Al Dykes wrote in part:
Fully understood and agreed. However, grounds can be noisy, and will enter the sheath when/if the 100baseTX unused conductors (pairs 1 & 4, blue & brown) are grounded by the NIC.
I don't think that's entirely true. Look up "Banana Skins" on a British EMI magazine. There's also staggering quantities of STP being installed in the EU to avoid EMI certification. Why would they?
-- Robert
peancor wrote in part:
Something is different from that electrical outlet. Either power or ground less noisy.
As other posters have correctly pointed out, low frequency isn't going to be a problem. But any high frequency, high power switchers can be. Those VFDs might not have high primary shaft speed, but that does not correlate to the switching frequency which might easily be in the kHz.
-- Robert
A google search for "emi Banana Skins ethernet" comes up with 7 hits, none of which seem to have anything to do with TCP over CAT-x wiring plant.
People have wrapped loops of CAT3 cable around arc welders, RF transmitters, induction furnaces, and lots of other equipment that generates EMI of one sort or another. Tests of IP error rates are near zero no matter what you do.
Newer CAT-spec cables are made to tighter tolerences and even more resistant to EMI-induced currents.
It would not be surprising for a mispaired cable.
-- glen
Al Dykes wrote in part:
A little restrictive search. I meant the "Banana Skins" column in the British EMC Compliance Journal:
Sure. But IP (and TCP even moreso) are remarkably robust to errors. The ethernet checksum catches many at a lower level.
A surprising number of networks operate surprisingly well for years with split pairs and packetstorms. People just say "the network is a little slow". Then some traffic pattern change happens, the network becomes _extremely_ slow and we get called in.
-- Robert
Do you use shielded on unshielded cable for the Ethernet connection ? e
If you use shielded Ethernet cables and you have high enough current flowing on the Ethernet cable shield this could be a problem.... can cause problems to sensitive electronics on either end of the cable.
True.
RFI/EMI and ground-induced interference with properly-wired Ethernet equipment is effectively non-existent in the real world when you use unshielded Ethernet cable for thw wiring.
When shielded cable (STP) is used then we have a system that can be sensitive to ground-induced interference.
Tomi Holger Engdahl wrote in part:
There sure could be! There's not supposed to be _any_ current flowing through the shield! If there is (and sparking connects aren't unknown), then there is a grounding problem. Likely disbonding plus a ground fault current. Electricians job.
In severe cases, one recommendation is to only tie one end to ground, or to tie the other through a resistor or capacity.
-- Robert
Tomi Holger Engdahl wrote in news: snipped-for-privacy@pippuri.niksula.hut.fi:
The cable is unshielded.
OK. Thanks. Got a ethernet-specific citation for EMI on STP? We rarely use in here in the states.
Is the OP working on a UTP or an STP network?
That will happen when both ends of the shield are connected, which I'm sorry to say happens a lot.
Rick Merrill wrote in part:
It does, and is theoretically best. Ground-induced interference should not happen because the ground system should have zero resistance and zero current within practical limits..
However, this is sometimes untrue. Then the stopgap advice is to ground one end and the other through a resistor or across a capacitor. But the grounding really should be fixed.
-- Robert
Once you get distances over a few 10's of feet getting a measured 0v potential at all grounds is nearly impossible. Unless you put in something like a 1/2" thick steel floor or some such and use it as a common ground. But that would introduce other issues.
David
DLR wrote in part:
Why? Are you saying that if I pull the main breaker there will still be a ground potential difference over short distances? (Yes, I'm aware there is about 3V across the Chunnel).
AFAIK, most ground faults are stupid things like conduit/tray corrosion and inadequate fasteners and grounding jumpers. Even these cause no trouble until there is a ground current from occasional induction or more frequent machine fault like a poor welding clip. But I'm no electrician!
Yes, these things happen and dcom is blamed. But it is really the canary in the coal mine. Something else is amiss. And if the location really is that EMI noisy, perhaps fiber really should be run.
AFAIK, STP is mainly popular in Europe because it eliminates a code-driven requirement to test new installations for EMI emissions. Of course it works the other direction too. But there are limits.
-- Robert
Brand new factory - no problem. Old factor with many mods - watch out!
I have seen ground differences in factories that draw a LOT of current, even on 3-phase. The largest (decades ago) was 48v!
I'm guessing you're from the right side of the pond so specifics do not apply but principles do. Even with current US standards you can get current flow through grounds in new construction after you hook up all the copiers, toasters, lamps, etc... And in older buildings, well watch out. And in much of the US while it is a VERY recommended practice to have all services that in any way shape or form come in to a building at a common service entrance and have a unified grounding point ... Well even new construction with this can have ground potential. You just have to plan for it. And with older buildings where the electrical, phones, cable, come in via various point and have separate ground established back before broadband, gig E, etc... were even dreams, well things can get to be a hassle. And with multiple buildings with separate service or worst yet 1 building with multiple electrical services, well things can get weird.
So you plan for bad grounds and go from there.
Basically I got to study the issue a lot when a client moved into an old warehouse district and finished out the spaces. The worst was one of the bays was grounded to a municipal water line. Except the water mains had been dug up a few years earlier and new meters and service established. We had no idea if the line being used for ground stopped 1" outside of the building or went 10s of feet under the street.
We EASILY got 5 to 10 VAC between the phone ground and the electrical service ground. Oh, yeah the phone DMARC was in the next bay over on a separate ground from anything in "our" bay and the phone company had absolutely no interest in our requests to change this setup.
We now have several areas with fiber between them and runs of ground cable from the electrical service to where other services come into the building. Lots of little quirks vanished after that.
Interesting. Just what are the EMI limits protecting? People or radio programs?
David
DLR wrote in part:
Why? all of these devices should be properly insulated and current flow in through hot and out neutral. Zero to ground. Now if you have the phases mis-loaded, you could get some potential difference and current between the neutral--building's groundstake and the transformers. But mis-loading phases is also a fault.
You shouldn't be able to get an occupancy permit without this.
Why study? Quicker to drive a code groundstake. Preferably straight through their stinkin' plastic pipe :)
So put in your own protector tied to the proper groundstake.
That is also a good solution
Both, I believe. Life-critical devices in some cases. I have seen snow on NTSC ch 2 from a short Cat3 carrying 100baseTX error-free.
-- Robert
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.