6

I am building a small (6U) self-contained networking rack in an enclosed case which will accept only power and coax from the residence. The only external connections on the rack are an IEC C14 inlet and an F-type coax connector.

Inside the rack case, I have various pieces of equipment like a UPS, PDU, and patch panel, which all have grounding terminal screws on the back. My plan is to bond them together with the rack rails functioning as a bonding busbar, and finally to ground the bonding system using the ground conductor on the C14 inlet.

Based on my understanding of the design and intent of these systems, this will ensure that if a short occurs with the rack case or any of the cases of the bonded equipment inside of the rack, this voltage will have a safe path to the building's electrical service panel via the wall receptacle ground. (Be assured I will test continuity between all parts of the bonded system.)

Have I correctly interpreted the best approach to grounding for this build, given that the finished case is not to be installed in locations where a separate building ground system will be available? Or is have I misunderstood everything and it would actually be safer to not bond or ground any of the equipment in the rack beyond their regular 3-prong grounded power cables?

NReilingh
  • 165
  • 6

3 Answers3

10

Yes, that's correct.

What you Really don't want is for a device to have a hot-ground fault... and then, the chassis of that device is now floating at 120VAC from ground. And because it's in a rack, the chassis of all the other devices are floating too. And then you touch anything, and BLAM!

But if you are using UL listed* equipment (as you should), this should all be done for you by the equipment.

* or competitor NRTL, such as CSA or ETL.

Harper - Reinstate Monica
  • 300,628
  • 26
  • 286
  • 734
3

Are mechanical bonds required for proper grounding or safety of rack mounted equipment that is connected with 3 wire (grounded)connections?

The bonding of the exterior case in this case is more of a shielding than a safety.

Most rack mounted systems have faraday shielding this is to block the radio frequency interference not as a safety.

My most recent example of this was a properly grounded system that was picking up noise from the adjacent rack mounted equipment to the point it delayed the opening of a surgery suite that had been completely remodeled it is powered by a UPS (very clean power) the mfg added shielding to the device and this eliminated +90% of the electrical noise the bonding was the same the only change was the shielding that was added.

So the bonding of rack mounted equipment is not so much a safety for electrical shorts but to prevent or reduce the emissions or reception of radio frequency interference.

Ed Beal
  • 102,735
  • 4
  • 75
  • 155
  • You can get shielded racks, but common racks are not shielded, and bonding is only to ensure equipotential. – vidarlo Oct 17 '21 at 11:23
  • @vidarlo, all equipment must meet FCC guidelines for emissions, however with multiple pieces of equipment mounted in close proximity the shielding is needed it has nothing to do with electrical safety , but many internet only electricians do not have the background to understand this , thus the reason I provided the real life example. – Ed Beal Oct 17 '21 at 21:09
  • Stop it. Go look up some more or less random rack cabinet. There's not a *single* mention of noise in the manual - but there's plenty of references to safety earth. And yes, I do work with instrumentation, I know a thing or two about noise. As I said shielded racks do exist - but the primary role of the normal stamped metal cabinets is holding servers, not providing shielding. – vidarlo Oct 18 '21 at 05:03
  • As a pro I see this very issue and provided an example , but you can blindly follow , I guess the reason why licensed electricians don’t hang around, just saying! – Ed Beal Oct 18 '21 at 19:17
  • Yes, you claimed that the bonding was to provide shielding, *not* safety. Sorry, that is plain wrong. As shown in the manual the *main* function according to manufacturer is safety; equipotential bonding. Furthermore, ad hominem attacks that is plain wrong leaves something to be desired. For what it's worth, I'm certainly what would be considered professional in the electrical trade as well. So far, you've not backed up the claims of your answer; only attacked me. – vidarlo Oct 18 '21 at 19:42
  • Let’s compare state certified electrical license’s , you state I am wrong , parallel grounding paths is a code violation the frame grounding is for RFI. – Ed Beal Oct 18 '21 at 21:30
  • Sigh. It's not parallel grounding paths. It's equipotential bonding of exposed metal parts. It's not RFI; the manual states it's protective earth. To provide RFI dampening, metal gaskets would be far more effective. So far, you've repeated your claim that it's mainly for shielding purposes. Furthermore, the world is bigger than the US. – vidarlo Oct 19 '21 at 05:36
  • All I'm asking is that you provide some references to substantiate your claim that the bonding is primarily to provide EM shielding, as you in your question state Most rack mounted systems have faraday shielding this is to block the radio frequency interference not as a safety. - which in my opinion could be borderline dangerous assumption. – vidarlo Oct 19 '21 at 05:39
3

Presuming you're in the US where it is possible to have non-grounded power sockets, I would suggest:

  1. hard-wiring the input lead to the rack case, and to a 3 pin US style plug that cannot be removed.

  2. clearly label the outside that a 3 pin earthed supply is required for reliable functioning of the unit

  3. Add the same information into any paperwork/documents/agreements.

You still can't stop some end user from cutting the plug off and wiring on a new one, using a ground-stopping adapter, or some other shennanigans, but you're now on a better footing if they do.

Naturally you'll use 3 pin grounded connections inside the case and have it all secured down against vibration. Also put a good tamper-evident seal on any access panels.

Criggie
  • 10,616
  • 2
  • 24
  • 70