Voltage rating of SMT chokes for 600V buck converter

[Miyuki]

Hi folks,
I have a question about the voltage rating of small chokes, do they have any voltage rating issues?
It will run from rectified 400V so about 600V DC, using Viper26k
In reference design, they are using just a small SMT one 1mH
But in the datasheet of similar choke they have just 120V max 
And bigger one has in that marketing part just "For High Voltages up to 400V" without any specification

[langwadt]

does it actually see any significant voltage relative to anything other than the air around it?

[Miyuki]

does it actually see any significant voltage relative to anything other than the air around it?

Buck from 600V to 20V so it sees all the difference, it is plenty of voltage

[langwadt]

does it actually see any significant voltage relative to anything other than the air around it?

Buck from 600V to 20V so it sees all the difference, it is plenty of voltage

ah, since you said choke I guessed it was an input choke, not the main inductor in a buck

[fourtytwo42]

Hi folks,
I have a question about the voltage rating of small chokes, do they have any voltage rating issues?
It will run from rectified 400V so about 600V DC, using Viper26k
In reference design, they are using just a small SMT one 1mH
But in the datasheet of similar choke they have just 120V max 
And bigger one has in that marketing part just "For High Voltages up to 400V" without any specification

That's a very interesting question, as you say manufacturers data sheets typically don't quote it even one I found from Wurth that says the inductor is a "high voltage" type  This is going to depend on the quality of the enamel, the design of the terminations and the care or not of the wire during winding so a lot of variables. Whenever I have used inductors at these voltages they are hand made (by me) and often use TIW wire (sectional bobbins can also help) but as for small SMT I am afraid I have no experience to offer. Your best bet is to pick a few friendly manufacturers and Email them.

[T3sl4co1l]

On the upside, it's only functional insulation, safety regs don't care a lick.

They're basically never going to rate it, because it requires some special circuit to measure across the winding.  Which for inductors, is high frequency, so you have all the problems of HF breakdown, and dielectric / partial discharge and all that.

Noteworthy that, if it's a ferrite chip type, it's probably not a good idea to apply much voltage to it -- the pads are terminated on the ferrite itself, which is slightly conductive.  NiZn is higher resistance than MnZn, if you can get some idea of what material they used.  (Easy enough to tell with a "streak" test, NiZn is brown, MnZn is black.  Most likely they're almost all MnZn though.)

And this may still be true even if the leads are simply glued in/on the core, where you might expect they don't have a direct (resistive) path; but at high frequencies you still have the capacitance of that glue layer, which might be quite thin.

If you need it to be correct and safe and reliable, in the design itself, I would recommend getting a custom part made, with insulation rated for the voltage.  It can still be enameled, doesn't have to be TIW automatically -- but it also won't be the bog standard material.  Can also get shapes/styles to your liking, might as well have it all if it's going to be custom, right?

Hacky solution: use a converter transformer, and just ignore the windings.  Flybacks have significant inductance, just shop around for the right one.  Bobbin affords high core isolation, and presumably the wire is rated for at least as much as it is (a few hundred peak, if maybe not 600).  Will be a bit bulkier, and much more expensive, than needed.

Tim

[NiHaoMike]

What about use multiple inductors in series?

[Miyuki]

What about use multiple inductors in series?

Interesting idea, just like with resistors 
What a time we live in, we have integrated switcher with 1kV transistor on the same chip and can't get adequate inductor to match it

[TimNJ]

Noteworthy that, if it's a ferrite chip type, it's probably not a good idea to apply much voltage to it -- the pads are terminated on the ferrite itself, which is slightly conductive.  NiZn is higher resistance than MnZn, if you can get some idea of what material they used.  (Easy enough to tell with a "streak" test, NiZn is brown, MnZn is black.  Most likely they're almost all MnZn though.)

Tim

Just adding.

For wire-wound inductors, I have yet to see any issues related to turn-to-turn breakdown for any inductor. I would say, worry less about the wire itself, and more about the pin terminations where the insulation is removed.

And for chip inductors, my experience is the same. I tried to use a 1206 size multilayer ferrite bead in series with two Y-caps across the AC line, terminated to earth (CM suppression). Ran a 1KV/2KV lightning surge test and best believe those multilayer beads self-destructed. I don't know if it was 100% voltage related (maybe surge current?), but I think voltage makes more sense. Doesn't mean multilayer inductors/beads are worthless...just be careful where you use them.

[TimNJ]

I should add that all wire-wound inductors I've designed have been for <300V AC/DC converters...Obviously for some HV application, YMMV.

[T3sl4co1l]

It is interesting that I often see bobbin-wound E-core inductors in stuff like CFLs, LEDs, etc.  Not obvious if that's because they need just a little more [reactive] power than afforded by chip style parts -- but, I mean, I've seen composite and >15mm SMTs that surely handle more -- or because they need something extra, like breakdown voltage.  Or even cost, who knows, maybe those are just cheaper to make in China in huge quantity?

Tim