance with a given amount if series resistance, and then look at what sort of inductance corresponds to ripple at various switching frequencies. it's really easy to use.

[e*clipse]

Thank you!

- E*clipse