Take the 2-minute tour ×
Electrical Engineering Stack Exchange is a question and answer site for electronics and electrical engineering professionals, students, and enthusiasts. It's 100% free, no registration required.

I am re architecting something that works. (laser range finder). This is a cost down excercise. In the existing system I have separate boards for laser driver and afe. Now I am combining those and the clock for laser driver and the output of the afe will be coming and going on the same fpc cable. We are measuring tiny signals therefore usually the whole afe is shielded including the cables in the existing design.

Now imagine a 20 lane 0.5mm pitch fpc cable where pin 1 is laser clock and pin 20 is afe output. Apart from putting them opposite sides of the cable what else I can do to avoid cross talk. I though about insertin gnd lanes in between but that is very much it.

What is the best practice in these cases?

share|improve this question
add comment

2 Answers

up vote 4 down vote accepted

Above all, do not put critical signals on the ends of the cable! In your case, that is pin 1 and pin 20. The wires on the ends of ribbon-like cables are the most prone to emitting and receiving EMI.

From a radiated emissions point of view, the worst thing you can do is put a clock signal on pin 1. Putting a sensitive signal on pin 20 is the second worst thing you can do.

Here's how I would do it:

Pin 1: Ground. Pin 2: Clk. Pin 3: Power. Then put a 0.1 uF decoupling cap at the connector (both ends) between Power and Ground. If you don't want power on the cable then use two ground pins, but still put a decoupling cap between power/gnd. Do the same things for the AFE signal, with power/ground on either side and a decoupling cap.

The next thing to do is properly terminate your signals. A signal that is perfectly terminated will not radiate any EMI. Of course we can't perfectly terminate anything, but the closer you get the better. Controlling the impedance of your PCB traces and cable is important since it goes hand-in-hand with properly terminating your signals. Do you even know the impedance of your cable or the traces on your PCB? It's probably time to find out!

All of this, from which wires on the cable to use, to proper termination, to controlling impedance all falls into the category of "signal integrity". This is a huge subject that we can't cover here. I highly suggest this book: High Speed Digital Design: A Handbook of Black Magic. It's expensive, but you will save every dollar by not having to respin your PCB an extra time or two.

share|improve this answer
    
RE: "A signal that is perfectly terminated will not radiate any EMI." Not strictly true. Some transmission line geometries (including microstrip) are inherently lossy, regardless of the termination. The radiation may be small, but it's there. To absolutely eliminate radiation, you need a fully shielded transmission line. And even then, you need to be careful about how it's shielded. –  The Photon Mar 31 '12 at 17:36
    
@ThePhoton If you want to be pedantic then you're absolutely correct. The better statement would be "a perfect differential transmission line, with perfect termination, will not have loss or radiate any EMI". Of course there are no perfect anything. Transmission lines have dielectric and skin effect losses, and it is almost impossible to perfectly match impedance through vias, connectors, and cables. For most people, controlling impedance and getting proper termination is 99% of the problem. The remaining 1% can worry about skin effects and dielectric losses. –  user3624 Mar 31 '12 at 18:00
add comment

Gnd lines between the clock and the afe signal are the usual best practice.

Other things that will help:

  • try to make the clock input high impedance, so that there is as little current as possible.
  • slow the clock edges by adding a series resistor and or small parallel capacitor. Verify that the clock is still good using an oscilloscope.
  • Make the afe signal as low impedance as possible by buffering it at the other end. Even better, amplify it.
  • Synchronise the sampling of the afe signal with the laser clock is possible.
share|improve this answer
1  
Sometimes making the clock input high impedance, or adding an RC filter to the clock, is the exact opposite of what you want to do. This could mess up your signal termination, causing more radiated emissions and thus more crosstalk. Slowing clock edges can also cause more jitter which could effect sensitive time measurements. Proper signal termination will eliminate both ringing and radiated emissions without effecting clock edge rates. –  user3624 Mar 31 '12 at 16:15
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.