RFI Suppression Techniques over long wires

[Sepiroth]

Hi all,
I would like to have any kind of help to prevent RFI noise over a long wire. In the project im currently working, i've central servo driver card to send PWM signals over a distance which varies 1m to approx. 10m. Also we have 2.4GHz radio in the system, so whenever we use long wires for PWM signals, servo's continuesly glitches, i mean not staying stable. As a first solution I tried to use shielded twisted pair cable with connected to ground at one end, and as another solution, I winded the servo cable around a toroid. Both didn't worked well, whats your opinions and suggestions to solve this mess?
thanks in advance,
Sep

[Dillon]

Try putting a ferrite bead over each wire on both ends of the wire.  This should help.  IF this fails you might try a 10ohm resister in series with each wire lead but on one end only

[pickit2]

I would try a normal cat5 patch cable, or sheilded if you have one.
Also ground all unused wires in the cable at conroller end.

[solutions]

Winding around the toroid will kill off your rise/fall time, making narrow pulses on the PWM next to impossible, so not surprising that one didn't work.  The twisted pair also is not a good idea unless your are sending a differential signal, which you probably aren't.

Given you want to keep bandwidth and have a common mode signal, yet suppress RFI, perhaps a heavy gauge twinax cable might work.  You also need the shield to earth (which is not necessarily "ground") if you're to create a shield for your radios.  If you're on a budget, two coaxes may also work.

[adidav]

In did, to make workable the shield you should put to the "earth" and not to the ground.
If anything isn't work perhaps a differential transmission will work better for you. Also and STP cable will do the job very good. The connectors for the both end must to be with the metallic shield.

[Alessio]

Even if discussion is three month old, I suggest you PROFIBUS:

http://en.wikipedia.org/wiki/PROFIBUS

http://www.sea.siemens.com/us/Products/Automation/Industrial-Networks/PROFIBUS/Pages/PROFIBUS-Connector.aspx

[Sepiroth]

hi all, i constructed a circuit with a highspeed optocoupler, to see if i can transmit pwm signal over a long wire without any precautions. i used something like 5meters cable, connected it to the circuit i've made (6n137 opto with a clean supply). and connected this circuit as close as i can to the servo itself. without any RF signal present(in my case, my telemetry modem) circuit worked fine, but as i turn on the modem  and placed near to the cable, problem occured again. hope i can describe the situation well; while i looked at pwm signal on scope, there is a tiny signal over pwm signal,which travels through the pwm. i mean, this signal sometimes comes over the 0v level, and sometimes over 5v signal. and here when jittering occurs, when this signal reaches to the edges of pwm signal, it changes its duty cycle continually with unknown levels. i hoped opto would eliminate those signal, but i think i should also connect a schmitt trigger as well. btw, RFI is radiated one ofcourse, when i said i have a radio in the system, i thought its not needed to say radiated rfi :$ and also, i couldnt remembered which type of toroid i've used before but i think it was ferrite ofc.

[engamor]

If the 6N137 is on the receiving side, as it should be, and your PWM signal is still affected, then the problem is that the disturbing RF is already there, in the servo itself. I mean it reaches the servo anyway, not just along the PWM signal path. The problem is certainly aggravated by the long wires for the PWM transmission, but there are possibly other paths that conduct the noise. First of all you should check this possibility, detach the PWM cable and look in the servo circuit if RF is present anyhow.
The noise may travel over the power supply wires or simply be received in the servo because of RF coupling.
May be a physical shielding of the servo is required, as well as decoupling of the power supply etc. Try to fix one thing at a time. First check on the "overall" bad noise coupling, later look at the PWM transmission ...

[Sepiroth]

thats what i thought so, in order to minimize coupling over power supply line, i used discrete supply near servo and supplied it from there. i've seen this opto solution in a diy uav blog, they only used an opto nothing else, and stated they run servo's without jittering..6n137 is on receiver side ofc, and long cable aggravates problem for sure, but thats what i wanted to do. normally i've solved the problem by using a servo driver board for near each of servos, but it makes my overall system more complex. you think if i use a lpf with approx. ~100hz cutoff freq, can i suppress those RFI on the cable?

[Ichan]

Did you isolate the ground?

Attaching a drawing of your opto circuit would be good.

-ichan

[Sepiroth]

yes, idd i isolated two grounds of system. (at least i thought i did so:|) i attached a drawing which is what i did so far, opto circuitry is the same as its defined in its datasheet, 100nF bypass cap's etc. http://yfrog.com/3voptoisop

[engamor]

detach the PWM cable and look in the servo circuit if RF is present anyhow.

Have you done this yet? If yes, please confirm what the result is.

[Sepiroth]

yes i did, there is no RFI present at servo input.

[engamor]

So if you are positively sure that the cable is the culprit, you should try the following: use a high quality shielded cable. You need  2 wires + shield. Connect the two internal wires as in the drawing you have posted. The shield shall be connected on one side only to one of the three possible endings : originator signal ground, target signal ground, servo signal ground right side of the isolator. Connect one ending of the shield only. Of course the referred ground shall be a good quality one, close to the power ground, far from the signal paths. If this does not work the only safe way is to shield the board in a metal case and enter the signal via shielded connector or similar means. UNfortunately once the rf signal is on the board there is little to do. You must prevent it to reach the board, in my opinion.
By itself the optoisolator does nothing for this kind of problem. It is NOT impervious to radio frequency!

[Sepiroth]

you are totally right, at first i also thought faraday casing would solve the problem in a way, but there is a little problem there; those servo boards and and servos etc, are going to placed in a plane so more metal case=more weight and my lovely aerospace engineer mates gonna kill me lol idd optoisolation isnt a major problem solver here, but i thought if i use a clean regulator in opto iso circuit and place it as close as i can, it should work, big failure goes for me:P btw, can you refer me a good/simple book for pcb design, specially which covers rfi/emi interference solutions, good grounding etc?

Posted on: November 25, 2010, 10:00:53 22:00 - Automerged
btw, just thought a second, what i think is normally opto should work, without RFI at the right side of the isolation circuit. but servo jitters whatever i did, so as i said before which there is no RFI at servo side, i guess i was wrong. duh >_> i guess i should use a lpf or so..

[engamor]

Well ... if you have not answered correctly to my question about the presence of RF interference on the board itself, and I mean with the PWM cable detached ... I cannot be of any help. It takes time and efforts to be here and share one's knowledge, but it is pointless if the correspondant "already knows" things that are to be tested instead! So, once again: what happens to the RF if the cable is not connected? Does the servo still jitter ( it should be still at the rest position, as it does not receive pwm signal) ? Also : keeping everything as in the previous test, can you move the servo to another position and verify what happens to jiitter.
Another point: provide more information about the servo operation. How is it operated by the PWM signal? The PWM, is it converted to an analog voltage and tracked by a servo potentiometer? Or is it all digital? 
As for the circuit PCB design, there are book in the book section of sonsivri, I am not an expert on the subject. Hope some other guy will answer, BUT: how is made the servo board? Dual layer PCB? It may be necessary to have a 4 layer with a good ground plane in the sandwich...

[Sepiroth]

at the moment, i remade the test, and servo does not jitter (stays at steady state) while pwm cable is detached.
i attached pwm cable back in, and moved servo a little bit away from rfi source, and jitter is there again but lesser. btw servo is a hitec one, as far as i know, its analog one.

Posted on: November 26, 2010, 08:54:05 08:54 - Automerged
oh and forgot, servo driver board is 4 layered one, but as im not the one designed it(and the original designer left), i dont know much if its groud plane sandwiched etc. and also, thanks for all of yours help really:)

[engamor]

And what do you report about shielding the cable? Any improvement?

[Alessio]

Someone talk about ferrite, there are a little bit here from Laird Technologies:
http://www.lairdtech.com/Products/EMI-Solutions/Ferrite-Products/Ferrite-Cable-Cores/