I have an old (1970s) PCB that generates a (baseband, composite) NTSC video signal. It is a bare board without a containing chassis or case. I am using a shielded cable for the video output line and connecting that to a monitor. On the monitor side, the shield connects to the outer ring of the RCA plug as is standard, and it connects to a composite video connector. (i.e. not a TV/RF signal)

My understanding of shielding on audio/video signal cables has always been that what you're trying to do is protect the signal inside by essentially making the shield of the cable an electrically continuous "part of" both device's chassis on both sides of the cable, so that interference moves around the "outside" of the overall setup and allows the signal to pass unmolested inside:

enter image description here

With my video board, there is no chassis to connect to. If I leave the PCB side of the shield just unconnected, then there is interference and distortion in the video signal. If I connect it to the ground plane on the board, then I get the expected clarity in the signal.

But I'm a bit at a loss as to why that is, or rather, how I should think about it. I (probably!) understand what's going on when both sides of the cable shield connect to the respective chassis which are usually then connected to earth: interference travels around the outer shell of the system, doesn't mess with the circuits inside and is shunted to earth. But what is going on and why does it work to connect to ground plane?


  • \$\begingroup\$ It doesn't change the answer, but is the signal baseband NTSC, or is it modulated onto a TV channel? \$\endgroup\$
    – TimWescott
    Feb 5 at 22:49
  • \$\begingroup\$ @TimWescott baseband NTSC (= composite video but happens to be monochrome) \$\endgroup\$
    – BZo
    Feb 5 at 22:51
  • \$\begingroup\$ I know I'm being picky, but edit your question for clarity? I would have picked it up if you'd just said "baseband NTSC" or "NTSC (baseband)". There's lots of potential question askers (and question-readers) that may mistake one for the other. \$\endgroup\$
    – TimWescott
    Feb 5 at 22:52
  • \$\begingroup\$ @TimWescott: Sure. Just to clarify, I want to make sure we mean the same thing here: this is a video signal that connects to a composite input on a display-- not a television signal. Assuming the term baseband is correct for that context, I'll edit. (I avoided the term composite because my understanding is that that implies a color signal) \$\endgroup\$
    – BZo
    Feb 5 at 22:54
  • \$\begingroup\$ Yes, composite to a display (not to the antenna on a TV) is baseband. \$\endgroup\$
    – TimWescott
    Feb 5 at 23:03

1 Answer 1


The proper cable for that is 75-ohm coax.

A better term for the conductors in a coax cable are the inner and outer conductor. The outer conductor on a coax cable does, indeed, serve a shielding function (it's actually one of the figures of merit that pop up when you're shopping for cable). But perhaps more importantly, it's a conductor. It's the wire that's providing the ground reference for the signal going from your board to your monitor.

In a proper coaxial transmission line, the actual signal energy is confined to the inner conductor and the inside surface of the outer conductor. The result is a nice clean transmission path for broadband signals, with equal-length lines and constant impedance. Whatever current is flowing in the inner conductor is matched by an equal and opposite current flowing on the inner surface of the outer conductor.

When you left off connecting the shield to the board, the signal path became very odd. Basically, the forward path was still the inner coax conductor, but the return path (ground) became whatever connection happened to be in place between the monitor and the board. I assume by the fact that it worked at all that both the monitor and the board were grounded for low frequencies, but for the up-to-6MHz signals in NTSC, that would have been a torturous and uneven path, and one subject to stray "alien" signals getting in.

Hence your odd looking video and noise.

While I usually use the term "shield" myself for the outer conductor of a coax cable, it's really only proper if the cable itself already provides an adequate return path for signals such that the cable would work just fine if the environment were electrically quiet. You can see this in things like full 25-pin RS-232 setups where there's a bunch of wires including a ground wire inside a shield, or HDMI where there's a common return or each signal is on its own twisted pair. In both cases there's an entirely separate connection for the cable shield.

If there's just two conductors, though, that "shield" is really a full-fledged conductor that also provides some shielding.

  • \$\begingroup\$ Ahah that's super helpful. I misunderstood that the coax "shielding" was part of the signal circuit in this context-- I have dealt mostly with the other configurations like your RS232 example, etc, to which I suppose my original thinking still applies. And I appreciate the clarification that the outer side of the outer conductor in a coax cable is also performing the function of a "shield". \$\endgroup\$
    – BZo
    Feb 5 at 23:08
  • \$\begingroup\$ @BZo But it is a signal wire. In any context, a signal needs a forward path (middle wire in coax) and a return path (the outer shield of coax). In this case of your system, it just happens to be coax fot high speed signal, but s it does not need to be coax. For example a batteru also has two terminals, if current flows from one terminal to somewhere in a wire, it must return from somewhere to the other terminals via another wire. \$\endgroup\$
    – Justme
    Feb 5 at 23:35
  • \$\begingroup\$ RS-232 is not a twisted pair, not even in 25 pin setups as about 9 pins is stll used, maybe 10 if one pin is used for chassis ground. HDMI again is a differential pair with shield, so yes there is a separate return wire but it's not ground, even if shield is ground. \$\endgroup\$
    – Justme
    Feb 5 at 23:41
  • \$\begingroup\$ @Justme: well, I wasn't trying to say that; it's just how it looked in the end. Hopefully I've clarified. \$\endgroup\$
    – TimWescott
    Feb 6 at 2:10
  • 1
    \$\begingroup\$ "the other configurations like your RS232 example" -- actually, in that case you should probably bring the shield to the board ground, too. There are some very special cases where you only want to connect one end of the outermost shield -- but if you have to ask, you probably want to connect the shield to local ground on both sides. \$\endgroup\$
    – TimWescott
    Feb 6 at 2:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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