I want to us this sensor(107LN type) as in the following signal chain with differential-input DAQ board and a STP CAT6 cable as below:

enter image description here

Above the GND of the power supply is local/near to the DAQ. Someone told me this will be problem for differential inputs.

So I created a new ground on the sensor X (GND) from B (GND) on the sensor. Now along the transmission to the DAQ, the ground and the signal is twisted together. See below diagram:

enter image description here

Is the second diagram better for common mode noise rejection? Why?

  • \$\begingroup\$ none of the above. until you complete the schematic. Where is Earth Gnd connected? what is Signal level, impedance and BW? What noise exists, how long are cables, what is your required SNR, is there any Anti-aliasing signal conditioner ( Nyquist Filter) ? \$\endgroup\$ Commented May 18, 2018 at 5:38
  • \$\begingroup\$ What is that to do with anti-aliasing? \$\endgroup\$
    – floppy380
    Commented May 18, 2018 at 5:39
  • \$\begingroup\$ where does Diff Amp go? What environmental noise exists? \$\endgroup\$ Commented May 18, 2018 at 5:48
  • 1
    \$\begingroup\$ the question is asking about something else. \$\endgroup\$
    – floppy380
    Commented May 18, 2018 at 5:50
  • \$\begingroup\$ 100 dB CMRR Diff Amp does nothing if the lines are unbalanced impedance lines in the presence of high EM fields. especially on 500m of cable. I may look again tomorrow but you need to answer the other questions \$\endgroup\$ Commented May 18, 2018 at 5:51

1 Answer 1


The second diagram is better because the differential inputs don't share a wire with the power supply feed providing 0 volts to the sensor. That 0 volt feed will take current and, in the simple DC scenario will cause a volt drop that is an unknown value - that DC volt drop adds to the signal measured: -

enter image description here

X = unknown volt drop.

Your second diagram is better because the unknown volt drop (x) is disregarded by the differential amplifier.

  • \$\begingroup\$ Thank you sooo much for this! I really cannot find such info even in books. One more small thing, I also have a sensor with two signal outputs. SA-207 crlsensors.com/prodDocs/sa-307ln.pdf It is exactly the same sensor but two DC outputs/axis. In that case better use a separate signal ground for each signal? (X and Y twisted with their own ground) \$\endgroup\$
    – floppy380
    Commented May 18, 2018 at 12:45
  • \$\begingroup\$ That's the safe route and, if you are using twisted pairs, that is probably also the common-sense route. \$\endgroup\$
    – Andy aka
    Commented May 18, 2018 at 12:55
  • \$\begingroup\$ Cat5,6 PHY network CM hybrid transformers use at least 1nF to connect floating signal ground to earth ground to avoid line noise but shunt RF noise. Since user1234 did not answer my questions, this may be better but still sub-optimal. \$\endgroup\$ Commented May 18, 2018 at 17:59

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.