STP Cat 6 patch drain wire connection(s) to metallic shell?

Usually only connect a drain wire to shell at one end, to stop ground loops. Some may disagree.

Disagree!

Ground both ends, the equipment ground is referenced to earth.

Cheers

Empirical data taken from reality-based commercial product agrees with you. Loopy LARP-official intelligence believes it's best to bond both ends.

Both ends for high-frequency noise sources. One end (often the source) for low frequencies.

Note that you also have to consider the fact that the "ungrounded" end might happen to be the only *connected* end of the cable -- in which case, nothing is grounded.

If you have different ground potentials on the two ends of a cable, you likely have a bigger problem to address.

Ground everything as often as you can.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I once visited a research lab in LA and was having severe problems with mains interference. It turned out that the ground potentials on opposite sides of the room differed by about 50V. Powering everything from sockets all on the same side completely solved the problem. Probably the ground connection on one side was floating and leakage currents from all the equipment was producing the measured voltage.

Much more recently in the UK I came across a problem where a couple of PCs were simultaneously crashing. In that case the PCs were connected to one row of floor box sockets and the monitors were connected to the adjacent row. The floor boxes were part of a structured power distribution system and they all radiated from a single distribution box about 20m away. The problem was solved by cross-bonding the floor boxes. There was enough inductance in the ground connections to inject a noise spike into the PC motherboards through the HCMI connector shield when another device was switched on. John

In the US, "earth" is bonded to neutral at the load center (and to physical earth nearby; and also at the utility's control). The (metallic) case of most equipment is connected to earth in its power cord.

No current is intended to flow through the earth connection.

So, having shields (connected to chassis connected to earth) should never have a potential difference -- save for lightning strikes and other such events that affect "The Earth".

In the US (esp residences) the neutral exhibits such behavior as it can often be a long run to the load center and there is nothing preventing a consumer from switching on a ~KW load at the far end of such a run.

When the thermostatic control in SWMBO's (clothes) iron toggles, I can notice a change in the intensity of incandescent lights on that same branch circuit. Because she hasn't considered what else is on said circuit nor its location at the far end of the house.

By contrast, I can operate several KW appliances in the kitchen, concurrently, without a similar manifestation.

Most older US homes weren't designed for the types of loads they now see. E.g., it was common to run a branch circuit to feed multiple bedrooms -- before every bedroom had a computer, TV, etc.

I have a 60-drive (4x15) shelf array in my office. I have to be very careful to ensure not much else is powered on when I bring it into service -- even though the drives stagger their spin up. So, I only bring up all 4 shelfs when testing that software (it characterizes, wipes and installs software images on multiple drives concurrently; most of the time, it is sufficient to test it with just 15 drives attached)

Triac dimmers can be very sensitive to line voltage, and when dimmed down sometimes have on/off hysteresis. Turning something else on can shut down your lights. It's worse with LED lights.

Putting one incandescent lamp on a string of LEDs can help.

PC power supplies notoriously pull electrical systems out of phase. In my case, a mere two meter long power line once created problems.

As an aside, STP is overkill unless it connects two shielded ports separately connected to the common ground of respective separate appliances. In any other situation STP simply serves as a substitute for lower quality patch cables.

Telluric current mysticism may make the word "ground" evoke mythological imagery in the minds of many electronic engineers. It's expected behaviour because, as Bill Whitlock says, "Ground [is] a FANTASY invented by engineers to simplify their work. The 'uni-potential' fantasy assumes all ground symbols in schematics are at exactly the same voltage." Meanwhile, ham radio grapples with four distinct flavors of ground: safety ground, dissipation ground, RF ground, and common ground. It's a big mistake for me to try to treat all four grounds in a monolithic manner. Your Mileage May Vary.

Even more interesting where not all the sockets are on the same phase.

For me, where the ground is connected depends on application. For source that could radiate though the cables, the ground is at the source only, and open at the receiving end. For unbalanced audio, again, only one end is grounded to avoid line earth loops. For balanced line, ok to ground both ends, but could still get induced currents if there is significant volts between the chassis at each end.

Networking is another case, in that most network RJ45 connectors are all plastic, so the earth foil drain wire has nowhere to go anyway.

So, no hard and fast rules, it depends.

The connectors for shielded cable do have a metal shell which makes contact with the socket body, but the method of connecting the cable drain wire is not very satisfactory.>

In my case, part of the plug's metallic shell stays outside of the metallic data port after its snapped into place. So deft hands simply soldiered the drain wire to the permanently exposed metallic shell surface of the plug.

We had a product in a system where our box talked to another box over Cat5 S/STP cabling. It failed a formal EMI test because someone bought cheap cables with flakey grounds in the cable end RJ45s.

Always ground everything!

In my experience, emc testing and compliance can be a bit of a black art, and can require what seems like irrational remedies. Usually very product specific, and is difficult to generalise.

have done quite a bit of work around RJ45 and connectors for networking and other, and none of it seen in the past few years had any groundable metal shell. All plastic only, though cabling is balanced, 4 pairs, so the drain foil should proved some level of shielding, ground wire connected or not.

If the shield isn't grounded, it's a radiator.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics