2
\$\begingroup\$

DAQ Wiring

Hi all, I'm trying to build a portable data acquisition system to gather readings really close to high voltage powerlines. I was hoping to get some advice on how to achieve the best noise immunity. Here's my current plan:

I have a sensor with 4 analog outputs and I was thinking of running 4 shielded twisted pair wires from the sensor to the DAQ running in differential input mode. They all share the same ground so I was wondering if running 4 separate ground wires makes sense or if it's better to just run 1 ground and do everything single-ended and not use twisted pairs? My thinking was that then each pair would pick up the same noise on its ground and reading wires and therefore would cancel the noise. In addition, I will build a Faraday cage around the DAQ. Does this seem like a good plan?

Here's the datasheets for the sensor and DAQ in case: https://aerospace.honeywell.com/en/~/media/aerospace/files/datasheet/3-axismagneticsensorhybridhmc2003_ds.pdf

http://www.ni.com/pdf/manuals/378030c_02.pdf

Thanks!

\$\endgroup\$
  • \$\begingroup\$ Close to a high voltage power line is a good place to find strong magnetic fields that might interfere with what ever it is you really want to measure. \$\endgroup\$ – JRE Oct 8 '18 at 18:04
  • \$\begingroup\$ The magnetic fields are actually what we are measuring which unfortunately puts us in the worst case scenario for noise! \$\endgroup\$ – Alex Colbourn Oct 8 '18 at 18:26
  • \$\begingroup\$ The easiest way to deal with noise in connecting wires is... get rid of the wires. i.e. make them as short as possible. I did not see any distance requirement in your question. To me it means you can place sensor and DAQ really close to each other. Couple centimeters shielded twisted pairs with shields connected to DAQ shield should not have much noise, IMHO. \$\endgroup\$ – Maple Oct 8 '18 at 18:30
  • \$\begingroup\$ Also, it would be nice if you define "extreme noise". What frequency? It is a white noise or just a mains hum? And if it is magnetic field you are measuring then any magnetic field fluctuations are not a noise, are they? \$\endgroup\$ – Maple Oct 8 '18 at 18:38
  • \$\begingroup\$ @Maple Alex mentioned power "high voltage power lines". The largest noise in that case is 50Hz or 60Hz hum. The lousy news is that there may be both E-field coupling (high voltage) and H-field coupling. There may be powerline communication signals too (100kHz to 200kHz in the US). Alex, please confirm or challenge. \$\endgroup\$ – Nick Alexeev Oct 8 '18 at 19:37
3
\$\begingroup\$

There are two sources of noise, capacitive and inductive.

Capacative noise comes from a voltage potential difference between two conductors. In your case the power line would be one conductor and your device would be another, and can be modeled as a capacitor. To stop capacitive coupling, a faraday cage would be the ideal thing to stop any electric field that would produce a voltage on your sensors, cables and DAQ.

The ideal situation would look like the image below example c) The other problem is also grounding as the potential difference between your sensor and DAQ will most likely be at two different potentials. So try and make the whole system like example c) with shielding around the cable, sensor and DAQ. The shielding also needs to be continuous, so use shielded wire.

enter image description here
Source: http://www.7ms.com/enr/online/2010/02/notebook.shtml

Inductive coupling is harder to stop, it comes from magnetic fields and cannot be completely eliminated but can be reduced. Avoid loops as a magnetic field can create a voltage potential around the loop. Twisting wires also helps eliminate loops as it cancels out the magnetic field between loops.

enter image description here

Source: https://www.researchgate.net/post/What_is_the_basic_idea_behind_the_twisted_pair_Why_are_the_two_wires_twisted_How_does_this_arrangement_compensate_undesirable_disturbances

\$\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.