0
\$\begingroup\$

I'm planning to gather data from several Pt temperature sensors:

• Nominal value and tolerance according to DIN EN 60751 / standardized
• For temperatures between -70 ° C and +550 ° C
• Resistance values ​​of 20 ... 5,000 ohms
• Linear characteristic curve
• Quick response
• Good vibration resistance
• Material: Pt-Ni

I'll read temperature from metallic elements (aluminium, inox, copper) 30-50cm away, others n * 3m away (n integer).

However, in order to read about 10 sensors from a Pi Zero or Pico, I'll need some switch that will readout one sensor after the other, and communicate the sensor number at the same time.

How can I achieve this, and how to connect this to the Pi ?

More context:

  • I would like to keep the design reasonnably simple and cost reasonnable, for prototypes, but I'm also planning some industrial production, so I would like also to keep production cost and maintenance costs reasonnable.

  • My priority is exactness (tenth of C/K degree, if difficult I can have half or integer degree) and accuracy in time (latency, momentum); so the cycle period should be reasonnable (about a tenth of second).

  • I'll have temperature ranging from say 5°C to 120°C, and I may have sometimes 10°C shift in less than a second.


After the comments, I thought about combining both 24 bits DAC Microchip MCP3911A0-E/SS, and an 8-way multiplexer Onsemi MC14051B-D.

\$\endgroup\$
18
  • 1
    \$\begingroup\$ Any reason you can't use an analog multiplexer such as this one? Note there is some series resistance involved, but if you'll be amplifying (or buffering) the temperature sensor outputs via op amps, that might work. \$\endgroup\$
    – akwky
    Commented Dec 21, 2021 at 19:41
  • 1
    \$\begingroup\$ It all depends on wire distance, EMI nearby in order to get 60 dB SNR or <5mV noise floor using a 1 mA constant current source using analog shielded twisted pairs with a differential amp to a 16ch Mux.. Pls update your specs \$\endgroup\$ Commented Dec 21, 2021 at 20:39
  • 1
    \$\begingroup\$ Accuracy affects choices in the design , define your signal sensitivity and ambient noise please. read this yet ? jumousa.com/attachments/JUMO/… \$\endgroup\$ Commented Dec 21, 2021 at 20:52
  • 1
    \$\begingroup\$ analog.com/en/design-center/reference-designs/… \$\endgroup\$ Commented Dec 21, 2021 at 21:02
  • 1
    \$\begingroup\$ analog.com/en/analog-dialogue/articles/… \$\endgroup\$ Commented Dec 21, 2021 at 21:05

3 Answers 3

1
\$\begingroup\$

You basically answered your question yourself when you mentioned multiplexing. The main point is that you don't need to identify your sensors if you control the switching between them yourself.

The rest of it is just tuning up your design for your requirements. For example, the output of the sensor and the distance define whether you can multiplex analog outputs directly, or not.

If you cannot, then you may be able to pre-amplify the signals at sensor locations and maybe convert them to differential form, good enough to send over long distance to analog multiplexor.

Finally, if that is not possible, you'd have to digitize sensor outputs in place and use digital multiplexer to do the polling. In fact, if you use ADC with addressable interface, you may not even need a multiplexor.

The bottom line - it all depends on your requirements and environmental conditions. You have options.

\$\endgroup\$
1
\$\begingroup\$

You could use 18B20 probes instead. The original Maxim DS18B20 have ±0.5°C Accuracy from -10°C to +85°C, unique 64 bit codes and read out at a higher resolution.

As far as I remember, I tested up to 8 of the "chinese" versions on 1 meter cable each a few years ago. They could be read out, I did not check their accuracy.

If 15 is too much on two wires, you can always use two ports for the 1-wire protocol.

The advantage is that you do not need external components other than what is needed for the interconnect.

Note:
I just see that your upper temperature requirement is +120°C, the 18B20 can operate up to 125°C, the measurement is less accurate above 85°C.

\$\endgroup\$
1
\$\begingroup\$

You need to decide whether you're using 2/3/4 wires for the sensors (which affects cancellation of wiring resistance and resistance changes) and that will depend on the the absolute sensor resistance which, again, is up to you.

RTDs are high-output devices and your required bandwidth is low so you don't need to take heroic measures, but you do need to keep RF out of the front end or you can see apparent offsets and such like.

I would suggest doing your signal conditioning and filtering ahead of the multiplexer- more parts than trying to mux a raw sensor (since you need to duplicate the parts for each temperature measurement channel) but a much more reliable choice.

\$\endgroup\$
2
  • \$\begingroup\$ What is RTD for ? \$\endgroup\$
    – Soleil
    Commented Dec 25, 2021 at 22:19
  • 1
    \$\begingroup\$ @Soleil RTD (Resistance Temperature Detector) is the common name for the platinum sensors you are using. \$\endgroup\$ Commented Dec 25, 2021 at 22:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.