May 2014
I want to read and log temperatures from multiple points in my (unnecessarily complicated) central heating system. I want to use thermocouples because: 1) they are easy to attach to my hot water cylinder (stabbed under the insulation); and 2) I already have them — from IKEA meat thermometers.
I would like to select each thermocouple in turn, use a single amplifier/analog-to-digital stage (Max31855 ??) to obtain the reading, and move on to the next thermocouple.
I need ideas for switching between the different thermocouples without introducing unwanted switch resistances, etc. Would an array of reed switches driven by a shift register be a sensible solution? Is there a better solid-state solution?
Please log in to post an answer.
Besides solid state relays you can just use an analog mux/demux chip (that'll cost about $1). The mux/demux will take 8 inputs and, based on the address pins, route one of those inputs to the output. (Or, you can run it backwards and take one input and route it to one of eight outputs). Something like a MC74HC4051 or HEF4051B.
The challenge may be the fairly low voltage and current from the thermocouples, aggravated by long signal lines. You might need to put amplifiers at the thermocouples. Or, for a couple bucks, replace the thermocouples with sensors much better designed for remote sensing over narrow temperature ranges near room temp. Something like LM34DZ.
First, people have a misconception that the junction of dissimilar materials form a thermocouple. Actually, the junction creates a return path for current that flows due to the Seebeck effect, discovered in 1822. In essence, a temperature difference between the ends of a conductor creates a small EMF, or voltage, between the ends. A type-E thermocouple (constantan and chromel wires) has a 68 µV/°C coefficient, which means you'll see a 680 µV increase for a 10°C increase in temperature between the ends of the thermocouple wires. Of course, you need to know the temperature at the switches you use in your measurement systems. People often refer to this as cold-junction compensation, and many diagrams show an ice bath at 0°C. You just need an accurate temp sensor, not an ice bath.
Second, I recommend reed relays; either two SPST or one DPST relay will do the job nicely. Some people might say, "Wait, all the connections to the relays create individual thermocouples." Not so, the difference in voltage between the ends of the relay conductors, and associated conductors, amounts to a negligible amount. Just keep your equipment at a fairly stable temperature. Depending on the ADC you use, you might need an accurate instrumentation amplifier between the thermocouple signal from the relays to the ADC. Again, the measured voltage results from the temperature difference between the end of the thermocouple and the temperature measured at thermocouple wire ends in your measurement system.
For more information, I recommend the free ebooks, "Switching Handbook" and "Low Level Measurements Handbook," from Keithley Instruments: [url=http://www.keithley.com/knowledgecenter]http://www.keithley.com/knowledgecenter[/url]. Also, see the article at: [url=http://tinyurl.com/kuvd93o]http://tinyurl.com/kuvd93o[/url].
You can use a solid state relay. They are available in DIP packages of up to 8 switches (I believe) and can be controlled by your microprocessor.
