Characterizing a CT via a major B-H loop?

[MRKN]

I contend this cannot be done!

A popular method of modeling non-linear cores is to obtain the values for remnant flux density, saturation flux density, and the coercive force at some major hysteresis curve, and to then mathematically extrapolate a curve (a few other parameters are also needed to interface with the electrical quantities of a circuit - Ac, lm, lg, N). See "Nonlinear Transformer Model for Circuit Simulation" by Chan for the model used in LTSpice. So I figure this may be a good approach towards characterizing a CT- and I can use it in LTSpice!

The problem with simply hooking a constant voltage source, limiting the current by some resistance, and proceeding as shown http://www.cliftonlaboratories.com/type_43_ferrite_b-h_curve.htm" is that the value measured for the coercive force is dependent on the primary current- i.e. once we are in saturation, and a complete B-H loop is defined, current can be further increased, and Hc further increases! (Bs, Br remain constant). I have verified this in lab with a constant voltage source, limited by some large power resistors. I can source anywhere from 0-8A at 60Hz. I use a current probe amplifier (X input) and a passive 300k/2uF integrator, with the output of the capacitor on a 1x scope probe (Y input).

I believe the error lies in CTs being defined via a constant current, rather than a constant voltage source. I suspect it is for this reason that CTs are not characterized by B-H curves, but rather by excitation curves. Unfortunately, my ability to reason falls short here, and I was hoping for a clarified explanation, or any insight into where my reasoning so far has been wrong.

FYI: I am attempting to characterize a CTs response for very high fault currents (say, 5,000x that which would saturate it, when we are approaching the short time thermal limit- what happens there?), and would like an alternate model to the http://www2.selinc.com/techpprs/6038.pdf" . I have extended the idea presented there for calculating the magnitudes and durations of what trends towards secondary voltage impulses at high primary levels, but would prefer an alternative approach for examination.

Thanks a bunch.

 

[anorlunda]

ping @jim hardy

 

[jim hardy]

The problem with simply hooking a constant voltage source, limiting the current by some resistance, and proceeding as shown http://www.cliftonlaboratories.com/type_43_ferrite_b-h_curve.htm" is that the value measured for the coercive force is dependent on the primary current- i.e. once we are in saturation, and a complete B-H loop is defined, current can be further increased, and Hc further increases!

Doesn't that seem logical ?
If you push the core further into saturation, it gets harder to pull it back out.

I use a current probe amplifier (X input) and a passive 300k/2uF integrator, with the output of the capacitor on a 1x scope probe (Y input).

Great technique I've used it myself. I used a megohm and 1 μf,
did you see much 60hz noise on your flux signal ?

"Nonlinear Transformer Model for Circuit Simulation" by Chan

Just found it. It's a handful to digest at one sitting...
http://ltwiki.org/files/adventures_..._Transformer_Model_for_Circuit_Simulation.pdf
will try to figure out what he's saying.
I'm visiting family in Oklahoma so it may be a couple days

Ferromagnetic phenomena don't lend themselves to linear analysis because, well, they're nonlinear.
Best book i know of is Bozorth's "Ferromagnetism".
Your integrator has already shown you that.
If you have the ability to drive a triangle wave current through the primary winding, instead of your sine wave,
you'll see that flux doesn't follow current because of the time lapse necessary for magnetization to proceed from the periphery to the center of the core.
Simplest explanation for that is eddy currents in the core - your primary isn't the only current present - but old timer gurus at Westinghouse assured me there are other phenomena at play too.
EDIT I may be all wet on that statement - Ferrite should be way faster than ordinary steel

I believe the error lies in CTs being defined via a constant current, rather than a constant voltage source. I suspect it is for this reason that CTs are not characterized by B-H curves, but rather by excitation curves.

The excitation curve to me only conveys the maximum voltage i can expect the transformer to impress across its burden with any semblance of believability.
Once you get past the knee of the excitation curve you no longer have a transformer, you have a saturable reactor with a secondary winding that's reporting what is the approximate derivative of its very nonlinear flux.
They're constant current only within limits of that excitation curve.

FYI: I am attempting to characterize a CTs response for very high fault currents (say, 5,000x that which would saturate it, when we are approaching the short time thermal limit- what happens there?),

I don't think you can do that measurement with an iron core.
A Rogowski coil should do it, though.

Seriously - try a triangle current waveform , and at various frequencies.
The triangle current gives constant dΦ/dt so should make a square wave voltage
Deviation from that square wave will tell you how well behaved is your ferrite core at each frequency
I'd be interested to see your waveforms as it begins to saturate.

Sorry , i don't have my head all the way around your subject yet
but it is certainly interesting !

@Dr.D is interested in magnetics too, and i owe him a look st some related papers. His math is way better than mine.
Hopefully i can get some reading in tonight.

old jim

 

[jim hardy]

A 400 series tainless steel shaft approx 3 inch diameter
excited with a 3 hz triangle wave current(top)
note rounded corners of voltage wave(bottom)

your 43 should be MUCH faster

Thanks, Fair-Rite corp !

 

[anorlunda]

I'm visiting family in Oklahoma so it may be a couple days

Take your time Jim. This is a spring cleaning necro post.

 

[jim hardy]

I contend this cannot be done!

i tend to agree with you.
Chan did it for a plain vanilla transistorized mutivibrator , his fig 7, which has been around since at least the 1950's.
But you're pushing your core a LOT harder than does that circuit.

FYI: I am attempting to characterize a CTs response for very high fault currents (say, 5,000x that which would saturate it, when we are approaching the short time thermal limit- what happens there?),

and i doubt Chan's model will hold up - there's not enough of a memory term in it.
Currents of the magnitude you are considering will impart a "permanent" magnetization to CT's
and there's a niche market in power industry for devices to 'demagnetize'" them.

see http://www.spinlabinc.com/Demag Flyer.pdf
and https://www.researchgate.net/publication/276339546_Demagnetization_of_Current_Transformers_Using_PWM_Burden

for starters.

I hope you (or somebody else interested ) find this and post 'scope traces of flux and current for that ferrite core.

old jim