1
\$\begingroup\$

I have done EMC radiated immunity testing on 4-20 mA analog inputs and I noticed bad reading on these inputs when sweeping around 600-800 MHz at 5 cm at 100 V/m (antenna method).

To improve the noise susceptibility, I am planning to add snap-in ferrite beads around the 4-cable. There are two 4-20 mA sensors inside the cable.

The sensors are shielded and connected to earth on sensor side. On PCB side, the shield is terminated with a pigtail cable which is around 40 cm long (I plan to reduce the length as well).

The cable has 4 wires: +24 V / 0 V / 4-20 mA / 4-20 mA

The 4-20 mA wires are twisted and shielded. The +24 V and 0 V are not shielded as you can see on the picture below. enter image description here

My question is: where to place the snap-in ferrite bead?

  • Position A? 2 ferrites for each wire or one ferrite wrapping the 2 wires?
  • Position B? Around the whole cable?
  • Do I need to add one around the power supply? The PCB is supplying the sensors.
\$\endgroup\$
6
  • \$\begingroup\$ Why are there sensors inside the cable? If you are testing and get bad results (whatever that means), why don't you just do another test to establish what makes a decent improvement? \$\endgroup\$
    – Andy aka
    Commented Apr 11 at 13:41
  • 1
    \$\begingroup\$ Where exactly does the current in the two loops flow? The ferrite should be at a position where the currents sum to zero. \$\endgroup\$
    – CL.
    Commented Apr 11 at 13:49
  • \$\begingroup\$ Sorry that was not obvious. On the left hand side there are two sensors connected to the cable that transmit two 4-20mA signals to the PCB board (connected to the right hand side of the picture). The current returns through the GND connection. I edited the question. \$\endgroup\$
    – JeanB_01
    Commented Apr 11 at 14:04
  • 1
    \$\begingroup\$ A shield connected on only one side is ineffective at HF. In the IEC 61000-4-3 standard, the test for commercial equipment is level 2 at 3V/m, the industrial test is level 3 at 10V/m. Are you sure about 100V/m? \$\endgroup\$
    – Vincent
    Commented Apr 11 at 16:38
  • \$\begingroup\$ 100V/m is quite high indeed. Are you certain that the sensors themselves can even operate in such an environment? How are they shielded, is that something you added? It's peculiar they are earthed; have you done conducted tests yet, and if so at what level? \$\endgroup\$
    – Tim Williams
    Commented Apr 11 at 16:49

1 Answer 1

Reset to default
1
\$\begingroup\$

Pigtailing a shield isn't exactly the best way to reduce noise... however:

Most probably the UHF noise is entering in common mode on the cable (maybe through the shield) and rectifies inside your analog frontend. Another possibility is that the noise enter into the supply line of your sensor and the sensor itself is disturbed from the supply.

In my experience the second option is extremely rare (unless the sensor has design issues). The reasoning for ferrite placement is as follow:

  • The ferrite is essentially a resistor at its designed frequency range; following the signal path before the ferrite there is noise, after the ferrite there is low noise;

  • This is usually common mode noise; imagine the noise entering from the side of the cable and propagating in both directions (the middle is a simplification but usually you don't know where exactly the noise enters), something like every some centimeter some noise crawls in;

  • If you place the ferrite in the middle you don't really filter much (since noise has a lot of space to enter on both sides); if you place it near your input you reduce the noise coming from the cable (since there's almost no cable left toward the input side);

As an aside the ferrite works both way: it also keep your circuit noise in and avoid radiation from the cable itself. The best way would be to have the shield almost up to the circuit and put the ferrite there, so every conductor is covered; if you place it in your B position the whole section up to the board is susceptible (and then you'll really need to know where the noise is coming in); in A position the signal is OK but you risk radiating thru the power lines. Ideally the shield would be terminated to the 'junk ground' (i.e. the chassis) near the board (but then everything depends on enclosure, ground and bonding strategies and so on).

That said: 100V/m is quite a big electric field, you usually test at 10V/m unless you have some special requisite. I'd check the filtering on your amplifier, usually ceramics are quite effective (but then, 100V/m are a lot)

\$\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.