Making ferrite transformers for my 500V x 6MHz power supply?

High frequency, antennas

Making ferrite transformers for my 500V x 6MHz power supply?

Postby chrismb » Mon Oct 04, 2010 4:02 pm

{Not sure if this is for the RF, HV or digital forums!!}

I have built (a year or two ago) a high frequency PSU that can put out a few x100V square wave to a few MHz. It took me about 6 mths to research which components worked sufficiently well to actually do this, then another 9 to build it. Anyhow, I will put up all the details of it here quite soon when I start posting up, in earnest, some of the detail of the experimental work I've been up to.

Here is an example of the output, running ~400 V x 2 MHz into a load [25W bulbs]:

400V_2MHz.jpg


At this precise moment it is in a bit of a 're-fit', hence me asking this question now: The original intention was to run this PSU output on top of an up-to ~5kV DC potential. What I would ideally like is to have the flexibility to 'convert' this output into a bigger voltage output, but still square wave.

I have no knowledge of achieving such a feat of electronics but the way I see it is that if I feed this into a sufficiently sized ferrite core then as the current switches 'on' then the core will begin to magnetise and as it does so will generate a respective potential across the secondary for as long as the core is ramping up in saturation. Once switched off, the magnetic current will die off (how quick would it do that?!?) I'm looking for 1 to 3 MHz range so I guess it is likely not a big core that could accomplish this as there won't be that much energy actually flowing in each pulse [dunno, just guessing...].

Well, there it is. How do I 'transform' the output you see above into the same output waveform but with a higher peak voltage?
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Making ferrite transformers for my 500V x 6MHz power sup

Postby Doug Coulter » Mon Oct 04, 2010 6:55 pm

To handle a sq wave, you need a transformer that will handle a broad frequency range, as the fourier transform of a sq wave has all these components in it (odd harmonics). The rule of thumb is at least 10x to get something that still looks good on the other side -- eg get the harmonics through the 9th or so. You have to have at least 3 and 5 or it's going to look pretty ugly.

In other words, to have a sq wave coming out at 5 mHz, you need a transformer that will handle 5-50 mhz, roughly. It's a lot easier with a narrow band signal (sine) or if you don't care if the output is more sine-like. That happens all the time in our HV supplies -- we put in a sq and get a sine out pretty often. If you only need a sine, and you can deal with retuning when you change frequency, there's a bunch of easier (and cheaper) ways to get this done. The popular Pi network comes to mind for that, and can do quite high stepup/down ratios easily. (radio amateurs handbooks show these and some design info).

Yes, you use a ferrite, probably, for this, and not the same type as for a switching power supply, they have far too much core loss at these frequencies. You get a lower mu one designed for the range of interest. Usually in this range, you'd use a ring core of fairly low mu, and wind primary and secondary windings on it all around. The mu of the good HF cores is low enough that if you wind a primary on one end of the ring, and the secondary on the other end, there's enough magnetic leakage to make it not work very well at all. So you have to put the secondary winding over the primary, for example, and usually wind around the whole circumference. Users of large, low mu cores are kind of rare, so they can be hard to find and not real cheap.

Typically, you design for primary inductance impedance about 4x the driver impedance at the frequency of interest, if you can, so the transformer itself doesn't load the driver too badly.
You can't use an "open collector" type drive and get a good waveform, as you store some energy in that inductance, so you have to drive it both ways, and with no net DC across the primary.
Also, you have to stay inside the flux that the core will saturate at. Suddenly turning off a single sided driver will give you a "kickback" that can be many times the original drive volts.
(an auto spark coil goes to nearly 600v when the points open, on the primary, for example)

The main problem with a broadband transformer is going to be self resonance of the winding stray capacity with the winding inductance, usually the one with more turns is the one that gets you in trouble.

Here's a place that sells the right type. The K type for ham baluns looks good here, but of course you'll have to do the figuring on the windings.
https://www.amidoncorp.com/
You might find others. It's a weird business, most of the suppliers won't sell small lots, these guys will. Nearly all surplus will be the high mu low frequency type stuff. Or the deliberately lossy stuff used for ferrite beads to eat noise.

For a one-off, you're probably not going to need or want to find the very smallest core that can work, and more meat makes things much easier. The manuf's will usually give you some info on the inductance per n turns, but I find it's best to simply wind some and then measure by using a capacitor and a signal generator/scope to find resonance.

Hint -- when you first test, start with low voltages and gradually crank up to make sure you're not driving the thing into saturation -- your driver won't like that very much at all, as it will look like a short circuit when you attempt to drive past that point. Perhaps a small series R that driver can fry easily in there is a good idea, and watch for it getting hot at first.

Large turns ratios tend to get you in trouble with self resonance too low, as the inductance goes up with the square of turns, and the capacity also goes up linear with turns, so you hit a wall with that if not done carefully.

When you can, it's better to start with all the volts you can get to keep the turns ratio down. I find TV horizontal flyback driver tubes are the cheapest way to get to the kV range, and as beam power tubes, a lot easier to drive than old high power transmitting tubes with similar voltage ratings. You can run, for example, a 6KD6 at around 1.5kv on the plate supply for 3kv pk-pk with only about 100v drive to the grid if you hook it up right. Whereas a big transmitting triode might need a few times that drive, and also need to be "neutralized" so it doesn't oscillate at some odd frequency. (Chris, I know you don't want to have to build another driver but others may read this) 6kd6 is rated for 7kv peak on the plate, but at that point you need so much grid bias to turn one off, it's hard to handle. I am using a CRT cathode driver chip to drive my tubes grids, they work great, but I also bought a lot of them as CRT's are going the way of the dodo, and semi makers quit making cool stuff like that as soon as order volume drops. The chip takes me from a couple volts to 90 or so, with some oomph to drive the grid capacity.

At any rate, I'm doing some RF/arbitrary waveform work here (for similar reasons), and have the same problem. Right now it's on the burner behind the data aq /data mining stuff, but I'll be getting to it soon myself. Pretty much the same problem, only I need even higher voltage output and broader bandwidth (so I think right now). The transformer guys don't like to hand out the secrets of doing this well, or those AR or ENI broadband power amps wouldn't cost more than a new car (and that's used on ebay!) for long. The best I've been able to pry out of one is sort of "make the best air core one you can, then stick in some ferrite to get it to work at lower frequencies". Which is how I plan to start. In my case, I'm going to begin with windings on pvc pipe, one inside the other, and then stick in some ferrite rods....I'll need the insulation I get that way, and can maybe tolerate the weird impedance that's going to hand me on the driver side better, as I have some practice with that issue.

I believe we have some ace RF guys here who might know more than me about this -- ya'll chime in. I mostly just did ham type things, up to about 2kw or so.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Making ferrite transformers for my 500V x 6MHz power sup

Postby johnf » Tue Oct 05, 2010 2:01 am

Chris
You didn't say how many 25W bulbs
broadband operation is relatively easy at lower impedances.
Braodband operation into a load above around 600 ohms is difficult --all of this from a matched point of view ie RF.
I'll point you at a good book --you should be able to get it on interloan from your local library.
Transmission Line Transformers
By Jerry Sevick, W2FMI

ISBN 1-884932-66-5

Another of my prize books on my shelf @ home.\

PII match, L match, or whatever match used with tubes is inherently narrowband so your square wave will end up looking sine.
?'s do you really need the square wave if so another approach maybe needed ie low impedance supplies and very fast switches matched output impedance into home made high impedance coax (200 ohm is about the limit).

We might need a little more info on what you are trying to do

I had this problem driving a superconducting magnet
They wanted triangle response I gave them sine but at higher compliance. From -70% to + 70% of a sine amplitude the form is very near triangular.
End result resonant drive 1.5kW against 18kW if I forced the triangle wave non resonant.

ie what can you put up with in waveform distortion
johnf
 
Posts: 433
Joined: Sun Aug 08, 2010 3:51 pm
Location: Wellington New Zealand

Re: Making ferrite transformers for my 500V x 6MHz power sup

Postby Doug Coulter » Tue Oct 05, 2010 8:42 am

I'll have to get that book too, I see. Cool! Hey, you know we can't tell from Chis' explanation if he needs more AC amplitude, or just to put a bias on top of the signal he's got. If the latter is true, all he needs is AC (capacitor) coupling and a big choke on the DC bias....pretty easy compared to a large stepup. I think Chris is driving ion acceleration/control electrodes with this, and may not need a ton of actual power for what he's doing. He may be able to waste some from his driver (it's big) to get what he wants, and just kinda swamp SWR kinds of issues pretty much.

Edit: Found it on Amazon, used, it's on the way. TL transformers are cool for lower impedances where the C between the windings is part of the TL.

Chris, it's the rate of change of the magnetism that produces the volts on a winding. Once you have an inductor drawing current, removing the source drive means the voltage flips more or less instantly, only slowed by stray C's, and can switch polarity and go quite far (the auto spark coil) in the other direction. Capacitors have "inertia" in voltage, inductors in current. A capacitor left alone will stay at the same volts you charged it to, more or less. An inductor drawing current will try to keep that current flowing by adjusting the volts across itself until something external draws that current (or losses do, same idea as capacitor leakage).
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Making ferrite transformers for my 500V x 6MHz power sup

Postby chrismb » Tue Oct 05, 2010 9:33 am

Doug Coulter wrote:I think Chris... may not need a ton of actual power for what he's doing. He may be able to waste some from his driver (it's big) to get what he wants, and just kinda swamp SWR kinds of issues pretty much.
Yes, exactly. I designed 'well-over capacity' for exactly that flexibility. I dare say/hope the PSU will live on for service in future projects. Each channel (it is a 4 channel PSU) has a nominal rating of 1.5kW continuous and 6kW pulse, 24kW total. Yet the device it is intended to drive draws 100's of microamps, total! The actual reasons for this seemingly huge over-engineering is that the only FETs I found capable at all of the high voltages/duty cycle I wanted can handle 1 to 3 A continuous. Also, I was originally planning that I would need to terminate my transmission lines with 400 ohms or so of termination impedance, but this seems unlikely now.

Doug Coulter wrote:Chris, it's the rate of change of the magnetism that produces the volts on a winding. Once you have an inductor drawing current, removing the source drive means the voltage flips more or less instantly, only slowed by stray C's, and can switch polarity and go quite far (the auto spark coil) in the other direction.
I am well aware, though very unskilled, in these limitations but was thinking that there may be more physics possible in manipulating the pulse edges so that what I get is a 'flat top' of steadily increasing magnetic flux, then a reversal. I was just hoping there may be more to the subtleties of ferrite behaviour that might mean I can dodge some of these issues with a huge square wave capacity. I was just thinking that hope may overcome reality, if I hope hard enough ;) .
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Making ferrite transformers for my 500V x 6MHz power sup

Postby chrismb » Tue Oct 05, 2010 9:57 am

Doug Coulter wrote:I think Chris is driving ion acceleration/control electrodes with this
Yeah. The idea is that by creating a rotating electric field I can trap and maintain a circular ion beam (thus requiring only enough input to make up for ionisaton and Coulomb collision losses). So the intention for the circuit implementation looks like;

charge_pump_diag1.jpg


The theory of this says that the bigger DC1 is, the higher the total charge confinement. But if DC2 gets too small then you loose orbital definition. The limiting case is where DC2 tends to zero and you should then end up with a sort of Penning trap. If you wind off DC2, you face the prospect of losing the 'ion recovery physics' I am trying to create, and you end up with a low-focus cyclotron-like Bq=mw orbit that tends towards a Thomas-focus orbit ['squared off'] orbital shape:

thomas_orbit.jpg


As my device has both magnetic and electric field azimuthal variations, it all gets a bit hairy, and so the stronger I can make that voltage delta (DC2), the [supposedly] better focussed and circular the orbit shall be. You might be able to see from this image the square shape the orbit takes on, hence the desire to boost the available 'DC2' voltage.

P5020017c.jpg


This is getting a bit ahead of itself and I will likely need to go over all the theory of this in one place in the nuclear forum for the motivations and origin of this work to make sense. The way this work is going at the moment is that it has already drifted from the original concepts due to side-investigation on side-investigation that have distracted me away from the original concept.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Making ferrite transformers for my 500V x 6MHz power sup

Postby chrismb » Tue Oct 05, 2010 10:15 am

For interest, the insides of this PSU looks like this;

4_sq_PSU.jpg


One issue arising is that I need to flick between 4 discrete 'on' states between each of the output states I need, because the signal outputs have to rotate in quadrature. If I do that with commercial grade clocks and some XOR logic, then I end up with switch-on spikes as the registers change. So I also created a 'quad-stable' circuit to do this:

quad-stable.jpg


['AC', obviously, means an alternating signal whose Pk-Pk crosses the logic thresholds of the NOR inputs.]

I did try to get spice to simulate this circuit, and I could not get it to do so. This does not appear to be a circuit that submits to usual, or maybe even intuitive, analysis but it does do the job and it cycles discretely through 4, and only 4, output states.

The whole PSU is more cumbersome than perhaps it needs to be because I designed the back-end to be fully-float-able to 5kV and also put safety circuits and EMC mitigation into it. So it has several fully isolated and separate power supplies in it, some operating around ground along with the oscillator circuits, then 3 different supplies to feed the stuff the other side of the opto-isolators (middle of the lower cavity).
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Making ferrite transformers for my 500V x 6MHz power sup

Postby Doug Coulter » Tue Oct 05, 2010 11:01 am

Yes, Chris, we've been waiting for your main work on the nuclear area, perhaps "other forms of fusion" so the rest here know what you're talking about ;) I had the advantage of knowing already, but that's not fair to the rest. I did and still do think your idea is worth going after :D

The flat top on a sq coming out of a transformer is not very hard if you have enough leeway in amp/seconds in there -- the current in the primary is rising at a constant rate during that, but of course what, if any, current is being drawn by the secondary has an effect as well (large). So there's always some kind of jiggery-pokery that can be done, especially if you can afford to burn a little power someplace. In my experience, particularly with large stepups, it's the edges that give the troubles, due to stray C's and R's where you wish they weren't. Remember any current drawn in phase by the secondary tends to reduce the magnetism in a transformer -- pull it out of saturation. This fact has resulted in some of the "worthless for fusion" X ray transformers we've managed to obtain -- they just won't "go" at no load, or very light load, they just get hot fast.

I'm still not clear on the speeds and feeds here, the loads involved. Probably the best thing is to just build a transformer or two and fool around a little on a test jig -- you'll learn quick and get that all important "feel" for how it all works together -- then you'll be able to design what you really need/want much more easily. I've had plenty experiences where what was on either end either helped or hurt various aspects of the net performance....you'd have to be the judge of that!
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

TLT pdf

Postby Doug Coulter » Fri Oct 08, 2010 12:34 pm

Here's a pdf tutorial on broad band (but low ratio) transformers. I've seen better and will add links here when I find them.
TraskTLTTutorial.pdf
Trask TLT tutorial
(54.31 KiB) Downloaded 557 times

Here is another, same idea, different author.
sevickTLT.pdf
Sevick TLT
(208.42 KiB) Downloaded 502 times

A gov contract built 4::1 high bandwidth high voltage TLT (coax, air)
GvtTLT.pdf
More details
(1.44 MiB) Downloaded 640 times


Note, I found another that was too large to attach here, so I'll put it in the library.
Check the HighVolts dir, and I put the rest under RF.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA


Return to RF

Who is online

Users browsing this forum: No registered users and 2 guests