1
\$\begingroup\$

I have read about distance limitations of timing critical signals like communication over SPI. In order to get more distance between my Pi and my sensor I am considering using an analog sensor and an ADC.

Question: What should I know when considering the distance between my analog output and my ADC input? I am, at the moment, considering the impact on:

  • Linearity of sensor output and ADC readings
  • Delay between sensor output and ADC reading

Details:

  • Sensor output voltage: 0.25 to 4.75 V
  • ADC input voltage: 0 to 5 V
  • I hope to sample the sensor output at 50 kHz (seems a bit fast and too fast for SPI over a 1m distance)
  • I'm not really sure how to calculate the amperage the ADC will draw from the sensor output.
  • Sensor.
  • ADC.

Insight and references are welcome!

Edit:

I think a minimum of 8 significant bits is what I can work with. With a 500 unit range that's a measurement of 1/(2^8*500) = ±2 units.

\$\endgroup\$
  • \$\begingroup\$ Delay is not significant. Noise is a more serious problem. What is the output impedance of your sensor? How many usable dB/significant bits do you want? \$\endgroup\$ – pjc50 Jan 19 '17 at 20:44
  • \$\begingroup\$ (essentially these are the same problems as faced by audio engineers with microphones on long cables, for which balanced XLR is preferred) \$\endgroup\$ – pjc50 Jan 19 '17 at 20:46
  • \$\begingroup\$ Interesting, I will read about the audio analogy. I think I need 8 significant bits and I'm trying to garner the output impedance from the datasheet. \$\endgroup\$ – nate Jan 19 '17 at 20:58
  • \$\begingroup\$ Cable propagation delay is around 6ns/m -- so not an issue. 5V / 256 (8-bit) = 0.02V of resolution -- should be doable using common good practice. Now consider SPI -- 50kHz sampling rate probably requires around 1Mbps or higher SPI rate, that is 1us cycle time, definitely not too fast. Even using normal 3.3V/5V logic, noise margin can be seen as being in the 0.3V range, that is 10 times higher than the analog margin (0.02V). Follow good practice and series terminate the lines, I wouldn't rule SPI out. The real disadvantage is the need to supply power to SPI device. \$\endgroup\$ – rioraxe Jan 19 '17 at 22:53
  • \$\begingroup\$ Thanks both, for the comments. Could you provide either some links to relevant material when considering voltage fluctuations on my signal wire - preferably including the factors that influence this. Otherwise a few more terms to narrow my search (It seems impedance is relevant - is there a theory or law I could start with?) \$\endgroup\$ – nate Jan 19 '17 at 23:29
0
\$\begingroup\$

Modern ADCs charge up a tiny capacitor, and then use 20_questions (binary search) to consume that charge with binary-weighted decisions.

The capacitor is approximately 10pF.

The average current will be FCV, or 50,000 samples/second * 10pF * 5volts.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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