Multiplexing analog resistance signal?

[Mickey]

Hello.

I'm interested in how can resistance signals (like Pt100) be
multiplexed?

I mean how can I make a analog switch to switch 16 RTDs
to just one output? Three wire Pt100 connection.

For analog voltages and currents amalog multiplexers like
4051 can be used but how about resistance (around 100
Ohms)? (analog multiplexer chips have pretty big resistance)

 

[LameDuck]

Not entirely sure what you're up to, but I'd
venture to suggest that the individual sensors
be buffered by inexpensive opamps. The opamps
might also employ outputs that could be digit-
ally switched into a high-impedance state; in
turn giving the mux effect you're looking for...

 

[Mickey]

Not entirely sure what you're up to

I'm interested in how to switch 8 or 16 Pt100 temperature sensors
to one, so to select one sensor at a time to sample and make A/D
conversion on it.

 

[[email protected]]

I mean how can I make a analog switch to switch 16 RTDs
to just one output? Three wire Pt100 connection.

The Fluke data logger (Hydra 2625) that I've used simply has a bunch
of relays in it. It is essentially a nice digital voltmeter with
a 2 wire, 21 position switch (made out of relays) on the input leads.

You can't take readings very fast with it - you might be able to read
each RTD once a second with 16 connected - but the relays add very
little resistance (a fraction of an ohm) to the input signal.

If you decide to do this, get relays that are made for "dry contact"
(very low current) service, instead of power relays. Power relays
have contacts that are designed to be cleaned by arcs; "dry current"
relays don't depend on arcs for the contacts to work.

For analog voltages and currents amalog multiplexers like
4051 can be used but how about resistance (around 100
Ohms)? (analog multiplexer chips have pretty big resistance)

If the resistance of the multiplexer chip is fairly constant, you can
measure it for each channel and then subtract it out of your reading.
But the chip's resistance will probably vary with temperature enough
to make it hard to correct for, unless you add a 17th RTD for "chip
temperature".

Matt Roberds

 

[Tim Wescott]

Hello.

I'm interested in how can resistance signals (like Pt100) be
multiplexed?

I mean how can I make a analog switch to switch 16 RTDs
to just one output? Three wire Pt100 connection.

For analog voltages and currents amalog multiplexers like
4051 can be used but how about resistance (around 100
Ohms)? (analog multiplexer chips have pretty big resistance)

How much control do you have over the circuit? Most circuits for these
sensors drive a controlled current through the sensor and look at the
voltage. With your three-wire sensor you'd also get half a wheatstone
connection to the sensor, so in theory you can null out the resistance
effects of the sensor leads by calculating the drop in the paired sensor
wire and subtracting it out.

So there's two possibilities:

1: pretend that your switches are matched well enough that the 3-wire
circuit will correct for switch resistance, hook it up and go. When that
doesn't work, use option 2.

2: Go ahead and drive your constant current through the sensor using one
pair of switches. Monitor the return voltages through another three
switches (it's a three-wire circuit, remember?). If you can find a
well-explained 3-wire circuit out there you should be able to see where to
put the switches.

 

[Mickey]

The Fluke data logger (Hydra 2625) that I've used simply has a bunch

of relays in it. It is essentially a nice digital voltmeter with
a 2 wire, 21 position switch (made out of relays) on the input leads.

Thanks Matt. We have in a company where I work analog multiplexer for
32 channels, done with relays, but the relays are for Ex (Explosively)
dangerous enviroments, and it is a pretty large box, so I thought it could
be made not so large for not difficult enviroments. I also thought it could
be made without relays with something more suffisticated, but I guess
not relays and thats that.

Thanks.

You can't take readings very fast with it - you might be able to read
each RTD once a second with 16 connected - but the relays add very
little resistance (a fraction of an ohm) to the input signal.

Time is not critical.

If you decide to do this, get relays that are made for "dry contact"
(very low current) service, instead of power relays. Power relays
have contacts that are designed to be cleaned by arcs; "dry current"
relays don't depend on arcs for the contacts to work.

Thanks for this suggestion.

If the resistance of the multiplexer chip is fairly constant, you can
measure it for each channel and then subtract it out of your reading.
But the chip's resistance will probably vary with temperature enough
to make it hard to correct for, unless you add a 17th RTD for "chip
temperature".

The resistances aren't paired (or simmilar betveen contacts) so it is
going to be impossible to use analog multiplexer chips.
I did think of converting resistance signal to voltage and then the
analog multiplexing chips resistance wouldn't interfere much. But how
to easily convert resistance to voltage for 16 resistances (3 wire)?

 

[Mickey]

How much control do you have over the circuit?

I don't know what you ment but this, all the control I want I guess.

Most circuits for these
sensors drive a controlled current through the sensor and look at the
voltage. With your three-wire sensor you'd also get half a wheatstone
connection to the sensor, so in theory you can null out the resistance
effects of the sensor leads by calculating the drop in the paired sensor
wire and subtracting it out.

If the resistance of wires is the same for all 3 wires, and yes that is the
point of using 3 wires for measuring resistance.

So there's two possibilities:

1: pretend that your switches are matched well enough that the 3-wire
circuit will correct for switch resistance, hook it up and go. When that
doesn't work, use option 2.

Well, pretending works pretty badly in engineering stuff and I think that
Murphy's law can apply here in case lets say when you have 2 possibilities
for doing something, one is slightly easier it is probably the other one
that
is the right one ))
In this case it would be a shoot in the dark.

2: Go ahead and drive your constant current through the sensor using one
pair of switches. Monitor the return voltages through another three
switches (it's a three-wire circuit, remember?). If you can find a
well-explained 3-wire circuit out there you should be able to see where to
put the switches.

This idea isnot bad, I'll give it some thought. Thanks.