Archived Forum

Depending on the laws where you live, that one might require a certified/registered electrician, and thus be very costly to fit (depending of course on which wire(s) you will be fitting it). Other than that, it looks as if it should be usable. You should get 1 V rms at 80 A with the recommended 10 Ohm burden, which is fractionally low for an emonTx but certainly usable (and higher than you get with the standard 100 A split-core one at the same current with its 18 Ohm burden).

Bear in mind that as it has a steel core, not ferrite, you MUST make sure you never operate it without a burden, otherwise - besides the danger of a flash-over - you will need to demagnetise it to restore its performance.

Depending on the laws where you live, that one might require a certified/registered electrician.
I'm resident in Norway, so the above mentioned rules does apply. (But that's not a problem)

Bear in mind that as it has a steel core, not ferrite, you MUST make sure you never operate it without a burden
Thanks for pointing that out, I failed to see that myself!
Is it enough with an appropriate resistor between S1-S2 as a safety/fallback burden?

Might this be a better choice?
T22-100 (http://www.yhdc.com/english/productshow.asp?id=202)

Obviously the TA22-100 is somewhat larger and has a higher VA rating because it can develop more voltage, so you would have a wider choice of burden resistor - any value up to 50 Ohms - to give you the output you require at the current you want to measure. I'm sorry if that sounds nebulous but I don't know what your maximum current is nor anything about the input you are feeding into.

If there is any risk of detaching the burden resistor, then it is best moved and soldered directly across the terminals, alternatively (or as well) you could put a pair of zener diodes at the c.t. terminals to clamp the voltage at a little above the maximum peak voltage that you expect. If you do move the burden to the c.t and connect with a plug (as in the emonTx), then you could replace the 18 Ohm resistor on the pcb with a high valued one (say 18 KOhms) which would make no measurable difference to the calibration but would leave the analogue input biassed correctly if the c.t. were to be unplugged.

Thanks for your help so far. I've ordered one of each for further testing/tinkering.

I'm sorry if that sounds nebulous
Not at all. It's a general answer to a general question ;)

I don't know what your maximum current is nor anything about the input you are feeding into.
Probably maybe possibly the wrong forum thread to elaborate on that one, but in the absence of a "My Project" thread, here goes:

I'm in the process of renovating my house, including the electrical wiring. I'm also changing my fuse box. I was lucky enough to get a couple of these ones for free: "Schneider Prisma plus G" (floor standing type)
This gave me loads of free space to play with.
There will be some 16-20 circuits, and I intend to monitor each one separately + incoming supply, delivering fine grained data to my home automation system. To do this, I'll need to make a custom shield to accommodate all the CT's.
The circuits will be 16A@230V with the exeption of a couple of 25A ones. Intake breaker is 63A

Don't worry about this thread being inappropriate.

Now, I have a better picture of what you are doing. Because you are starting from new, I would certainly suggest you get lower rated c.t's for the outgoing circuits. The Atmel processor ADC is 10 bits wide, which basically means it can measure with reasonable accuracy down to 1% of full scale (10 counts peak-peak). So an 80 or 100 A c.t. is not the best choice for a 16 A circuit! (Though OK for the incomer).

You need to think about the input voltage you need and the voltage that your ADC will run at. If you have a 5 V supply and use 5 V for the ADC reference, then for maximum resolution you need 5 V p-p at max  current (16 A). The problem here is small c.t's have small VA ratings and distort the wave shape when they saturate, which happens if you increase the resistance of the burden to get more voltage. You can run at 3.3 V like the emonTx, or run the ADC reference at 1.1 V like Atmel's application note AVR465: Single-Phase Power/Energy Meter with Tamper Detection. You will have a wider choice of c.t. if you have a 1.1 V p-p input, because there are many cheap c.t's designed to give a voltage output of 0.33 V rms. And of course you do not need split-core c.t's, so there is yet more choice. 

If you design your own shield, you must take great care to separate the analogue and digital circuits.  There is some evidence that people who have used prototype pcb's read powers of up to about 50 W when no current is flowing, due to noise pickup.