Thanks for the info. I still can't justify the overpriced
transducers. I was looking at analog.com at some of their power
metering IC's like the ADE7757. Are there other products like this to
simplify getting the info into some form analog signal or
communication protocol like RS232?
If measuring the average absolute value of the waveform is acceptable
(that is, if you don't need "true RMS"), and if the input resistance
of the PLC's 0-10V input is high (say >100kohms), there's an easy way
to do this, and the accuracy will be quite good, even down to currents
under an amp, for 20A full scale.
You need to get current transformers (CTs) that can output a fairly
high voltage--at least 12 volts RMS. Expect these will have a high
turns ratio. For example, the 2000:1 CR8350-2000 from CR Magnetics
(thanks, Martin, for the link) should do the job. The catalog page
lists the effective turns ratio, Te, as 2037:1. We'll be detecting
the average of the absolute value of the waveform. Assuming it's a
sine, the average is about 0.900 times the peak (2*sqrt(2)/pi). We'll
put a resistive load on the CT, such that 20A RMS sinewave in the
monitored wire results in 10.0V average magnitude across the
resistor: 10V*2037/(20A*0.900) = 1131 ohms. Actually, we'll make it
a bit more, and add a way to trim it down to the right value to
calibrate it: say 1.21k ohms 1%, paralleled with (10k fixed in series
with a 20k trimmer). We won't connect that load directly to the CT
output, but rather through a bridge rectifier. There is some
advantage to using Schottky diodes, but it's minor, and you can just
as well use 1n4148 or similar silicon diodes. The resistive load goes
across the bridge output. Since the CT looks like a current source,
the diode voltage drop is unimportant, so long as it's not so large
the CT can't put out the voltage. Now we just need to average the
output; what's across the load is the absolute value of a (nominally)
sine wave. We could just put a capacitor across the load, but that
requires the CT output to slew quite a few volts rapidly; better to
isolate the capacitor with at least a resistor, and even better, an
inductor. For the very low DC current involved, you can use the
winding of a small audio transformer to get high inductance, tens of
henries for a nominal "20k ohm" winding. Then use something like a
47uF cap at the output to smooth out the voltage. You may want to add
a little resistance in series with the inductance to reach critical
damping; then the response of the circuit will be reasonably "crisp"--
much more so than if you just use a resistor instead of the inductor.
After construction, calibrate it: put a known 20A through the
monitored wire, read the voltage on the PLC, adjust the trimpot so it
reads 10.00V. Now reduce the current to 10A, and make sure the PLC
reads 5.00V. Check the linearity further by putting in, say, 1A, and
checking for 0.500V from the PLC. If it's not as linear as you'd
like, consider applying a calibration in the PLC. But with the
rectifier connected to the CT and the load to the rectifier output,
you've removed a significant source of nonlinearity.
Cheers,
Tom