A *slightly* over the top power feed for my test bench

[rhb]

When I lived in Austin 30 years ago I was plagued by EMI from a powerful FM transmitter nearby.  More recently I have become aware that Chinese test gear may produce no EMI in front but  lots in the back and sides which will be picked up by any power cords running nearby.

Recent purchases of old HP gear have provided me with a fairly comprehensive test bench, but also a lot of power cords to plug in.  I briefly tried a power strip but I had a very fat bundle of cords which proved to be rather problematic.  I'm using a 7' x 10' space and there simply was not room for all those power cords.  I'd already had to move the bench 10" from the wall to accommodate the depth of my 8560A. 

Rather than have power strips feeding power strips, I decided to build a shielded power feed for my instruments using  EMT and flex conduit.  The flex was a bit of a headache.  I had to buy a 250 ft roll when I only need about 50 ft.  It's not quite as flexible as the Liquatite brochure claims, but it's not too bad, about 1.25" minimum radius.  I also plan to use it for shielded test leads.

The flex is Liquatite SL-516 and the clamps and anti-short bushing are Thomas & Betts products designed to  fit this unusually small conduit.  The clamps are 3301-TB and the bushings are 390 . I'm using Appleton 4SS boxes which are 3.75" x 1.5" x 1.5" and almost too small.

I'm cutting off the regular IEC connectors and sliding the cable through the flex.  I heatshrink the IEC end and use flex conduit clamps at the box. .  The first photo shows the basic assembly and the second shows a completed distribution box.  Once all the distribution boxes are done I'll cut EMT to space them at the appropriate distances and run #12 wire to feed all the boxes.  The soldering equipment and the bench light will be on a separate switched circuit so that turning off the light cuts the power to the soldering equipment.

Cost is $4-5 per instrument excluding the IEC cords.  I had lots of those, but bought a bunch more from a local thrift store to make sure I had thin ones which would fit the flex.

[Gregg]

From many years of experience in data centers, if I were doing what you pictured, I would terminate my input electrical ground at the junction box and run a separate isolated second ground all the way back to where the ground rod connection is located in the electrical panel and tie the isolated ground to the individual cables going to the equipment.  Than I would add at least another ground rod to ensure a good earth ground and tie it to the panel ground bus.  Also make sure any buried metal pipes are bonded together and your hot and cold water pipes are bonded together at the water heater.  As I’m sure you are well aware, anything can be an antenna. 

[rhb]

I'm probably going to do a run of 12/3 BX all the way across the attic to the panel.  But I thought I'd actually scan the power lines to the room with my spectrum analyzer first.  I'm also going to finish reading Henry Ott's  "Electromagnetic Compatibility Engineering".

I just finished making up a whip and noticed it looked odd.  There was a splice in the flex!

I don't really know if this will help much with EMI, but it will certainly be a lot more convenient.  Since the photos I added a pair of zip ties on the heatshrink.  One on the flex end and one on the IEC end.  Originally I was just going to make up a custom power cord using the IEC cords but without the flex.  Then I was probing EMI sources with my Instek scope and discovered the huge amount of SMPS noise coming out the back and sides.  That made adding the flex attractive.

[sokoloff]

Cool idea! 

I would not be comfortable with that hard-wired to a breaker panel (or even a disconnect), but would be comfortable with that with a C15 male plug as the input power feed. (That makes it device/appliance wiring, rather than premises wiring which are likely subject to NEC in most jurisdictions.)

[rhb]

If I run the BX  will plug into a normal 20 A outlet in a metal box. Otherwise it will connect to one of the regular outlets.  This is just a fancy extension cord/power strip.

It will  have a master switch to turn everything off.  That won't be used much with a GPSDO and voltage references running, but I do want a panic cutoff switch that doesn't require crawling under the bench to pull the plug.

[CopperCone]

I wish I had conduit and armored but really, if your gonna do it, they sell EMI shielded special conduit/armor cable that has like 30+ more dB of shielding in broad band.


If you went this far ,you might wanna get quality shielded IEC connector cables, and also replace any corded instrument cables you have with shielded corded ones.

I also modded some old equipment I have with a dremel to feature a mains filter (like 1960's shit)

Ground measurement is also fascinating. I thought about doing stuff like making a separate grounding rod array outside my laboratory window and running a thick ass cable braid to it, for specific experiments, but its difficult to determine how well it works or when it should be used etc... I need to learn alot more before I understand that situation well.

And then you also wanna spent 2000$ to get 5kHz to 10GHz TEMPEST filters for your mains input, and get a big reactor for your entire house, and run everything with 4 gauge welding wire, remake all your instrument chassis with 3/8 inch plate iron with copper blast welded to it, send out HARM missiles to destroy all nearby radars and radio stations, break into your neighbors houses to install linear regulators, start hacking appliances to run on linear regulators, add house wide water cooling to cool all your appliances, and determine if a huge shield around your house is counter productive because of parasitc capacitance (might be better to destroy all other transmitters then shield......)

Probobly better to install some kind of AC to DC converter on the power pole far away from your house and run everything on DC right? Why put a RF feed going to your house?? 


I think it might be a good idea to seek out a planetoid made of pure copper... before I end up starting a reign of electrical terror

Then half the time you probably wanna go into deep interstellar space far away from everything to conduct experiments without large nearby parasitic ground planes anyway... or just start making your own universes to satisfy physics requirements

Sometimes I think about the electrical situation in the world and it kind of makes me feel like trying to read a book at a techno party or mosh pit

'hey power company, there is an excessive 0.000002 Hz noise peak on my mains at 37aVrms, can you fix it??'

then in the end you end up using 3.5 digit multimeters and budget arduino boards in your low noise lair 

[rhb]

Here are a couple of photos of the almost complete unit.  I still need to attach  a plug for connection to the mains and actually connect all the whips.  The box in the air is for the master kill switch.  I'm going to swap out the box used for the 90 with a regular handy box and install an unswitched duplex outlet to power my GPSDO which uses a wall wart,  I got an ECO after I finished the conduit work ;-)

Happily everything just fits in the available space.  So it was worth the hassle of having to special order stuff.  But I've got 230 ft of SL-516 and 35 4SS boxes extra :-(

Pushing 3 pieces of stranded 12 AWG THHN threaded through that 5/16" conduit was a chore.  I'm glad it didn't need to be longer. It went fine until I got to the last foot. In retrospect I should have run a fish line.   If I need to expand it I can go up with 1/2" EMT from the box that the power cord enters.

My original plan was one run like this for each shelf, but with only 3 instruments on the bench I decided that was excessive.
Once this is in place I'll connect the spectrum analyzer to the power line and see how much noise there is.

[TerraHertz]

I hope you realize that spiral metal flex conduit is intended for mechanical protection mainly, and isn't rated for RF shielding at all? Since there's no guarantee of how well the metal spiral is making electrical contact longitudinally. It's really just a big helical antenna, with intermittent shorts on a lot of the turns.
If you actually want RF shielding, braided wire cover is the only way.

Also, what are you going to do when your number instruments exceeds the number of IEC cords you've wired in from your wall metalwork? About an hour after you finish the installation.

As for EM radiation entering or leaving instruments via the mains cord, is there any gear at all that doesn't have an RF filter block on the IEC entry point?

An easier improvement might have been to run a nice fat copper grounded bus bar along the back of the shelf/bench, with lots of screw holes for ring-lug attachment of short, fat ground wires to a chassis ground point on each instrument.

If you have external  sources of RF strong enough to be a nuisance in the lab, the only way to deal with that is to shield the whole room, with copper/brass mesh or foil stuck to the walls, and sealed well around doors and windows.

[Brumby]

Also, what are you going to do when your number instruments exceeds the number of IEC cords you've wired in from your wall metalwork? About an hour after you finish the installation.

I'd call that an evil thought .... if it weren't true.

[IanB]

I'm disappointed.

The last "OTT" power feed that I saw was a large (several tons) transformer with an 11 kV secondary routed around the walls of a power supply room to amplifying equipment that needed lots of power 

[rhb]

Um.  Have you heard of magnetic fields?  I've not heard that those were blocked by  conductors. But low carbon steel is the most effective option short of mu metal which I definitely don't need and can't afford.  If high frequencies are an issue that I can't squelch at the source, I'll wrap the flex in conductive foil tape.

It's trivial to expand.  It's not as if there is more room on the  benchtop and shelf  it is designed to feed. If I want to feed another shelf all it takes is 8" of EMT to go up and another 5 ft  horizontal run to feed two more shelves.

If I need to expand it I can go up with 1/2" EMT from the box that the power cord enters.

This all comes under the heading of:

I need a better way to power 10 instruments than using a power strip and a large wad of power cords.  The flex is mostly gratuitous, but easier to do at the start than later.

I had been considering running BX, but I found a 20 A EMI filter Dad had.  I haven't swept it yet, but if it looks as if it will do the job,  If so, I'll skip the BX and just put the EMI filter in a box between the instruments and the mains feed.

I'm less concerned about conducted EMI on the mains cord than radiated EMI from the device.  My Instek 2000E scopes are horrible in that respect.

More recently I have become aware that Chinese test gear may produce no EMI in front but  lots in the back and sides which will be picked up by any power cords running nearby.

BTW Have you ever priced copper RF screen  material sufficient to screen a 10' x 7' x 8' room?

Sigh... I'm happy I don't need 11 KV or even 11 KVA

[TerraHertz]

I'm less concerned about conducted EMI on the mains cord than radiated EMI from the device.  My Instek 2000E scopes are horrible in that respect.

The first lunchbox scope I ever used (a Tek, can't recall model) used ccfl backlight tubes in the LCD. With a plastic case and absolutely no shielding of the several KV 40KHz power to the tubes. There was no point trying to do any mV level signal testing anywhere near the scope, front or back. My opinion of Tek nosedived.

Do current cheap lunchbox scopes use LED backlighting? (I hope)

BTW Have you ever priced copper RF screen  material sufficient to screen a 10' x 7' x 8' room?

Actually, thin copper foil is not toooo bad. And you can solder the seams. I didn't bother doing that in my workshop because there's not an interference problem. Plus the building has aluminum foil moisture barrier in the walls (well overlapped) and sheet metal roof, with continuity of the foil up to the plane of the roof.

But, I once passed up an opportunity to buy a huge roll of brass wire flyscreen mesh, at a local hardware store going out of business sale. Kicked myself ever since.

[CopperCone]

https://www.electriflex.com/shield-flex-flexible-conduit/

graph



The dotted line indicates a comparison to standard unshielded liquidtight flexible conduit Type LA.  The spectrum of test frequency is from 1 MHz to 10 MHz Electric Field, to 100 MHZ to 1 GHz Planewave Field and 2 GHz to 6 GHz Microwave Field.  Tests were performed per MIL-STD-285 on 1″ trade size conduit using standard liquidtight fittings from Thomas & Betts series 5300.  Results are based on controlled laboratory conditions and may vary in actual field installed conditions.

LA conduit:
https://www.electriflex.com/products/type-la/

So armored cable is not horrid, but its horrid compared to a real solution

slipping a braid over it iwll give you 10-20dB, while buying their best will get you 100dB at 700MHz

[DaJMasta]

That graph really threw me for a loop initially... seeing attenuation go up instead of signal level go up on the Y axis just seems strange 


I don't have experience with this stuff, but are there power line cables that use a twisted pair for hot and neutral for lower noise that's picked up on the line?  I guess that may not even make a difference unless the stuff in the walls was the same, though.  Also are there external common mode filter modules that are right on the end of the IEC cable that could help with something like this, either in addition to or in place of shielding the cables?


While living near a powerful transmitter seems like something I really just want to avoid, it is sort of fascinating to me how equipment and systems are effected by it.

[rhb]

Thanks for the link.  I was going to retort, "And have you priced it?".  But eBay came up with 1/2" at $9.80 for 100 ft plus $22 shipping .  It's not flexible enough for my power strip though.   For the power strip I'm really just trying to suppress SMPS noise so the SL-516 should be fine.

https://www.ebay.com/itm/ELECTRI-FLEX-LIQUATITE-EMS-11-100-WIRE-CONDUIT-1-2-78811-RFI-SHIELDED-COPPER/183242913388

None of the usual suspects stock it, but I'd guess it's many dollars per foot.

[CopperCone]

keep in mind the LA stuff is the baseline they show. I don't think it differs from the armored cable you use, it is just rubber coated.

I would suspect that a unarmored but braided cable would give you better performance in the RF region past like high KHz, while the armor cabled will give you better performance in the magnetic region, while the solutions listed on that graph (other then the grey line, which is what you have already).

[CopperCone]

I am also thinking about what you made, I am not sure what the hell is going on electromagnetically, it would probobly work better if you terminated the metal bit on the chassis, but its complicated because of the coaxialish helical antenna ground loop thing that would be formed. Their solutions just meant to go to a junction box not between the junction box and the thing being powered.

i kind of wonder if your way of doing things may result in a some bizzare peaks or resonances, because its not a true double shield.

it is confusing because there is already a shield on the IEC cable right, so your running a shield through a partial shield? i confused myself and I don't want to work on this on saturday to be frank

and it can also be more confusing depending on how your grounds are arranged, like do you terminate the shield of the conduit/armoreed cable to the ground wire, or do you use separate ground wires (to absolutely minimize the number of interconnects and failure points and minimum possible saftey impedance) going to the electrical ground?

like does the ground wire from the 3rd prong on that cable, go to a twist nut that connects the junction  box, or does it go strait through to the panel per each outlet? I believe that going strait to the panel is preferred for safety but not a requirement, but I don't know the electrical code where you are).


Home wiring drives me fucking insane, I am waiting to bite the bullet and rewire the whole fucking house, but I don't know what I should do really. Right now I got a mess.

I need to layout different options and compare them and perform some measurements on the stuff I got now, I believe now I routed 2 seperate cables going to the box, for 2 outlets, but for some rason i get a 50 volt ghost reading if i plug in a multimeter between the two outlets, and i can get zapped from equipment connected to different outlets if i touch the ground, idk if its the GFIC or a ground loop or what, the amount of work required to tear apart the wall, repaint, run the wire, do the sheetrock, etc is too much  ]

I tried to star wire it as much as possible when I did the electrical for this room, but i got that bizzare fucking problem that needs investigation. I am dreading it.

i suspect that if you look at home wiring requirements from a low noise rf prospective you will not be happy

[T3sl4co1l]

Interesting, but, how are you grounding the conduit to the equipment chassis?  I've seen very few boxes with RF fingers on the receptacles.

Without that ground bonding, it's just a resonant stub with a sheath over it. 

It would at least be better to have the cord's ground bonded to the conduit at the end of the conduit, so that carries ground into the receptacle; but that's still quite a length of unshielded wire.  And the ground lead is rarely straightforward in RF terms: sometimes it's bonded to the chassis inside the receptacle, sometimes it's a wet noodle looped around until finding a lug and shakeproof washer somewhere deep inside. 

Tim

[rhb]

Typical IEC cords are not shielded.  The conductors form a spiral in the cord.  I cut off 12-13 of them.

Ground should run all the way to the panel.  The box  should also be bonded to the ground and that will in turn bond the conduit.  That is NEC practice.

You probably don't need to rip out walls,  But you do need to check the outlet wiring.  First thing is to get a 3 neon lamp plug and go around and test all the outlets for wiring errors with that.  Then get a line tracer from Harbor Freight and identify which outlets go to which breakers.  Don't rely on it telling you which breaker.  They get close but are not exact.  Use the loss of signal to prove which is the correct breaker.

Many years ago I was detailed by my dad to put screws into the sides of the room HVAC units at his motel so the front panels  would not fall off when the door slammed.  At one I got zapped by my grounded drill.  I went and dropped the breaker for the room and started pulling the 220 outlet apart to check the wiring when KERPOW!  Turns out the outlet was wired across two different breakers.  So I had only dropped one side.  I don't recall why the ground was hot.  I didn't get shocked, but a 120 V dead short when you have your hand in an outlet box will certainly be memorable.

The point of the steel conduit is low frequency shielding, not high frequency.  SMPS noise is around 30-50 KHz.  I'm mostly concerned about the noise from an Instek scope.  But I've also got multiple computers and monitors in close proximity.  Mostly the point is not to have this large wad of cords.  If I want to feed more shelves it's an easy connection from the box that forms the 90 for the switch.

I finished the setup, but made the mistake of using 12 AWG stranded which was a bitch to get in the Appleton 4SS boxes.  Being somewhat OCD, I'm going to redo it using 16 AWG.    I'm a bit uncomfortable with the connections mashed into the boxes the way they are.   The peak draw if I power everything up at once is 6.5 A and 3.8 A running.  I should have measured the load before I started. 18 AWG would be adequate, but I have several spools of 16 my dad left me along with several pickup loads of boxes, conduit fittings, etc.

Someone cleaned out the eBay seller.    It's now listed at $145/100 ft  which is probably still cheap.  While it has stellar performance it's not flexible enough for this application.

The photos show the finished power feed and the installed  There's a single screw into the shelf inside the switch box.


Edit:  In considering CopperCone's wiring problem this morning, the most probable cause is the hot and neutral leads being swapped on an outlet.  It's an easy mistake to make and if you don't check your work will go largely unnoticed unless someone gets a shock.

Here's a link to the type of tester I mentioned

https://www.amazon.com/Power-Gear-Receptacle-Indications-50542/dp/B002LZTKIA

They are available at most hardware stores.

[CopperCone]

Well I got my hands on a neon ground tester, it looks like there is an open ground for one of my wiring runs.

I just need to find out if I fucked up with my triple checking or if this bozo that told me he would take care of it failed to wire the distribution box right.

[rhb]

That would explain the 50 V reading.  If the ground were not connected at the panel, the capacitance would induce a voltage with very low current capacity.

It becomes a real pain if the fault is in a junction box that is inaccessible.  Locate all the outlets on the same circuit and visually inspect them to see that the ground is carried through each one in the chain.  You may be able to locate the fault with the neon tester.  The fault will be the last box in the string which has ground.

[CopperCone]

I don't know if the current capacity is particularly low, it hurts quite a bit if I touch two instrument chassis connected to two different outlets. Thinking about it,  I figure that if the ground was not connected at where I did the wiring, the parallel floating plate capacitor formed, would be rather small.

I don't have a tactile feel for this though.

I basically wired up this room when I remodeled to have a separate cable for each wall, and 3 walls are wired, so I basically know what cable it is.

I did not want all the RF amps, switching high current supplies, computer, bench lights and heavy equipment running on the same outlet as things like 6.5 digit meters, electrometers, spectrum analyzers, etc. I made something along the lines of a star ground.

I just need to get some work lights and clean a bit so I can work under the work benches comfortably to inspect it. I am 90% sure its not my fault but I need evidence.

[kitkatbar]

BTW Have you ever priced copper RF screen  material sufficient to screen a 10' x 7' x 8' room?

Actually, thin copper foil is not toooo bad. And you can solder the seams. I didn't bother doing that in my workshop because there's not an interference problem. Plus the building has aluminum foil moisture barrier in the walls (well overlapped) and sheet metal roof, with continuity of the foil up to the plane of the roof.

But, I once passed up an opportunity to buy a huge roll of brass wire flyscreen mesh, at a local hardware store going out of business sale. Kicked myself ever since.
[/quote]


Sometimes we miss out on golden opportunities - no need to kick yourself - believe you may be able to purchase pretty cheap online or on ebay perhaps - what kind of opening are we looking at when it comes to this copper screen?  Have you guys tried local scrap yards?  I have some 100 x 100 (100 wires per inch - www.bwire.com - this is the best measuring guide..)

I believe this screen is too "tight" to get the job done - the wires are so close together, the opening are almost too small to even see through.  Whats the opinion out there on this medium being a practical option for me to use when shielding?

[T3sl4co1l]

As long as the wires are electrically connected together, finer is better.  If you need to see through it, obviously, it's going to be a bit of a challenge there...

Even just putting up chicken mesh is a good start for longer wave radiation (VHF and below, say).

Tim

[rhb]

1/8" mesh hardware cloth did a fine job screening my Z400 PSUs.  I have to get my scope probe very close to detect anything whereas before the mesh I could see it from a foot or more.

Wide aluminum flashing with aluminum tape on the seams ought to work and be fairly cheap.  Use a small piece of copper foil to make the ground connection.

I will eventually have a shielded room, but lots to do first.