Archived Forum

No personal experience, but I think I've read somewhere in these forums, that if you're building your mid-rail with a voltage following op-amp, you can get away with one for all the sensors, but if you're building it with R dividers, you need one per sensor, otherwise they interfere with each other.

Indeed. Robin Emley uses the 'op-amp voltage follower' in his diverter.

Thanks

can you just put a diode after the voltage divider for each CT sensor?  Here's is what mine looks like:

I see no diode. What you have there isn't any better, indeed it might be a little bit worse, than the standard arrangement. There's no advantage in having two 10 μF capacitors against a single 22 μF one, you might still get some crosstalk especially if one CT is seeing a high current and the other one sees only a small current, and with the double-size capacitor, it will take twice the time to settle from switch-on.

Here's the diode idea.. originally I thought the capacitor was for the CT sensor, not the input voltage..

The capacitor is there to provide a low impedance to ground for alternating voltages and currents, effectively anchoring the midpoint and supplying or accepting the charge taken or dumped by the sample-and-hold capacitor as the multiplexer switches. 

I still don't see the point of your modification. All you're probably doing is dropping the input range to about 700 mV rms, so the max current will be 21 A rms or thereabouts. That's because the capacitors will only charge to around 1 V, so the input can only swing 1 V negatively before it hits the GND rail. The capacitor voltage might just creep up above that, but whether it does depends on the balance between the diode current at whatever forward voltage it has, versus the capacitor leakage current.

You're saving the cost of 2 resistors and adding the cost of 2 diodes. Is that a saving?

I guess I'm not understanding how adding a second voltage divider network off of the same 3.3V rail would reduce any noise/crosstalk between the two CT sensors.

I guess you're right there. You need to draw a diagram showing what you have above, then extend it to the right and draw in the input pins, the multiplexer, the sample & hold capacitor and the input of the ADC (see the data sheet for details of those components). Then figure out, as you change channels and the switches in the multiplexer operate, how the charge gets on and off the S&H capacitor and where it moves from and to, and which voltages change as a result. The worst case is when one channel is sitting at 3.3 V and the next is sitting at 0 V. Remember, your CT is pretty much a short circuit to all this (actually, it will be about 66 Ω - the burden, in parallel with the secondary winding inductance and resistance). 

There was a big discussion about this two or three years ago.