A question for the SMPS guys

[Amphiprion]

OK, I know Jon is an SMPS guy, and I got a question or three for him and anyone else with knowledge in this field. Question regards HF inductors and transformers which I know you SMPS guys use.

I need an inductor or transformer for high frequency, high power applications. I'm thinking I'm going to need MnZn or NiZn ferrite cores.

Basically, I'm making something to spark a gap for about 200ns at a rate of 10,000 times a second, and I'm trying to use inductive kick to do it.

Basically, something I can feed <100 volts into, get a DC current flowing through it, throw a switch (IGBT) to cause a massive di/dt spike and get the inductance of either an inductor or the secondary of a transformer to give me on the order of 10kV @ 10 amps for that 200ns pulse.

Any idea what company I might go to to have something like this made?

Also, I need something similar for a much lower power but higher voltage app. On the order of a few watts continuous. We previously wound our own ferrite but had a hell of a time insulating it without internal arcing.

Also, why don't we see audio inductors with these materials? Steel laminate sucks compared to what I've read about ferrites. Is it simply cost?

[Wilk]

Any idea what company I might go to to have something like this made?

The only company I am sure builds things along those lines is Avtech

Avetech

They specialize in making that type of equipment. I see them all the time where I work. I unfortantly don't have the knowledge to help you make one though. I got a few ideas, but most of them would probably end up hurting you very badly. Avtech does have an excellent enginnering staff though that may give you a few good/safer ideas.

There are a few other companyies that make simular units to be used for impulse testing in a EMC lab. Not exactly what you are looking for, but may be close enough.

Solar
Noiseken
Amplifier Research RF division

Most of these companies equipment can be found used at www.used-line.com, that should save you some money if you head that route.

[JonMarsh]

Mark,

I'm presuming you want DC at the output, right?

10 kV at 10 A is nothing trivial, even without isolation. Especially since you also have to limit that voltage so that the reflected voltage at the primary doesn't overvoltage the switches there. Problem with some of the classic ways to do that as noted are break down voltage, plus the reflected blocking voltage issues, and the fact that no transformer is perfects, so there's leakage inductance spikes to deal with. The normal non-isolated way to do that would be a boost circuit, but that's not even remotely feasible from 100V to 10kV.

Bottom line, it's not a "DIY" kind of project. Most HV supplies don't operate at such a high current- just what is your load and what are you trying to do? This is soundling like the kind of stuff done in parallel banks for fusion igniters. Not low budget.

Server and telecom power supplies are more my area of expertise- how to get power densities of 25 watt/cu in and up off line. Small 1 kW power supplies.

[Amphiprion]

This is a $750,000 SBIR project, budget not a problem We are working on a solution in parallel with our planned method (HV high speed switch from Behlke). Someone wanted try this as well as the switch is very sensitive and expensive. My switch system should be ready to test by next week, and if I toast the Behlke I need a backup to work on while we wait on another one (long lead times).

The really funny thing is, when I got told this had become a priority, I had to order a transformer "by the end of the day". As in overnighted from Digikey. You know, one of the ones they keep on the shelf

This is for some new kind of methane reformer. I'm making an arc inside a plasma tube or something. I don't get told details of what the physicists and chemists want, other than spec's the electrical system must meet.

[JonMarsh]

The really funny thing is, when I got told this had become a priority, I had to order a transformer "by the end of the day". As in overnighted from Digikey. You know, one of the ones they keep on the shelf

Oh my Gawd!!!!! :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl: :rofl:

[digital desire]

Thats funny.
I certainly have nothing to add to the thread, except seeing the amount of energy and Jons avitar. I can only think of the scene in BTTF where Doc yells out "Did he say 1.2 gigawatts!?" or words to that affect.

[Amphiprion]

Remember, it's only being delivered for 200 nanoseconds, so in Joules it's a bit less impressive.

[JonMarsh]

True, but with some significant repeition rate. that's a peak power discharged of 100kW, which implies a decent impedance but a lot of voltage. There are IGBT switches and we have in research Silicon Carbide switches that with a few stacked up in cascode could deliver a 10 KV switching capability; then you just have to have a power supply charging up a HV cap stack that can store the energy between pulses every 100 usec. It's about like switching 20 kW every 1usec out of 100, my question is what kind of load would be self limiting to 10A when hit with 10 kV, or is this more like a capacitive discharge where the cap has to be recharged every 100 usec? My real point is that there hasn't been enough information provided about the load and V/I characteristic of the load to really start designing a way to conceptualize an adequate design.

But you certainly aren't going to find it at Digikey! not that I don't buy a lot of parts for eval boards from Digikey... even Mouser sometimes!

~Jon

[Amphiprion]

I'm hesitant to say too much just because I don't know what part of it we are doing is proprietary/confidential (I believe we get the IP rights to whatever we do under SBIR grants). I just hope the Behlke switch works, in which case we've got enough other stuff to do I think this transformer idea *might* fall by the wayside. There's a cool phase I SBIR proposal we're submitting on visual speech analyzers for the deaf for the NIH that I want to get in on. Anything with signals and audio and DSP. No more pulses and high voltage stuff for a long time would be nice. They hired me because of my analog and DSP skills (as well as general coding and MCU abilities), and somehow I am doing this stuff. I'm real green at a LOT of this (this is my first real job after college).

Besides, I've already earned the nickname "smolder" in the first four months of working there. So yeah, a 10KV @ 10A pulse generator is just what I should be working on

[Amphiprion]

BTW, why don't we see these fancy cores in audio? Is it simply cost?

[JonMarsh]

Ferrite cores or MPP cores are good for HF magnetics, because of their low losses at high switching frequencies; in an SMPS transformer, or power inductor, this is critical.

The problem is that their absolute flux density capability is not very high. Several times lower that silicon steel laminations used in transformers and E-Cores.

For any inductor, the linearity issue is a function of permeability droop versus core flux and core hysterisis. MPP and silicon steel have a soft permeability droop, that is, you can push them a certain amount wtih very little change, then increasing levels start to make a steady drop in permeability of the core which is not too fast; SMPS power inductors often being designed for a nominal N/L permeability thats 2X to 3X the full load permeability. Permeability doesn't vary too much with temperature, and powdered cores like MPP or HighFlux can use different binders and particle size to wind up with what is a distributed gap and various initial permeabilities.

Ferries are seen in cheap speaker crossovers (ferrite bobbins typically) because they are actually relatively cheap to make certain types, and the initial permeability is fairly high. Ferrites are nice for HF and transformers because of the low Hysterisis. BUT, when they saturate, boy do they saturate- permeaibility tends to drop VERY quickl. That means that if you overdrive them a bit, the inductance can fall very rapidly.

I've heard this in commercial speakers, such as some of the Legacy models being demo'd at the factory, where playing Terry Evan's "Down In Mississippi", you'd get a crack like sound on an overdriven bass transient that wasn't the woofer being over driven (I was watching for that), but core saturation in the crossover.

In general, all cored coils introduce some level of inductance variation with current (we call that distortion where I come from), so unless the inductor is very carefully designed and overbuilt, the linearity will nevere be as good as an air core inductor. Cored inductors are always about cost reduction. And that's why I don't use cored inductors for the last ten or 15 years.