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

[e*clipse]

Yeah, that inductor is certainly impressive.

Wow - 48 pounds!! That's the weight of 6 Leaf modules.

This one weighs about the same, if you have $1500 for an inductor..
http://www.coilws.com/images/HCS-161...%20REV%20A.pdf

Hmmm, I wonder if it's possible to make a torroidal inductor that weighs less in this power range. There doesn't seem to be much off the shelf stuff in this range; it's all custom pretty much. I need to review my notes on that...

I can definitely see the advantage of increasing the switching frequency. In this case, going from 10kHz to 25kHz dropped the required inductance to 110uH. That would certainly be a good argument for SiC switches.

The control's switches aren't the problem - three TO-247's can handle these numbers. And, as anyone who's looked at real switches knows, there's a huge range in current handling capability (proportional to price) in that package size. They don't all have to be spec'd on the most impacted part's requirements.

However, it sure seems like based one weight and $$$, it might be best to KISS (no boost) and just buy more batteries.

If the constraints are 50kW with 2X boost for the cheap model this could work out - I'll re-run some #'s for that.
***edit*** that doesn't help. Reducing the power helps, but increasing the boost amount offsets any gains.

- E*clipse

[e*clipse]

OK YEA! I may have some good news.

I've been working through a torroid core inductor design. It can get pretty involved, but the first stab at it with some big fudge factors thrown in is really encouraging.

There is really good info on inductor design here:
Powder Core Documents | mag-inc.com

The first thing you need to do is calculate the L*I^2 value; it's in millijoules.

So, using the worst case of 0.273mH from and 200A prior examples:

LI^2 = 0.273E-3 * 200^2 = 10.92milliJoules

Now, for a first stab you look it up on one of a zillion charts in the core catalog. That's the first reference in the link. These charts give rough suggestions for appropriate cores based on the L*I^2 value.

For my first stab at it, I looked up kool Mu cores and standard powder (MPP) cores.
On pg 77, there's a kool mu core #77095 that could work. It's OD is 48mm and it weighs 86 grams. Yes, grams.
Working out more details, you calculate how many winding wraps it will require. There is a number called the "inductance factor" Al, used in these torroids. In this case, it's 57nH/T^2. Divide 0.273mH/(57nH/T^2) and find the number of turns: 69.

Now, you can't just use the 69 turns number, because it's based on no load. In a number of examples (remember the zillion charts) they came up with about a 70% difference. I'm sorry about the fudge factor thing, but this is REALLY just a rough stab at it. Adjusting for the difference, I came up with 98 turns.

Using two #6 AWG wires with a current loading of 800A/cm^2 allows 200A. Again, I don't know if this is good, but it will handle the current and will fit in this size of torroid.

If you don't trust these numbers, and I really wouldn't blame you - LOL! The biggest torroid they sell has an OD of 166mm (6.5") and weighs about 2.2 kg.

So let's say half the inductor weight is core and the other half windings. That would be about 4.4kg in an absolutely worst case scenario.

Long story short - even with big fudge factors and using a huge torroid, it would be possible to cut the inductor weight to between 5 and 10lbs.

- E*clipse

[P-hack]

It is like a white whale I've been through the boost thing quite a bit, about 100 pages back, it isn't your friend. It is probably going to disappear from hybrids once they figure out that they sell cells that can hold more than 1.2 volts.

edit: or they simply switch the motor vhz rating on the fly, a-la series-parallel/wye/delta/phase groupings/etc. I would trade no inductors for more switches if circumstance called for it.

[MPaulHolmes]

Originally Posted by P-hack

It is like a white whale

What is like Oprah's Caucasian sister?
haha

For my part, I LOVE talking about inductors. There are so stinking many AC industrial motors, and I think there are a lot of interesting possibilities for using them for low power "after russia nuking everybody" mad max relatively low power but very functional vehicles. LOL. And, using them as a necessary boost stage for when you want the inverter to be used as 240v split phase too. There are so many interesting applications! I agree that there must be a relatively cheap and simple to assemble toroid inductor. Please don't stop talking about it! (Well, if you get bored, you can. haha)

[P-hack]

well, I'm not suggesting it for an EV since it is deadweight (and you'd need a large cooling plate), but theoretically 12 of these should get you 300kw, and might have stationary uses as well.

And I wouldn't be surprised if a couple fell in someones toolbox while @ u-pull-it... But nobody seems terribly fond of the idea mmm... 12 interleaved boosters... mmm...

edit: heck, you might be able to get rid of the capacitor if 6 converters are dumping current at any given time. though 12 high side gate drivers for regen is a bit much.

[MPaulHolmes]

OK I"m not going to get it finished and tested tonight. I'm changing the raw temperature to real degC. That is probably one of the more important ones.

[P-hack]

random motor, actuals depend on lots of stuff, but useful boost argument, wondering about the next level (field coil current as limiting factor) and starting with star/series and finishing strong w/delta (blue lines) and a second peak (and turns in hand or pack selection or ...).

36*650/72*300 ~ %92 efficient, pretend that is the boost converter I recon, I'm pretending it is the baseline for star delta switching (72v being delta kv curve, 36 being star).

[e*clipse]

Here is a quick run-down on a more detailed toriod inductor analysis.

It turns out I was quite a bit off on the first stab - the LI^2 calculation was off.

Fortunately, that big core I mentioned will work at 25kHz, with a max design current of 200A.
I reduced the current density to 500A/cm^2, and this requires 3 parallel windings.

Specs:
L = 110uH
Design current: 200A Absolute max current 300A
copper loss: approximately 400W
core loss: 13W
Inductor OD: 228mm (this includes the windings)
Inductor thickness: 203mm
Core weight: 2800g
Winding weight: 10.9kg total weight: 13.7kg >> 30lbs.
3 windings of 6AWG wire; 127 turns >> very high fill factor of 66%


So, it will work for the booster switching at 25kHz. Unfortunately it's much bigger than I had wanted, but losses are low. I guess it's ok at 30lbs, but that's pretty chunky IMHO.

Most of the weight is windings. A quick way to reduce weight would be to increase the current density to 800A/cm^2. This would make a 2 parallel winding coil. It may not be so bad, as the DC current is 175A. The 202A is the combination of the DC+AC, so with a duty cycle of .3375, the two windings would probably be ok.

I need to sleep.

- E*clipse

[MPaulHolmes]

Me too! I think 30 pounds is very reasonable.

[P-hack]

since it is kinda related, I'll just leave this here. But the idea is to push the fields to their current limit, and switch to delta well into the field weakening part of wye (and careful switching back) so you can push the field current back up (though that is going to affect bemf...).

this quick and dirty needs reverse blocking on the delta switches, so it is 3x dual igbt + 3 single reversed, though it may be possible to reconfigure the inverter or something and use less switches.

I'm just guessing here, probably time to sort out a dyno (or read bajillions of speculative papers and all the hieroglyphs therein).