6
\$\begingroup\$

I may have an issue with conducted EMI through the ground connections between devices in a system. There is specifically a device which clearly creates noise on the measured signal from another circuit with a scope (noise is reduced when it's off). That scope has to be connected to both that noisy device and the measured circuit so I'm suspecting HF noise is conducted through the ground.

I am thinking about putting ferrite beads on the ground connections for example on the ground connection from all the devices (especially the noisy one) to the scope, but that layout looks like odd to me. Does it make sense to have a ferrite for the ground lead of a coaxial cable? Is there any other way?

Additionally, the scope is directly connected to the measured circuit, which is a resistor with one grounded lead (It's a current consumption measurement). Another approach I'm thinking about is a differential amplifier, does this approach looks better? Can even be necessary a combination of both?

\$\endgroup\$
  • \$\begingroup\$ You got some good general answers. I think however you would benefit from making the question clearer. Maybe make a drawing. And clarify if you mean "board" when you say "system"? Often system refers to something consisting of multiple pieces of equipment. I don't understand your question well enough to give you a good answer. \$\endgroup\$ – Rolf Ostergaard Nov 27 '13 at 17:02
  • \$\begingroup\$ I found a discussion here: electrondepot.com/electrodesign/… My idea is that this noisy device does not carry the signal, it only receives a trigger pulse but in return generates all that crap so I think I don't need to care about it's ground perfectness as far as it's working, does this make sense? \$\endgroup\$ – Joan Nov 28 '13 at 11:41
8
\$\begingroup\$

Increasing the impedance of the system ground between various devices is a bad idea. Decouple the power supply feed points, filter signals between blocks as necessary, but keep the ground connections as solid as you can.

If you do manage to decouple the ground, like by adding inductance in series, then the noise will be even larger on all the signals coming out of that block. By allowing the ground to float at high frequencies, you make everything common mode noise, so it will show up on all signals too.

\$\endgroup\$
  • \$\begingroup\$ I think you are missing a point here, a ferrite is not an inductance, they behave different, please refer to the link in the comment above. And in addition, as I put in tthe comment above, that device does not send/ receive analog data. \$\endgroup\$ – Joan Nov 28 '13 at 11:43
  • 1
    \$\begingroup\$ @Joan: No, ferrite beads most certainly are inductors. That's their point. It is this inductance that adds impedance that causes "decoupling" at high frequencies. Your question makes no sense at all if you don't intend these "ferrite beads" to be series inductors. \$\endgroup\$ – Olin Lathrop Nov 28 '13 at 15:56
  • \$\begingroup\$ Even a straight conductor, even a via, you know, has its inductance. Said this, everything is an inductor plus something else. The main purpose and I think difference when using ferrites instead of loops of wire is that they become lossy at high frequencies (inductors don't, just inductive). If ferrite is for you synonim of inductor, you should fix the entry for "ferrite bead" in wikipedia then. \$\endgroup\$ – Joan Nov 28 '13 at 16:04
  • 1
    \$\begingroup\$ @Joan: Ferrite beads, also called "chip inductors", are one type of inductor. They do usually have a relatively high resistance compared to their inductance, and the lossiness is actually useful in many applications. However, these things are still first and formost inductors. No, "ferrite bead" is not a synonim for inductor. All ferrite beads are inductors, but all inductors are certainly not ferrite beads. \$\endgroup\$ – Olin Lathrop Nov 28 '13 at 16:19
  • 1
    \$\begingroup\$ A more precise definition of ferrite beads would be very low-q inductors, one's that have an unusually large resistive region centered at their cross over frequency. They still behave as inductors at frequencies below this region, and behave like capacitors at frequencies above. They generally don't have a huge amount of inductance (1-2 µH or so) and for their intended frequency range, will dissipate the energy as heat instead of radiating like an inductor/antenna would, but as Olin Lathrop correctly answered, it is unwise to ignore their otherwise typical inductive behavior for lower freqs. \$\endgroup\$ – metacollin Jun 9 '17 at 1:49
2
\$\begingroup\$

It sounds like your problem may be common mode noise. Typically common mode noise flows equally through the + and - rails on your circuit and back through a ground path. Putting a ferrite bead in the ground connection won't help against this kind of noise. What you need is a common mode choke, which is an inductor with two windings that have the same polarity. Try connecting this choke between your supply and your circuit.

\$\endgroup\$
  • \$\begingroup\$ That may be or may be not. I'm not sure what's the return path. If the noise goes out through the ground it may return through the wire (then it'd be differential mode) or somewhere else (then common mode) \$\endgroup\$ – Joan Nov 28 '13 at 11:46

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.