11
\$\begingroup\$

What's a canonical method for reducing / eliminating electrical noise induced in ADC from a small brushless DC fan? For reference below, I'm also using this wall-wart and this 5V linear regulator.

My baseline system looks like this:

enter image description here

I have some options in my head and I'm wondering what the best one might be.

My first idea is this [OPTION A]:

enter image description here

My second idea is this [OPTION B]:

enter image description here

My third idea is this [OPTION C]:

enter image description here

There are probably other options that I haven't thought of, as well as permutations and combinations of these options. What are the merits and drawbacks of these options? What option would you choose and why?

\$\endgroup\$
9
  • \$\begingroup\$ All of your pictures look fine solutions BUT you'll nearly always get a problem if you don't implement correct PCB or wiring practices. You have to ask yourself why you are getting noise and how it is getting to where it does. These pictures don't help anyone answer. The most important thing you haven't mentioned is how you proved the ADC noise came from the fan - this is vital to uncovering a decent solution. \$\endgroup\$
    – Andy aka
    Commented Sep 25, 2013 at 14:38
  • \$\begingroup\$ @Andyaka This blog post blog.wickeddevice.com/?p=423 documents in gory detail how I determined the fan was causing a problem in my system. I know routing matters too, feel free to talk about that in an answer! \$\endgroup\$
    – vicatcu
    Commented Sep 25, 2013 at 14:44
  • \$\begingroup\$ I don't think you should expect best answers without providing this info directly into the question. \$\endgroup\$
    – Andy aka
    Commented Sep 25, 2013 at 14:45
  • \$\begingroup\$ I'd also try putting a small ceramic capacitor across the motor's terminals, as close as possible to the motor. \$\endgroup\$
    – m.Alin
    Commented Sep 25, 2013 at 14:47
  • 1
    \$\begingroup\$ You've done that already because all of your pictures are potential working systems BUT you haven't addressed the vital issue of proving whether the interference is air-bound or conducted through your circuit. Your pictures appear to assume the interference is conducted and this is only half the potential story. \$\endgroup\$
    – Andy aka
    Commented Sep 25, 2013 at 14:51

2 Answers 2

Reset to default
4
\$\begingroup\$

You need to decide how the interference is getting to your ADC input. The best way of dealing with the problem is at the source of the problem. The problem could be: -

  1. radiated EM emissions from the fan - try moving it closer or further away
  2. radiated EM emissions from the wiring to the fan - try moving the wires closer or further away
  3. conducted emissions from the fan/wiring causing power supply problems such as upsetting the ADCs voltage reference
  4. Bad practice on laying out the circuit board such as ground and power currents to the fan also sharing copper with sensitive inputs.

These are just a few and until you resolve what is wrong and where it is wrong, asking for canonical solutions is premature.

There may be several solutions, all of which will work with any of the pictures in your question and these might involve: -

  1. Redoing the PCB layout to avoid shared copper for power and signal
  2. Adding decoupling caps to various places
  3. Using balanced measurement techniques on the ADC
  4. Improving wiring and adding shields
  5. Ferrite beads and clamps.

Work out what is causing the problem first and why.

\$\endgroup\$
3
\$\begingroup\$

Usually fan noise is pretty low frequency so the ferrite bead in option B won't work very well. I tend to use a PTC fuse element with a largish capacitor (100uF+) to ground on the fan side and a ceramic capacitor on the other. The PTC is nice because it's a resistor and it will also provide short current protection, but sometimes you have to use a regular resistor to get enough filtering. You can optimize the values by looking at the power rail with a scope.

\$\endgroup\$

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.