Current transformers: is this the expected behavior?

Discussion in 'General Electronics Chat' started by someonesdad, May 18, 2011.

  1. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    I have a box with a Variac in it. I want to put in some control circuitry that turns the AC power to the Variac off if the current gets above an adjustable threshold. Thus, I need to measure the AC current.

    One method is to use a low side shunt, but I don't like that because it's not insulated from the line. I suppose a small isolation transformer could be used, but I haven't tried that yet (and I don't know how well it would work, as the voltage dropped across the shunt would need to drive some current through the transformer, dropping the shunt voltage).

    The method I first decided to focus on was to use a current transformer. I've not used one before; hence my question (I'm a hobbyist, not an EE).

    I want to measure AC currents between 0.1 and 10 A. To do this, I put around 49 turns of wire around a small ferrite toroid that was in my junk box (see attached picture). I ran the AC line through the toroid and measured the output voltage with a scope (see attached waveform, which was for a current of 8.6 A(rms)). I was surprised to see this nonlinear waveform when I expected a sine wave (the load was a resistive heater with a power factor of 1).

    My question: is this the normal behavior for current transformers? Here, I'm referring to this non-sinusoidal output for a sinusoidal input current.

    My suspicion is it's not because I can tell by the peak's phase shifting as a function of current that this is likely due to the shape of the BH curve of the ferrite. From the little I've read, typical current transformers are made with steel toroids, not ferrite.

    Second question: would you recommend I just use a commercial current transformer (I found one at Digikey for under $10) instead of trying to make my own?

    I like to make my own stuff, so if you think I could e.g. take a stack of steel washers and make a suitable current transformer, lemme know...

    BTW, I characterized the 0 to peak voltage as a function of line current (see attached graph) and things are crudely linear from 1 to 10 A or so. Thus, I might be able to use some kind of sample and hold circuit to generate a DC voltage needed to compare with the trip current pot.
  2. Adjuster

    Well-Known Member

    Dec 26, 2010
    The waveform suggests that the core is saturating at really quite low current levels, so that all you see is a brief spike induced in the secondary whenever the field in the core reverses. There are probably two issues here: one is the nature of the core, which probably would be better built with metal laminations (or perhaps a tape wound toroid).

    The other point though is that current transformers are normally operated into a relatively low load impedance. This helps to avoid saturation by reducing the degree of magnetisation, but also is normally required for safety. This is important because commercial current transformers normally have fairly big ratios and can develop hazardous output voltage if not loaded. Note that even your transformer was putting out 28V spikes!

    You might like to try the effect of loading your secondary coil into a low resistance, say a few tens of ohms, to see if you could get a more linear result, but on the whole a commercial job could be a better bet. Don't forget the secondary loading though, never open-circuit the secondary, rather, short-circuit if no load is available - and yes, before you ask this is exactly the reverse of what would be safe with a voltage transformer. Otherwise, you may harm the transformer or create a fire / shock hazard.
  3. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    Thanks, Adjuster -- I'm aware of the hazards of an open secondary, but that's mostly for commercial stuff monitoring tens to thousands of amps. I should have mentioned that I did try loading with a resistor (I connected my GR resistance box across the output of the coil), but it made no qualitative difference except to reduce the signal into the noise at low enough resistances.

    Based on your advice, I think I'll expend the effort needed to wind myself a coil with some flat steel washers because I can have that on-hand for an experiment this afternoon rather than wait for some ordered on line. Normally, this wouldn't be a big deal, but with my friend Arthur and the fingers, this isn't as pleasant as it used to be. :p

    I'll post the data and photos when they're available.
  4. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    OK, this turned out well -- thanks for the advice, Adjuster.

    I went out into my shop to get some steel washers to make a new toroid current transformer and came across something interesting that I had: a bunch of precision cobalt washers 23 mm in diameter and a 9 mm ID. These were given to me by a maintenance tech where I used to work 20+ years ago. They were for a sputtering machine that we used to sputter a cobalt alloy for magnetic disks and, since we had sold the machines, he was going to toss them out. At the last minute he asked me if I wanted them because I was the thin film process engineer and he thought I might have a use for them.

    Anyway, I took a stack of 6 of these washers and put 56 turns on one. Then I tested it as in the previous tests by putting a wire through the hole carrying an AC current going to a resistive heater. There was a fair bit of electrical noise from the cable, but I used the low pass digital filter on the scope to get rid of that crud and was left with a beautiful signal (see attached scope picture). I plotted the RMS output voltage as a function of input current; I put a 1 kohm resistor in parallel with the coil. The second plot shows the data.

    Cool -- I can use this!

    Now I just need a way to turn that sine wave into a DC voltage proportional to amplitude. While I could buy an RMS chip to do this, I'd rather do it with crud I have on hand (op amps, discrete parts, etc.) -- this will also help save toy money (I've got more time than toy money). The other day I built a precision rectifier with an op amp (straight out of Art of Electronics) and it worked perfectly. I put a 10 uF cap on the output to turn it into DC and it worked pretty well, but there was still a bit of ripple left. I'll add some resistance to it to get a time constant around 100 ms or so. But, the rectifier is only a half-wave rectifier. Can you guys recommend a better way of doing this AC amplitude to DC transformation?
  5. Jaguarjoe

    Active Member

    Apr 7, 2010
  6. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    You're right -- for the money, that looks like a very nice solution!