Shielding components from EMI/RFI... Help please

Well done, Bryon, and also very well described.

I suspect that grounding the ethernet switch resulted in significant benefits because it grounded the shield for the 50’ shielded ethernet cable running between the computer and the switch.

I suspect that you're right. The shield is grounded at the other end, via the computer, but the resistance of the very long run presumably lessens the effectiveness of that ground from the perspective of the switch.

02-29-12: Almarg
One reason that [grounding the switch] would tend not to accomplish anything is that the inductance of the ground wire would make it an ineffective conductor of high frequency (RF) energy, which is what you are attempting to dissipate in this case.... Presumably and hopefully whatever RF energy is picked up by the shields of the ethernet cables from the conductors they contain will be dissipated effectively in the metallic structures of the two components, and perhaps also further upstream.

I was perhaps focusing too narrowly in these statements on RF noise related to the high frequency and very fast risetimes and falltimes of the signals being conducted by the long cable, and the need to prevent that noise from radiating from the cable to other points in the system. Lower frequency grunge presumably was also present, perhaps associated with the computer's switching power supply, power line distortion, emi pickup, etc, the effects of which may not have been entirely eliminated by the reclocker. Your ground connection is presumably a much better conductor at those lower frequencies than at the very high signal-related frequencies (for which a braided ground strap would be necessary to provide an effective path, although doing that could very conceivably worsen the results by providing a path for RF noise to bypass the reclocker and get into the G68).

In any event, congratulations on the excellent improvement!

Best,
-- Al

Thanks again, Al. A few final questions...

You mentioned a couple times that a ground must have an "effective path." I don't really know what the factors are that make a ground path effective.

To ground the ethernet switch, I drilled through the metallic housing of the block of ethernet ports and connected a 14 gauge wire. The wire runs to an aluminum screw that passes through the switch's housing. You can see a picture here. A second 14 gauge wire runs from the switch to a screw on the power conditioner. Does that seem like an effective path?

Also, you mentioned that the shield of the 50' ethernet cable is grounded at the computer end. I couldn't find any information about that online. How can you tell?

Bryon

Hi Bryon,

The 14 gauge wire is certainly an "effective path" at low frequencies, and probably at frequencies ranging up into the ultrasonic and perhaps low RF region. But the inductance of a plain piece of wire will cause it to have a significant impedance at the high RF frequencies that constitute the spectral components of the ethernet signals. It will therefore not conduct those frequencies effectively. The impedance that is presented by inductance, as measured in ohms, is directly proportional to frequency. The inductance, btw, is directly proportional to length.

As I mentioned, for good conduction of RF frequencies to ground a braided ground strap, such as this, is needed. However I suspect that if you were to use a braided ground strap on both the network switch and the G68 you would wind up with worse results, because it would create a path that would conduct RF noise from the switch to the ground point, and from there to the G68, bypassing the reclocker.

Also, you mentioned that the shield of the 50' ethernet cable is grounded at the computer end. I couldn't find any information about that online. How can you tell?

I should perhaps have qualified my comment by saying that that is the case with the desktop computers that I am familiar with, which primarily means those I have built myself using Asus and Gigabyte motherboards. On those motherboards, at least, shielded jacks are provided for the ethernet ports, the shields in turn being connected to the circuit ground on the board, which in turn is connected to the metal case of the computer and to AC safety ground.

It's easy enough to verify that on your particular computer. Disconnect the ethernet cable from the computer, and look for two small metal tabs on either side of the jack. Check for continuity from one of those tabs to the computer case and/or to the AC safety ground pin on the computer's power connector.

Best,
-- Al

Hi Al - I checked for continuity between the tabs in the computer's ethernet port and the computer's IEC ground pin. You were right, the port is grounded.

Correct me if I am wrong, Al, but that seems to help explain something that puzzled me at the time: the dramatic improvement I heard when I first replaced the unshielded ethernet cables with shielded ones. In my posts on 2/16, I was mystified about how a $7 cable could make such a difference. You offered me a speculative explanation at the time, but I was left with a certain amount of puzzlement. What I didn't know at the time (but you did) was that the shielded ethernet cable was also grounded. Knowing that now, it seems to help explain how I could have heard such a dramatic improvement.

There is one other point I've been wondering about with respect to ethernet grounding, and that is: Is it better to ground an ethernet cable at both ends? I poked around a little and I found a discussion of this topic, in which an evident expert said this...

In high-speed digital applications, a low impedance connection between the shield and the equipment chassis *at both ends* is required in order for the shield to do its job...

In high-speed applications involving low-impedance circuitry, most of the near-field energy surrounding the conductors is in the magnetic field mode, and for that problem, only a magnetic shield will work. That’s what the double-grounded shield provides. Grounding both ends of the shield permits high-frequency currents to circulate in the shield, which will counteract the currents flowing in the signal conductors. These counteracting currents create magnetic fields that cancel the magnetic fields emanating from the signal conductors, providing a magnetic shielding effect.

For the magnetic shield to operate properly, we must provide means for current to enter (or exit) at both ends of the cable. As a result, a low-impedance connection to the chassis, operative over the frequency range of our digital signals, is required that *both* ends of our shielded cable.

One thing I'm still unclear about is whether the grounding I've created at both ends of my 50' ethernet cable is "operative over the frequency range of [the] digital signals." I know you said that...

the inductance of a plain piece of wire will cause it to have a significant impedance at the high RF frequencies that constitute the spectral components of the ethernet signals.

...which, in combination with the information I quoted above, seems to suggest that grounding the ethernet cable at both ends should have no additional benefit. But I DID hear an additional benefit when grounding it at both ends. Perhaps that is because, as you said...

the resistance of the very long run presumably lessens the effectiveness of that ground from the perspective of the switch.

...but I'm not sure I completely understand how that works. In other words, I'm unclear about how the following 3 things can all be true at the same time:

1. The effectiveness of a ground method for an ethernet cable's shield depends upon whether the ground method operates over the frequency range of the ethernet signal.

2. The ground method I provided my ethernet cable (at the switch) is ineffective at the frequencies range of the ethernet signal.

3. I heard an improvement when grounding the ethernet cable at *both* ends.

Assuming I'm right about #3, maybe you can say a few more words about how those things can all be true. Thanks!

Bryon

Hi Bryon,

Good find. Whatever Dr. Johnson says, is so! He is one of the world's leading experts on high speed digital signal transmission.

As it happens, I took his course on high speed digital design about 15 years ago, in connection with my job. This is the associated textbook. You'll be amused at its sub-title, "a handbook of black magic." As you certainly realize at this point, grounding and shielding are among THE most arcane and mysterious aspects of electrical engineering, with problems often being resolved by not much more than blind trial and error. I, btw, am by no means an expert in that area.

With respect to the second paragraph of your post just above, where you said "correct me if I am wrong," I believe that everything you said is correct.

With respect to the apparent paradox cited at the end of your post, I believe that two things need to be considered.

First, for shielding to be effective at high frequencies, as Dr. Johnson indicated the shield should be grounded at both ends. But I believe that the key element of what he is referring to by "grounding" is a connection at each end between the shield and the metallic structure of the component, rather than a connection to some external ground point. I believe that circulation of noise currents from the cable shield into the metallic structure of the components will dissipate their energy significantly, although perhaps less so in the case of the network switch due to its small size.

Second, concerning the improvement you noted when connecting the 14 gauge ground wire to the network switch, my speculation is essentially as I commented yesterday:

Lower frequency grunge presumably was also present, perhaps associated with the computer's switching power supply, power line distortion, emi pickup, etc, the effects of which may not have been entirely eliminated by the reclocker. Your ground connection is presumably a much better conductor at those lower frequencies than at the very high signal-related frequencies ...

You were probably reducing the amount of low frequency noise that had been present at that point (the sources of that noise being unrelated to the ethernet signals themselves), that was making its way through the circuit grounds downstream, resulting in jitter that was not being entirely eliminated by the reclocker.

All that is obviously fairly speculative, but those are the only explanations I can think of that seem to fit all the facts.

Best,
-- Al