Homemade Wideband Current Transformer

We made a current transformer for measuring ~~100 amp, nanosecond pulses. It was just a small ferrite core with maybe 5 turns, a shunt resistor, and some coax to the scope. Seemed to work fine.

A Rogowski coil needs no core so can be very fast.

Nice.

I am hoping to use this for checking cable emissions for EMC, so it needs to be reasonably sensitive.

Making a CT is easy beans. A hundred turns on ferrite into a suitable resistor (a terminated transmission line, in this case) will easily do MHz, maybe 20 if it's somewhat small. Amazingly, commercial CTs can suck eggs: a Triad CST306 (I think) has its lowest resonant mode at something like 300kHz!

It gets better if you wrap it with aluminum or copper tape, because this reduces the leakage inductance and coupling between winding segments, suppressing the helicotoroidial resonator modes. Probably, if lossy tape were used, it would help dampen as well as shield the resonances. A high burden resistance also discourages resonances, forcing the feedpoint to not be a good current antinode.

It helps a lot to use fewer turns, raising the resonant frequency. Segmented windings, arranged carefully, may help to discourage certain resonant modes. If a large ratio is required, multiple stages can be used.

Past maybe 100MHz, with any approach, you're pretty much working in black magic territory. Good luck.

Tim

-- Deep Friar: a very philos>

I think you just need to stop the current from flowing, so a layer of copper tape with some other insulating tape between the overlapping layer section should work.

George H.

To get a response curve that makes sense for an RF current transformer, you have to be able to excite it with a known current. The usual way to do so is by inserting the transformer into a suitably sized and carefully terminated coaxial transmission line. The line's characteristic impedance ensures that the current is flat over frequency (to within a few dB or so). The 'suitable size' bit is to make sure that the insertion of the transformer does not make a discontinuity in its impedance. I check this through reflectometry: The line with the transformer in place should look like a smooth transmission line all the way through.

That's the way I test beam current transformers here. In its final application, the primary is not a wire; It's a particle beam. I get bandwidths of about 3GHz. OK, my kit is upwards of 35k$ a piece. Then again, it's all custom made, has to hold a vacuum too, be radiation resistant, etc. That has a way of driving up the cost... (Just the ferrite cores are 1k$ a piece, and I use five.)

Come to think of it, the thing is warped such that many wouldn't even recognize it as a transformer. We call them "wall current monitors". Here's a picture, if anyone is interested:

Jeroen Belleman

Ha! I had a thought last night it might be something like that. Be a big coax for my one though. I did wonder if a sort of giant microstrip (megastrip?) would also work :)

It looks like that same company I linked to also does calibration fixtures:

Cool, thanks!

Word of the week ^^^^^^^^^^^^^^^

Yes I must try to use that one more often.

But, Success!

I built the transformer, along with the finger-shredding copper tape electrostatic shield.

Seems to work well, as far as I have been able to test.

Transimpedance is 20 ohms.

Looks beautifully flat response when you look at the first 20MHz.

Generally flat response continues to about 200MHz, with some +/- ~10dB wiggles up to 1GHz+. Don't know how much of this is the response, and how much the excitation. I might try my megastrip idea to test it further.

Will try to post some pics later.

OK, here it is:

I made a mistake, it is more like 7 or 8 ohms transimpedance. About 1 ohm impedance on the "primary",as could be expected from the 7 turns and

That is with 1mA from the tracking generator, so the 7.87mV display is

I love the display and output of HP stuff from that time.

my scope has some silly parallel port printer module, and there's some painfully slow windows software that runs over a 9600 baud serial port to do screep captures like you posted. the fonts seem optimized for use with a plotter or something like that.

Yes it's nice - picture a bit blurry since was just my mobile phone.

The spectrum analyzer expects to plot to a plotter, so that plot was acquired over GPIB, then run through a HPGL to PNG filter. You can tweak options to get different resolution and colours / line thicknesses for the various elements of the plot.

Have you used any GPIB to USB converters that function and didn't cost $500?

It seems silly to run out and buy a PCI card for this and be chained to one desktop forever.

No, actually.

Yes, I use a old SFF PC just for that, running a linux distribution. It does all the "GPIB stuff", has scripts to automate various "experiments". It's headless, just a box, but I can remote in to it of course. For the rare times I go "on the road" when I need GPIB I do have a NI USB interface, but it is not quite as reliable.

The thing that causes most problems is handling a "screen dump" from the various instruments. I think they all want to be controlling a HP plotter then, instead of being controlled by a GPIB master. I never really got to the bottom of it, but I needed the PCI card to get it to work reliably.

tweak

for

to

Personally i use the Prologix, LLC devices; nicely better price. Seem to work just fine. Back in the day i used to be a bit if a GPIB guru. Ask me by PM to discuss the instrument to plotter things. We may need to trick the interface a bit to read the control words that activate the plotter. However they are usually settable in the instrument GPIB menus. Some configurations do not need the control words (promiscuous transmit and receive).

?-)

• Takes forever in load attempt; i gave up.

• Should work fine; ground ONLT one end; the other leave open.

• Takes forever in load attempt; i gave up.

Sadly, the link to the drawing fails.

Yes, I'm not very surprised. It's a php script putting out a TIFF picture. It has 'header("Content-type: image/tif");' as the first line producing output. That works fine for Mozilla- type browsers, but IE doesn't seem to understand it. So I put in a quick hack that makes it work in-house at least. Of course, you are outside.

If you have a suggestion towards a good fix, I'm all ears.

Jeroen Belleman

I have used that too, worked fine for simple things, but ISTR it seemed to get lost when doing sequences of commands.

Or I did.

Thanks Joseph that is a generous offer :) I would love to take you up on it sometime but I am up to my neck in other projects at the moment, would not be able to do it justice!

Thanks,

I've heard good things about this adapter that is alleged to be 100% compatible with Agilent's controller. And the price can't be beat.