Capacitor Bank for Open Revolt: What Size?

[Nevyn]

So I'm working on our controller using the new and improved controller PCB, the SMD current sensor, and the VLA based IGBT driver from mrbigh.

Here's what we got for IGBTs: 4x Fuji 300A 1200V

Running these IGBT's together would give 1200A. We plan on setting it to 1000A max current.

Based on what Paul said long ago in the main thread:

The ripple current rating is basically how much current can rush in and out of the cap without it overheating. The bad bad bad boys above can take a very large amount of ripple current each, so like 10 or 20 in parallel would have an extremely high ripple rating. (like 500 amps or something) You only need like around 5% of the max current for ripple rating, so they say. 50% is the worst case scenario.

5% of 1000 is 50A; I'd rather have a little more room for error, so let's say 8% (80A).

Is capacitance tertiary to getting a low ESR and high ripple current rating, or does it matter just as much?

Looking for best as well as least expensive route.

Thank you all!

[DJBecker]

Picking capacitors is really, really hard. It's all about trade-offs.

The primary task is to clamp the voltage spike from drive turn-off. If you can't absorb the energy, the voltage spike will avalanche the MOSFETs or IGBTs. So you need at least a minimum capacitance to handle the inductive surge.

We can't tell you how much that will be, since it's highly dependent on the batteries, the battery condition, and the details of the battery cable runs.

If you have the bare minimum, the capacitors will have to deal with a high ripple voltage/current. They'll need a low ESR to not overheat from the current, and a high ESR to limit the current. You can't have both, so you'll need more capacitors to both lower the per-cap current and spread the heat out.

If you are designing for this regime, you pretty much need to use polyester (Mylar) film capacitors.

If you increase the capacitance beyond the minimum, with capacitors that can handle the current, the ripple voltage will decrease. That will decrease the dV/dt, which will widen the range of capacitors you can use. At this point you can start using aluminum electrolytic capacitors.

Continuing to that corner of the design space, you can minimize the ripple voltage rather than just barely handle it. The means much more capacitance, generally a dozen or more high density, moderate ESR capacitors. Generally these are connected close to the battery cable input to the bus bar, with low ESR film caps connected close to each switching device itself to handle the inductive spike from the bus bar.

A side note about this: mixing capacitors with different characteristics can be effective at reducing noise and the peak voltage. Or it can create its own ringing as current "sloshes" back and forth between capacitors. Sometimes you just have to built it and measure.

In general, when picking capacitors you should avoid physically large surplus ones. They are usually designed for power supply filtering, especially handling brief power flickers. They can handle a small ripple voltage at 60Hz, and a partial discharge every once in a while. If you have them absorb the same ripple voltage at 15KHz, that's 250x the power they were designed for. They will heat internally and they don't have the surface area to volume ratio to get rid of the heat.

[Nevyn]

Now, why would do something like go and confuse me like that? Lots of info in there, some of it is helpful, some of it makes me realize I've forgotten a LOT of stuff and need to research to fill in some holes. That's why I'm doing this with friends; one is a mechanical/electrical engineer.

If we're going with twice the amperage, do we need one, all other things being equal, that's twice as many caps (of the same type) as the 500A controller? Can we go with the same number of caps if we have twice the ripple current handling ability?

[jackbauer]

My original igbt design use 2 x 1500uf 450v and 2 x 4700uf 450v caps. They don't even warm up. Driving for over a year. My motorcycle controller uses 2 x 6800uf 250v and my new liquid cooled design 3 x 4700uf 400v. My theory being that its a pain to deal with lots of little pcb mount caps and its really a game of swings and roundabouts. I also used 1uf 400v film caps. one across each of the big boys.

[Nevyn]

Sounds good, Jack - but I'm also wondering about expense. I'll have to see what I can rustle up from the guy at the welding repair shop - I might be able to get free caps! (there will be a different thread explaining that later today).

[sawickm]

For those interested in DC Link Capacitors used in IGBT based inverters.

This is a link to the Capacitor Manufacture that EVnetics is using in the 1000 Amp Soliton:

SBE inc. (SB Electronics) Developer and Manufacturer of Film Capacitors

-Mark

[MPaulHolmes]

Those are incredible. how did you find that out?! I don't think they were very excited about that information getting out.

[MPaulHolmes]

Future carries them:

SBE INC | 700D10896-348 | 700D Series 600V 1000 uF ±10% Metallized Polypropylene Power Ring Film Capacitor - Future Electronics

Well, they are out of stock, but HOLY COW THAT'S A CAPACITOR!!!!!!!!!

[sawickm]

Originally Posted by MPaulHolmes

Those are incredible. how did you find that out?! I don't think they were very excited about that information getting out.

Hi Paul,

I knew they were using those Ring Capacitors for more than a year now.

The only reason I shared this info was it became public on May 12th, on the SBE's website.

-Mark

[mrbigh]

Originally Posted by sawickm

For those interested in DC Link Capacitors used in IGBT based inverters.

This is a link to the Capacitor Manufacture that EVnetics is using in the 1000 Amp Soliton:

SBE inc. (SB Electronics) Developer and Manufacturer of Film Capacitors

-Mark

I'm thinking that..........the Soliton was ground designed around the reel capacitor...