pt100 and 4051

HC4051s are fine. Arrange the circuit so that the switch resistance doesn't matter. You'll need three mux chips.

Here's a 4-wire mux'd RTD thing, similar concept:

R324 is a Susumu thinfilm resistor, and everything is ratiometric off that.

So, your intent is to switch the excitation current in one lead of the RTDs but to leave the other lead's excitation current common between the eight units? That takes an 8-to-1 switch, then you use the other two 8-to-1 switches to probe the third wire and the common node of the unswitched excitation wiring.

There's no need here (unless your voltage sense amp bias current draw is significant) for the switches to match in resistance. Only the excitation wires on the two poles of the RTDs have to be matched (as in all three-wire resistance schemes).

If, rather than leaving one excitation (current-carrying) wire common, you wished also to switch it, the problem of switch resistance mismatch does show up, and the multiplexer has to grow to four-wire complexity (i.e. you'd use four 8-to-1 switches), even though the device wiring stays in the three-wire configuration.

Il 13/01/2013 18:22, John Larkin ha scritto:

I'm trying to fully understand your schematic. One channel of AD7793 reads the mux'd RTD's signals while the second channel measures the voltage drop across R324, that is the current flowing into the selected RTD.

In this way, the Rds_on of any channel of the three 4051 is different doesn't matter at all?

Thanks for your hint Marco

Il 13/01/2013 19:06, whit3rd ha scritto:

No, I was mean to use a switch for each lead. So three 8-to-1 switches.

I will try to draw a schematic and I will post here to be sure I understood correctly.

In your solution doesn't matter, but in mine does.

Yes, now I understand my error.

Thanks Marco

Right. We separately measure the voltage drop across R324, then the drop across the RTD, and do the math. In neither case does the switch resistance contribute any error.

We would get an error if the switch resistance in U237 were to change between those two measurements, from self-heating maybe, but in real life that's not significant. The voltage reference doesn't matter, either, as long as it's short-term stable.

Hardly at all, not enough to matter. Only one switch needs to carry current, and it just needs to be short-term stable.

What sort of RTD current did you have in mind? 200 uA is a little low for my taste... only makes 20 mV across 100 ohms.

Il 14/01/2013 00:01, John Larkin ha scritto:

and

Right, the other two legs doesn't carry any current (ok, it's very very small -> negligible)

I usually power my RTDs with 500 uA. No evidence of self-heating and the voltage is not very difficult to read using a 24-bit converter and a good layout.

In this particular case I may rise the current over 1 mA because I can power each RTD for a short time. I need to get a reading every several minutes.

Thanks Marco