Current transformer with good coupling?

I don't think that this is entirely correct.

The transformer equations are

V1=3D L1.dI1/dt + M. dI2/dt

V2=3D M.dI1/dt + L2.dI2/dt

where M - the mutual inductance is =3D K.(L1.L2)^0.5 and usually comes out better than 0.99 for bifilaR windings on an ungapped toriod, always assuming that the winding is "non-progressive", which is to say that you haven't made a current loop in the plane of the toroid, by winding your coil around the toroid as a single - say - clockwise progression.

The book "Coaxial AC Bridges" by Rayner and Kibble

discusses the matter and recommends winding the first 25% one way, the next 50% the other way - back over the first winding and past it, tehn reverswing agains for the final 25%, so that that widing ends at the swame point on the toroid that it starts.

Note that "K" has nothing to do with the winding resistance. The usual way of estimating "K" - by meauring - say - L1 with L2 first open- circuit and then shorted - will give you an inaccurate estimate for K if the winding resistance is significant vis a vis the inductive reactances.

-- Bill Sloman, Nijmegen

Metglas?

John

Try coax cable; say inner wire is primary and shield is secondary. The capacitance between inner and outer adds to coupling and extends bandwidth..

Hello Klaus,

What is the frequency range of interest? Is there any change on (small) DC component (as this will saturate a high-mu core quickly)? Can you give us some info on the specs (phase and amplitude accuracy versus frequency, etc).

Very tight coupling comes with large coupling capacitance also. If there are significant in band voltage components present, these may appear at the output and may ruin your measuring result. You need some special design to reduce the influence of these parasitic effects.

With kind regards,

Wim PA3DJS

ndary.

extends

Not a great idea for a current transformer - you get a lot less copper in the winding window than you do if you wind with enamelled coper wire.

Transmission-line transformers do have their applications, but this wouldn't seem to be one of them.

-- Bill Sloman, Nijmegen

"I want A so I am demanding B."

If you want accuracy, ask how to make an accurate transformer. Is the World's Most Tightly Coupled Transformer the way to get the accuracy you need, or would you be better off with a low-resistance transformer with coupling that is known and easily compensated for?

The low impedance on the output of a current transformer guarantees that the core flux is small, but winding resistance and the load resistor aren't zero. If they WERE zero, the core flux coupling would improve even beyond the 99% level (because the back-EMF of the secondary winding would not be present and the resulting change in the flux distribution favors complete coupling).

You might be able to squeeze that last percent of coupling by loading the secondary with a zero impedance (transimpedance op amp circuit) or even a negative impedance (to compensate for copper losses).

Really good (as in PPM accurate) CTs use a powered feedback winding and operate zero-flux on the core. But that's arguably not a real CT.

The dual-core DCCTs, like the Danfysik units, are phenomenal.

John

If I remember rightly, it is not so much dual core - per se - as stacked cores, and there are more turns around both cores than there are around the first core, and even more around the whole stack in a stack of three.

The book - "Coaxiasl AC Bridges" - that I referrred to earlier in the thread, may have something on that technique too

-- Bill Sloman, Nijmegen

The wideband DC current transformers really need two cores. The wire to be measured runs through both, the feedback winding ditto at the same phasing, but the saturation excitation/sense windings are reverse-phased on the two cores. That keeps the saturation drive and sensing from coupling into the other two. It's very clever.

You can do an AC zero-flux thing with a single core.

John

stor

Obviously not the application I was thinking about - which was for AC bridges being used in a bridge balancing measurement system when you still wanted some sensitivity when the bridge was way off balance.

Yours seems to be a sytem for measuring DC current by detecting saturation in a transformer core. One of my ex-bosses - one Colin Hunter - developed such a system for measuring big DC currents in busbars for the London Underground, where the core was a toroid wrapped around the busbar, and the sensing was done by a bifilar winding on the toroid that formed part of a classic Royer inverter, which relied on core saturation to killed the base drive to "on" switching transistor.

This has the disadvantage that it injects noise into the system being measured.

Your two core solution allows you to balance out the signal that drives the two cores into saturation - the excitation windings will generate equal and opposite voltages on the wire to be measured and the feedback winding.

The sense windings on the two cores will obviously have to be separate, otherwise you won't know the sense of the DC current driving the cores into saturation - the system seems to need to be a bit cleverer than you've bothered to tell us.

-- Bill Sloman, Nijmegen

It's not "my" system, although I did build some once. Danfysik and others have been doing this for a long time. I think Danfysik sold the line to LEM. Julian Bergoz does DCCTs too, I think.

Google it if you are interested.

John

Sorry, cannot say to much about the application beyond we have a push- pull configuration at medium frequency that is very succeptible to duty-cycles other than 50%. The drive needs to symmetrical from both rails, otherwise the transfer is not within specs. Also we have not much ferrite to do with....

Can you integrate up both currents, subtract and run a DC-servo to gently nudge the duty cycle if it wants to veer from 50%?

Also, if core saturation sets in you'll usually see harmonics pop up early on. Maybe that can be used to keep it from drifting off in duty cycle?