Shielding components from EMI/RFI... Help please

A recent experiment with a product designed to reduce EMI/RFI left me curious about other ways to reduce EMI/RFI in my system. In the past ten days, I've stepped onto a slippery slope, at the bottom of which is surely some kind of insanity...

I've been experimenting with copper plates in an effort to absorb, deflect, diffract, and block EMI/RFI. I've tried copper plates under components, on top of components, and inside components.

This is the point where you tell me I don't know what I'm doing and I'm likely to short circuit something and/or electrocute myself. Consider me duly warned. This is also the point where you tell me to get some balanced interconnects, or at least to get some shielded interconnects for Chrissake. Consider me duly informed. Moving on...

I'm hoping you can help me make the most of this experiment, and help me avoid killing a component or myself. My strategy so far has been to:

1. Place copper plates at locations that generate a lot of EMI/RFI, e.g., components with switching mode power supplies or high frequency clocks. The system has a total of 3 SMPS and 3 clocks.

2. Place copper plates at locations that are vulnerable to EMI/RFI, e.g., under the amp, near the transformer.

3. Place copper plates inside noisy components -- in particular, my Meridian G68 preamp/processor. I've begun to build 2 partial Faraday cages, one for the SMPS, and one for the analog output stage.

4. Ground the copper plates either to the component chassis (when plates are used inside a component) or to an independent ground point (when plates are used above/below a component).

Has anyone tried this sort of thing?

I use GIRON shields between my run of power cables and components.

Have a look at the store lessemf on the web. They have a variety of products. I picked GIRON because my power cables have large magnetic modules, and these were recommended by the manufacturer.

Yes, slippery slope to insanity, indeed.
Check page 14 of this catalog: ( I, personally, use the TI Shield(between all my components), but- the Stillpoints stuff sounds interesting too. The last page of the catalog has some new products from Bybee, for the power supply.
Check out the Trifield Electromagnetic detector.
I got mine for $100.00 on Amazon.
This meter is promoted as a tool to protect your health,
by many people.
It also measures audio equipment for magnetic, electric and radio/microwave radiation.
You will be able to measure the results of your modifications.
? $100? get a Greenlee non contact voltage detector $18 from Amazon

I use one, and it is better than most oter models of the sort of device.
Sniff out radiated rfi. and find out what actually blocks it!

I like black antistatic foam. (has to be grounded to really work well). I use it in DACs, and around Power supply umbilicals on my preamps.

I also use quartz crystals to block EMI
Thanks, ladies and gents. The TI Shield looks very interesting. I think I'll order some to experiment with. And I will get myself a meter posthaste.

Just build a faraday cage around your whole room;). Run all your cables through copper pipe. Your cell phone won't work anymore but who needs distractions?
Did I forget to mention that I already rebuilt my listening room?

Nice room. Warm and inviting.

I once wrapped the umbilical cords of my preamp with aluminum foil then teflon tape over that. It was cheap and did lower the noise floor. When I moved I decided to remove the foil and listen again. The sound improved quite a bit without it. The sound was more open and airy. Sure the noise floor was higher but not much.

So I suggest once you get used to the change remove it and revaluate its benefits. You may be surprised with the results.
That is good advice, Sarcher. I had a similar journey with analog interconnects. After trying a number of different shielded interconnects from well respected manufacturers, I settled on an unshielded pair. As you describe, the sound was more open and airy with the unshielded cables. I've also experimented with shielded and unshielded power cables, and found that I preferred them unshielded or minimally shielded.

I guess I'm learning that I have a different attitude about shielding analog signals and power cables vs. shielding digital signals and digital circuits. The digital interconnect I settled on is double shielded. Very happy with it. As I mentioned in the OP, I have 3 digital devices with high speed clocks and switching mode power supplies. I've built complete shields for 2 of the digital devices, and a partial shield for the other. I recently removed the shields to make some other changes and I discovered that I definitely prefer the sound with digital devices shielded rather than unshielded. It wasn't so much a matter of audible noise as it was a matter of high frequency grain and glare. I guess that I could sum up my (admittedly limited) experience by saying that...

Shielding analog cables and power cables tends to reduce openness and air :-( whereas shielding digital cables and digital circuits tends to reduce grain and glare :-)

I have no idea if that is generalizable.

That is an interesting observation about digital interconnects. I have a Linn DS which uses an ethernet connection from a NAS in a different room. I wonder if your theory would apply in my situation? I'm not sure if ethernet cables benefit from more shielding? Guess I could search the net to see what pops up.
I also use ethernet. My understanding is that most (all?) ethernet protocols employ a variety of error detection/correction techniques that ensure perfect or near perfect data transmission.

If that is true, then I would imagine that additional shielding on ethernet cables would have little or no effect. But what do I know? Maybe it would have some effect. Stranger things have happened in the world of audio.

Maybe someone who knows this stuff better will weigh in.
Bryon & Sarcher30, IMO shielding an ethernet cable will provide no benefit with respect to the integrity of the data it is carrying, or with respect to the amount of jitter that will result when that data is ultimately converted to analog (since the data is buffered at the receiving end). However, I would not completely rule out the possibility that shielding the cable could be helpful in terms of minimizing the amount of digital noise that could couple FROM the cable to arbitrary circuit points within the system.

From this paper by Steve Nugent:
Networked audio (Ethernet), both wired and WiFi is a unique case. Because the data is transmitted in packets with flow-control, re-try for errors and buffering at the end-point device, it is not as much of a real-time transfer as USB, S/PDIF or Firewire. The computer transmitting the data packets must still keep-up" the pace to prevent dropouts from occurring, but the real-time nature of the transfer is looser. Unlike with other protocols, there can be dead-times when no data is being transferred. Networking also avoids the use of the audio stack of the computer audio system since it treats all data essentially the same. This avoids kmixer on XP systems and the audio stacks on Mac and PC Vista. Because of the packet-transfer protocol of Ethernet and data buffering at the end-point, the jitter of the clock in the computer is a non-issue. The only clock that is important is the one in the end-point device. Examples of end-point devices are: Squeezebox, Duet and Sonos. This would seem to be the ideal situation, which it certainly is. The only problem that can occur is overloading the network with traffic or WiFi interference, which may cause occasional dropouts. The problem for audiophiles is that the majority of these end-point devices were designed with high-volume manufacturing and low-cost as requirements, with performance taking a lower priority. As a result, the jitter from these devices is higher than it could be. It should be the lowest of all the audio source devices available.

Best regards,
-- Al
Thanks, Al. That's what I suspected, but it's good to hear from someone who actually knows something (in this case, both you and Steve).

On the issue of ethernet cables themselves releasing EMI/RFI... I have no idea whether my ethernet cables are shielded. Is there a way to determine that without cutting into them?
Almarg, that makes sense. Thanks for the info.

Bc, I found some sheilded cat6 cable online. It was cheap enough I decided to give it a shot anyway. I have to run my ethernet fairly close to some power cords because my house is not prewired with it. I'm using some cat5 I made myself now. I know it is not sheilded. I think if your ethernet cable is sheilded it will say so on the jacket.

Also, in many cases either the letters UTP (unshielded twisted pair) or STP (shielded twisted pair) will appear on the cable.

Or if a manufacturer's part number appears on the cable, Googling that number will presumably lead you to a description of the particular cable, and whether or not it is shielded.

I believe that it is considerably more common for CAT 5e cable to be unshielded than shielded. Not sure about CAT 6.

Another point to keep in mind, I believe, is that the shield in most or all shielded ethernet cables will only be grounded if the jack it is plugged into makes that provision, and I believe that only shielded jacks will do so.

Best regards,
-- Al
I am guessing that consumer equipment is unlikely to have sheilded jacks if most ethernet cables are unsheilded? Oh well. They only set me back $25 with shipping.
Thanks, Al. My Cat 5 cables are unshielded. I just ordered some shielded Cat 6.

Yesterday I received the materials to complete the DIY projects described in the OP. Hopefully there will be some results. I will report back.

Bc, I should get my cat6 tomorrow. I will report back if I hear any change. I have my doubts though. Please let us know how your projects work out.

P.S. I like your real room. Nice and tidy. If not for the speakers you would not know an audiophile lived there.

Thanks, Sarcher. Interested to hear if the Cat6 makes any difference.

I have an update...

I have installed 7 copper plates inside my preamp/dac. They are each 16 gauge, or .06", which is fairly thick. I grounded each plate to the chassis. I bolted 3 plates around the analog output stage, 3 plates around the power supply, and 1 large plate to the underside of the lid.

The results are surprisingly good. The noise floor dropped noticeably. The resolution increased slightly. The high frequencies got smoother. I can turn the volume up about 3dB with the same comfort level.

I've tried quite a few tweaks and mods, and this is one of the most audible. The results are much more obvious than when I recently added ERS cloth to the preamp/dac. Total cost for this mod: about $100 in copper from an online retailer, who also cut the plates to exact sizes.

Next up: I'm building a copper/aluminum enclosure for my reclocker. The enclosure will be grounded to a dedicated ground rod I just buried under the house... be continued.

I installed my shielded cat6 today. I think it made a slight improvement but it could just be my imagination. I had to reboot my NAS and Linn DS afterward. It takes a long time for the NAS to reboot so it makes it difficult to do a proper A/B comparison. It looks like the Linn DS ethernet port has metal tabs on the sides so I think it is shielded. I don't know if they connected the port to ground though. My NAS also seems to have a shielded port. The only thing in the chain without shielded ports is the router. So I plugged in both ethernet cords next to each other and then wedged in a scew between the metal shields on the RJ45 plugs to tie their shields together. Does anybody think that will work?
The only thing in the chain without shielded ports is the router. So I plugged in both ethernet cords next to each other and then wedged in a scew between the metal shields on the RJ45 plugs to tie their shields together. Does anybody think that will work?
I doubt that is necessary. As long as the shields are grounded at one end, I would think that should be sufficient. And I would expect that which end is grounded doesn't matter, since what you are trying to do is to prevent rfi from escaping from the cable.

In other situations, as you may realize, if the signal flow is unidirectional and only one end of the cable shield is grounded it is generally desirable for that to be the source end, so that rfi that may be picked up won't be conducted to the destination component. But that is not applicable in this situation, because rfi pickup is not the concern (and probably also because the link is bidirectional).

-- Al
Almarg, I have heard of interconnects with the shield only connected on one end. I was not sure if that arrangement would work in this application.

Thanks again.

Sounds like the Cat6 experiment has produced uncertain results. I've already ordered shielded Cat6 from Amazon, so when it arrives I will let you know if I hear any difference. In the meantime...

I can now report that my shielding experiment has produced fantastic results. I hesitate to use the word 'fantastic,' because I don't want to diminish my credibility by overstating the facts. But the fact is that the addition of shielding has produced very impressive results.

I had an audiophile friend over who is familiar with my system and he said "It sounds like a different system." THAT might be an overstatement, but it is fair to say that the system sounds MUCH improved. Two improvements are particularly noticeable: Greater resolution and less high frequency grain and glare. The difference in these two categories is so pronounced that you can literally hear the improvements from the next room.

Here's everything I've done so far:

-Custom cut 16ga. copper plates around the Meridian G68's power supply, analog output stage, and the entire underside of the lid. All plates are grounded to the chassis.

-An aluminum/copper enclosure around the Empirical Audio reclocker. The enclosure is grounded to a dedicated ground rod under the house, and the reclocker is electrically insulated from the enclosure.

-An aluminum/copper enclosure around the Sonos. The enclosure is grounded to the dedicated ground rod, and the Sonos is electrically insulated from the enclosure.

The biggest gains were achieved by adding copper plates inside the Meridian preamp/processor. I added plates in three separate stages, and each time there was a clearly audible improvement. Apparently, in this case, more is more.

I don't know how generalizable my results are to other preamps, because my Meridian G68 is undoubtably much noisier than the typical audiophile preamp. But my G68 is also my dac, and I suspect that similar results could be achieved with other dac's or players.

It bears repeating that I am not an ideologue when it comes to shielding, as you can see in this thread. My (admittedly limited) experiences have led me to prefer unshielded or minimally shielded analog interconnects and power cords.

But digital interconnects and digital circuits are another story. For a while I've suspected that shielding most (all?) things digital is a good idea, and the results of this experiment confirm that suspicion. Of course this is a sample size of 1, so I recognize that YMMV.

Sounds like you made some worthwhile improvements. Your conclusions seem to support the idea of separates. I have separates in my system and some even have separate power supplies. I'm a believer.

It is interesting that with some careful shielding some integrated solutions can be improved. Makes you wonder why you don't see shielding like you have done in more high end integrated designs.

What's with the dedicated ground rod buried under the house?

The earth does not possess some mystical magical power that will suck, drain, RFI to it.

What you have created is a great path for a lightning strike to follow. There is a reason why all grounding electrodes of a building must be connected to one another....

Just for a test disconnect the shields from the dedicated ground rod and reconnect them to the chassis ground of the equipment. Post back your listening results.
Sean - I think my results, if they can be generalized, point to the value of extensive shielding in dac's and cd players, all of which have clocks, and many of which have noisy power supplies. I agree that it's strange that you don't see extensive shielding more often. The copper for the shielding in my G68 cost me about $100. That's a very small fraction of the current $12K msrp.

Jim - The ground rod wasn't based on magical thinking. It was based on the belief that the ground for a Faraday cage should be different from the ground for the equipment inside the cage. I gather from your comments that that is incorrect.

When I installed the copper shielding inside the G68, I grounded it to the G68's chassis. The dedicated ground rod was used only for the 2 enclosures - one for the reclocker and one for the Sonos. I have disconnected the ground rod and connected the enclosures' ground cables to a conductive spot on the G68's chassis. My initial impression is that I can hear no difference between that arrangement and my previous one.

After more extensive listening, I've concluded that grounding the new enclosures to the Meridian G68 resulted in sound that is slightly worse, particularly in "listening ease" at loud volumes. Here are the 3 arrangements I've tried:

1. Enclosures grounded to ground rod.
2. Enclosures not grounded.
3. Enclosures grounded to G68.

I can say with some degree of confidence that (1) and (2) sound better than (3), but I cannot say with much confidence whether (1) sounded different from (2). So maybe the ground rod is adding nothing other than additional risk.

As for why (3) sounded worse than (1) and (2), I don't know what to say. Maybe grounding the enclosures to the G68 made the enclosures act like antennas collecting RFI, and that RFI was then transmitted into the G68? I don't know. I'm out of my depth. Maybe someone has a theory.

Hi Bryon,
As for why (3) sounded worse than (1) and (2) ... maybe grounding the enclosures to the G68 made the enclosures act like antennas collecting RFI, and that RFI was then transmitted into the G68?
That would certainly seem like a possibility. In particular, perhaps noise originating from the enclosed devices is finding its way from the connection to the G68 chassis to the ground/signal return of the digital input to the G68. A noisy ground at that circuit point would certainly contribute to jitter, especially if the connection is unbalanced S/PDIF rather than balanced AES/EBU.

Question for Jim:

Although the additional ground rod may very well not be helpful with respect to minimizing noise, I'm wondering how it would represent a safety risk in this particular application.

There are two reasons I can think of why code prohibits multiple ground rods that are not connected together.

One is that if it is used to ground the safety ground of an electrically powered device, fault current may not cause the circuit breaker to trip. That is explained on page 8 of this Bill Whitlock paper.

The other reason I can envision, which is also mentioned in the paper, is that if lightning hits the outdoor electrical wires and is conducted to ground through the main grounding rod, thousands of volts may briefly appear between the two grounding rods, potentially damaging anything for which a connection path to both of them exists.

Neither of those scenarios seems applicable to what Bryon was doing with the second ground rod, assuming that the means of insulation between the Reclocker and the Sonos and their grounded enclosures is sufficient to withstand the voltage that would appear across it during the lightning strike scenario. Perhaps that isn't a good assumption, though. Can you explain further? Thanks.

Best regards,
-- Al
Almarg, I've always wondered about the code to use only one ground rod as well. I have never heard an explanation. Usually people say you can't do it but never explain why. I will have to read that paper you linked to.
Neither of those scenarios seems applicable to what Bryon was doing with the second ground rod, assuming that the means of insulation between the Reclocker and the Sonos and their grounded enclosures is sufficient to withstand the voltage that would appear across it during the lightning strike scenario.


You hit the nail on the head. The insulation would possibly have to withstand hundreds of thousands of volts. Though quicker than the blink of an eye, still long enough to damage solid state devices.
Odds? No idea.... But if the connection to the rod made no difference in sound why take the chance.

If memory serves me right a lightning strike can travel through the earth for 5 miles.

I can say with some degree of confidence that (1) and (2) sound better than (3), but I cannot say with much confidence whether (1) sounded different from (2). So maybe the ground rod is adding nothing other than additional risk.
02-13-12: Bryoncunningham
A Faraday cage does not have to be earth grounded to work.

Though usually, say in the case of a 2 wire cord and plug piece of electronic equipment such as a CDP, the Faraday cage (metal enclosure) is connected to the signal ground.

Electronics RFI shielding/Faraday cages of a Stealth Bomber are not connected to the earth.
Thanks Jim!

Best regards,
-- Al
Well I read the Whitlock paper and it explains it quite nicely. Thanks for the link.
Almarg, I've always wondered about the code to use only one ground rod as well. I have never heard an explanation.
02-14-12: Sarcher30


There is no limit of the maximum number of grounding electrodes (ground rods), only the minimum required.

What NEC does require is that they must all be tied electrically together. When all tied together they are considered by NEC as one grounding electrode.

Several 8ft ground rods is not necessary better than one 30ft deep driven ground rod. What matters is the rod/earth resistivity...... The lower the earth soil resistance the better. 3 ohms or less is considered very good.

I would bet very few houses even come close. NEC code says if the earth resistance is greater than 25 ohms the ground rod shall be augmented by one additional ground rod. The earth resistance could be 60 ohms but all NEC calls for is one additional rod. The earth resistance still could end up higher than 25 ohms...... NEC code is satisfied with the one addition rod.....

At least in my area, I have never heard of a residential electrical contractor ever checking the earth ground resistance. I have only seen it checked on industrial facilities. Special equipment is required for the test.
Jim, Interesting article. Do you think it would be worth while to have a second ground rod installed? It seems that even an eight foot ground rod would be difficult to install without drilling a hole first.

Thank you, Al, for your help. In light of Jim's advice, I've abandoned the idea of using a dedicated ground rod. Even though I live in Southern California where lightning is rare, there's no good reason to take the risk. It's really no sacrifice anyway, because I could not hear a difference between the enclosures grounded vs. ungrounded.

The improvement from adding the two additional enclosures (whether grounded or ungrounded) was perceptible but not dramatic. The big improvement came from adding shielding INSIDE the G68 (grounded to the chassis). With that in mind...

I've been reading that copper is effective at shielding high frequency RFI, but not particularly effective at shielding low frequency EMI. So I'm thinking about adding a layer of STEEL shields around the G68's power supply, bolted directly to the copper plates that are already in place.

Try this:

Looking inside the Meridian G68, I see that the power is grounded to the chassis at TWO points. A ground wire connects the IEC input to the chassis. A second ground wire connects the power supply to the chassis.

The unit has been modified extensively, both by me and by a professional modder, so I don't know how it was grounded when it came from the factory.

Is it okay for the power to be grounded to the chassis at two points? Or am I creating a ground loop?

Hi Bryon,

My suspicion would be that the connection from the power supply is connecting the ground on the DC output side of the supply to chassis, thereby connecting the G68's circuit ground to chassis. That is often done, although as you surmise it can result in a ground loop between the G68 and the components it is connected to.

I wouldn't play around with it, though, unless there is a particular reason to do so, and an understanding of why it was done that way. Other possible approaches, btw, besides simply removing the connection, being to connect circuit ground to chassis via either a resistor, a capacitor, or an inductor.

Keep in mind, also, that balanced connections have much less susceptibility to ground loop issues than unbalanced connections. I assume that you have a balanced connection to the Pass amp, and perhaps also a balanced AES/EBU connection from the re-clocker.


-- Al
My suspicion would be that the connection from the power supply is connecting the ground on the DC output side of the supply to chassis...

You may be right, Al, although the ground wire on the power supply is right next to the power INPUT. Here are some pics...

Modded PSU photo 1

Modded PSU photo 2

Modded PSU photo 3

Stock PSU

The power input is the blue/red wires in the top left of photo 1. The ground is the green wire in the bottom left of photo 1.

Can you tell from looking at those pictures whether the ground is on the AC side or the DC side? Can we infer from its physical location that it is on the AC side?
Hi Bryon,

Yes, you're right. After looking at the photos I believe that the two small blue objects just above the green wire are capacitors that are provided for purposes of filtering high frequency noise that may be present on the incoming AC. One capacitor is probably connected between AC hot and chassis, and the other capacitor between AC neutral and chassis, with the green wire providing the connection to chassis for both capacitors.

Ergo, not a problem!

-- Al
Thanks, Al. I will leave both ground wires in place. There is something else strange to report...

I installed a shielded Cat6 cable today, replacing an unshielded Cat5 cable. I expected to hear no difference whatsoever. Much to my surprise, there was a BIG difference. I know that sounds crazy. I don't understand it at all.

I know that the shielded cable might keep out EMI/RFI, but that doesn't seem to explain what I'm hearing, which is significantly more resolution. It's dramatic enough that it seems difficult to explain as simply less EMI/RFI. The difference sounds similar to the kind of resolution jump you get when you go from 16/44.1 to 24/96. In other words, it sounds like more information is reaching the system. But how could that be, if the ethernet protocol utilizes error detection/correction? I'm confused.

The old configuration was...

iMac -> 50' unshielded Cat5 -> Apple Airport Extreme -> 50' unshielded Cat5 -> Sonos -> S/PDIF -> Reclocker -> S/PDIF -> Meridian G68

The new configuration is...

iMac -> 50' unshielded Cat5 -> Apple Airport Extreme -> 25' shielded Cat6 -> Sonos -> S/PDIF -> Reclocker -> S/PDIF -> Meridian G68

So only ONE link in this chain changed: a 50' unshielded Cat5 was replaced with a 25' shielded Cat6. And the result was a big jump in perceived resolution. I'm scratching my head...

Hi Bryon,

Presumably the shielded cable is reducing the amount of rfi/emi that escapes from the cable and couples onto circuit points elsewhere in the system, most likely points that are "after" the reclocker circuitry. There are probably a number of ways in which that could occur, despite the shielding you have incorporated into the G68. One possibility would be that it couples onto the power wiring, and makes its way from there onto the grounds within the G68. Noisy grounds in the digital section could (and probably would) result in increased jitter. Noisy grounds in the analog circuitry would have unpredictable but conceivably significant consequences at audible frequencies, even if the noise itself is at rf frequencies (as a result of intermodulation, AM detection, or other effects that can cause rfi to affect audible frequencies).

That's my speculation, anyway.


-- Al
Al, I think you may be right in your speculation that the reduction of EMI/RFI provided by the shielded cable is resulting in less jitter. That is consistent with the change in sound quality. If that theory is correct, the system now has a LOT less jitter.

I spent a couple hours listening last night and I continued to be stunned by the difference. I have NEVER experienced anything like this with any other cable. In the past, I've had two different $2k analog interconnects in the system, and they didn't have anywhere near the effect I'm describing. I currently have about $4k in power cords in the system, and they didn't result in a fraction of the improvement of this $7 cable. It's mind blowing.

FWIW, I very seriously doubt that other systems would be similarly effected by a shielded ethernet cable. My setup is so idiosyncratic that my results are most certainly not generalizable.

I'm inclined to conclude that there was a significant PROBLEM in the system that the shielded Cat6 cable has rectified, a problem that has gone undetected by me for years. Things have sounded pretty good, so when I listened to the system, it didn't sound like there was a problem. But this is such a dramatic effect, I must have been listening to/through this problem all along.

As a footnote, it's interesting that you mention intermodulation, because when I added shielding around the analog output stage, walling it off from the digital circuitry, I said to a friend: "It sounds as if there is less intermodulation." I'm not even sure I know what I meant by that, but that was the concept that immediately came to mind.

"I've been reading that copper is effective at shielding high frequency RFI, but not particularly effective at shielding low frequency EMI"

It is because copper, being non-magnetic, does not shield against electromagnetic wave. In cable this electromagnetic wave induces current but because of skin effect it travels on the outside of the cable - shield . Same happens with copper plates around components. Skin effect won't work at lower frequencies where copper becomes useless against magnetic field of 60Hz transformer or even 50kHz from switching power supply.

In case of low frequency radio interference you need at least 1/10 of the wavelength antenna to "receive" interference - not likely. You are more susceptible to direct magnetic fields like leakage field from power transformers.

MuMetal is good suggestion - it works much better than plain steel (permeability of steel is few thousand while MuMetal reaches 100,000). It comes in different permeability because higher permeability Mu Metal is easier saturated by magnetic field. In such cases sandwich of lower and higher permeability MuMetals gives best results. Your magnetic fields are most likely very weak but just in case don't place MuMetal close to the source (transformer etc.). Excessive bending of MuMetal is not advised (has to be re-annealed).

You might want to do some research on the archives of Tweakers' Asylum regarding the use of ferrous, magnetic, materials. There may be better ways to shield for EMI.
Search shielding for EMI
Jea - Thanks for the advice. I will poke around on Tweakers and see what I discover.

Kijanki - Thanks for the help. You are always a good source of information. Much appreciated.

MuMetal is good suggestion - it works much better than plain steel.

I already ordered custom cut regular steel from an online retailer. :-( You think using regular steel is a waste of time?

Your magnetic fields are most likely very weak but just in case don't place MuMetal close to the source (transformer etc.).

Again, what if I use the regular steel I already purchased? It's T-304, which I believe is non-magnetic. If that's true, do I still have to be careful how close I get it to the power supply? How close is too close?

Now that I think about it, if the steel I purchased is indeed non-magnetic, will it work for shielding EMI at all?!

T-304 won't shield against low frequency magnetic field. It might protect against it if you build Faraday Cage but even then it is very limited. I'm no expert on Faraday Cage but as far as I remember it is used mostly as electrostatic shield. It works against magnetic field but poorly at lower frequencies.

We don't even know what is the nature of the noise pickup. If switching power supply has fast transients (high frequency) even non-magnetic shield would help cutting on capacitive coupling and creating losses (eddy currents) with magnetic field. I would stay on the course, since you already ordered shields. It will give you shielding at high frequencies allowing to isolate the problem.

I absolutely agree with Al's assessment on the Cat6 improvement. Jitter is not only function of the signal noise (not likely since it is buffered) but also system noise (receiver threshold noise) that is influenced by noise injected by current induced in the cable that finds return to ground thru the system. Ethernet uses differential signaling that cuts on common mode noise but it still couples high frequencies thru input capacitance to signal ground. Currents traveling on the signal ground are system noise. Judging by improvements you experienced your noise pickup might be of high frequency and your non-magnetic shields will work.
The TI Shield, that I mentioned earlier, will block all RFI/EMI and is VERY easy to work with(thin enough to cut with scissors, and bendable).