Analog switch applications

[mikael1985]

Hello,

I have been thinking about making a thermocycler project design for some time, and for that I want to have several temperature Measurements at high resolutions (using thermistors) at various sites.

I have been thinking how this project could be made cheaper. One thought was to use a single 12 or 16bit ADC and then gate the different sensors via an inexpensive multichannel analog switch (CD4066) to keep cost low. But I'm Little unsure if i actually can do that. (atmega328 or an arduino will be the MCU)

There is some resitance in the channels, which I suppose could be adressed with calibration, but can I use an analog switch for this, and if no can you provide an example where an analog switch is prefered to just using some transistors/mosfets?

nb. the stuff is from ebay as I want it to be an inexpensive and easy open source project

Best
Mikael

[JPortici]

What i do in this case:
Provided that for best accuracy and faster sampling frequency you have to buffer the signal before going into the ADC..
- If i have a spare op amp or can add another, i just buffer the output (common pin). In a non inverting configuration the input current is rear zero so the voltage drop across it is negligible (at least for 12 bits resolution). Then antialias filter.
- If i don't have a spare opamp i just go into the antialias filter. by playing with the resistor values you can recover the original amplitude, check the datasheet for the ron values variations in terms of Vin.

[Giaime]

Thermocouples or NTCs?

Professional multichannel switches (e.g. Keithley 2700 we use at work) use mechanical relays 

[mikael1985]

>JPortici
phh, you put me to Work , I will read up the antialiasing part and see if I can grasp it and convert it to an functional design. But it seems to be method for limiting the signal range, so only useable Measurement are within the voltage range?
but I havent really educated myself  here, but I can use an opamp to amplify my reading from the NTC, limit the signal and then use a cd4066 to multiplex the signals to the ADC?

>Giaime
I would use NTC, as they are cheaper than optocouplers.
Relays are they used because of their low resistance?

[Giaime]

>Giaime
I would use NTC, as they are cheaper than optocouplers.
Relays are they used because of their low resistance?

Hi! ok, NTCs are much easier. They're just resistors  I would check with the final circuit that you come up with if the on-state resistance of the optos is low enough not to interfere with the temperature measurement. A rule of thumb could be: if the resistance error induced by the opto is equal or less of the initial tolerance of the NTC (or, if it is worse, the accuracy of your ADC), then I would not worry about it.

You can easily calculate the voltage drop across the on-resistance of the opto by accounting the input impedance of your measuring circuit (ADC, buffers, etc...).
I haven't done the math but the usual 50-100ohm opto on a 10k NTC should be fine, if the measuring circuit is >1Meg input impedance.

[danadak]

Consider a Cypress PSOC, you get ARM CPU + OpAmps + A/D + DAC + Mux +....tons of other
analog and digital. Low costs, like PSOC 4 family, tools very GUI oriented. And routability
like FPGA. Dev boards sub $ 25, several < $ 10.

Just a thought.

Attached is a list of components (read internal) available in PSOC.

http://www.cypress.com/products/32-bit-arm-cortex-m-psoc

Take a look at this thermistor ap note for PSOC. Shows several muxed designs.

http://www.cypress.com/file/45091/download

Project files here

http://www.cypress.com/documentation/application-notes/an66477-psoc-3-psoc-4-and-psoc-5lp-temperature-measurement



Regards, Dana.