With the BNC, SFAIK that's a single-ended connection - the lead is in the middle, the coax is the return.
With an XLR, you may want to float ( open ) one end of the shield for a balanced connection. There may be a case ground potential difference from box 1 to box 2.
The only question is which model works best for USB.
formatting link
I see a D+/D- pair. And a ground. Looks balanced to me. You may want the shield connected only at one end.
Given all this, USB cables are not usually that long, so the chance of any case ground potential is pretty low. But you never know.
Some aplication notes even recommend this - in the form of a ferrite bead in the USB ground connection. however,it seems that such ferrites are a common cause of USB compliance failures as they also slow down certain state transitions which use common mode signalling as well as the differential mode which is used for ordinary data transfer.
For getting through EMC compliance testing, bonding the grounds together as solidly as possible works well.
Take a look at any 10Base2 Thin-Ethernet card. The connector (cable shield) is isolated from the metallic card frame and the computer frame. Also when building 10Base2 network, the cable shield must be grounded at exactly one point, usually at the other terminator. An unisolated expansion barrel connector touching some grounded support structure may ruin the network, now that we have two grounding points and hence ground loop currents.
As far as I understand, in the US, the ground and neutral are connected together at the service entry as well as connected to a physical grounding electrode.
As long as you connect two devices powered from the same service entry, the ground potential of the data cable shield ends is the same.
A power supply connected through an EMC filter will spew interference current into the ground connector through the Y-capacitor, thus causing unequal voltage drops in different ground conductors, thus, there is some high frequency interference current flowing in the data cable shield.
Not directly of course, but look at the big picture.
Take a look at en.wikipedia.org/wiki/Earthing_system and especially TN-C-S
Do you expect that the Neutral (N) and protective ground (PE) wires would run the whole way back to the power plant. That would be a true TN-S system with total of 5 wires. In practice the distribution system is some form of TN-C-S, in which the PEN (combined protective earth and neutral starts at the distribution transformer star point. This PEN conductor is split into PE and N at some point towards the customer.
This splitting point could be at the main distribution panel, in any sub distribution panel or in apartment specific distribution panel. It would not at all be strange to have a common PEN conductor in the raiser cables in two separate stairs and hence different ground potential between stairs.
Some older regulation allow the PEN to PE and N in each mains socket. Thus, the equipment mains cord, is the only -S part in TN-C-S
If the equipment signal ground or shield are connected to the equipment local PE ground, there may be ground potential differences between devices even in the same room, same apartment and at least between buildings.
** Amazing how most of the worlds's microphones manage to work so perfectly - ain't it ??
FYI:
Connecting a cable's shield *directly* to metalwork at the signal receiving end diverts RF energy harmlessly away - this has been standard practice with audio and RF devices for centuries.
Connecting the metalwork of various audio devices that are all linked to a common safety earth can sometimes result in ( low level) hum loops - even with all balanced set ups.
Those who do *not* deal with a particular area of technology have no right to pontificate about it.
10b2 is not USB. It is transformer isolated, and was notoriously unreliable.
At one point. But there should be no current in the ground wires, hence no ground-ground voltage differences. And difference will probably come from magnetic coupling from line currents, which will be small and relatively high impedance. The shield of a short USB cable, connected at both ends, can significantly short out, and reduce, and ground-ground voltages.
There's also high frequency stuff, like spikes and ambient RF. A cable with one end connected is a monopole antenna. The open end can have serious amounts of RF on the shield, especially at resonant frequencies, and that is coupled hard to the four wires inside. Ground the shield at both ands, and the RF goes away.
All the more reason to ground the shield on both ends.
There's yet another reason to ground the shields. The shield forms a transmission-line transformer with the wires inside it. So delta-v between the shield ends becomes (at higher frequencies) delta-v in the wires, forming a common-mode-rejection transformer for free.
Ground on both ends.
--
John Larkin Highland Technology, Inc
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Not quite. They are tied together at the service entry and the grounding electrode is in tied at that point. They are also tied together at the transformer (center tap) feeding the building.
There can still be a significant difference between the two, particularly under fault conditions. The AC difference can be substantial (and ruin your EMI compliance day).
It's a balanced connection anyway. It's a three (or four) wire connection. The problem is the length of the cable means that it's possible for it to cross power domains.
Because there may be 50-100V on the shield, and not just some capacitive coupling (I've seen it). Connect the two together and you'll see sparks and smell smoke. XLRs are generally connected through isolation transformers. For a reason.
But safety (of both the user and the equipment) trumps making your EMI compliance easy.
You snipped the critical point. There is little chance of a dangerous ground differential, so ground both ends.
If you break the shield at one end, you convert that huge current into a huge voltage... coupled directly to the wires inside the shield. Fun things will then happen inside the box.
Few of us operate USB gear inside a big transmitter. If we did, the best arrangement would be to ground the shield at both ends and add clamp-on ferrites to reduce the current in the shield.
--
John Larkin Highland Technology, Inc
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
If there's low impedance 50-100 volts between power outlet grounds, the building wiring is defective and dangerous. And it will blow out the USB electronics.
If that 50-100 volts is high impedance, like from capacitive coupling, a continuous shield will short it out and protect the electronics. As it also shunts RF to the chassis on both ends.
Even transformer coupled audio signals have some finite common-mode rejection at the receive end. An open shield is an antenna that jams cm voltages onto all the wires inside the cable.
How is an open shield safer than a continuous one? Which is safer for the USB target chip:
Shield connects first and zeroes out any ESD, ground loop voltage, and EMI
or
Let the signal wires handle all of that.
Can you buy a USB cable, or an XLR audio cable, or a shielded RS232 or CAT6 that has the shield connected at just one end? How about a BNC-BNC coax that has a disconnected shield on one end?
Right. Ground the shield at both ends.
--
John Larkin Highland Technology, Inc
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
That discussion was about XLR (audio). As I noted, USB is a different argument completely.
That's why they're transformer coupled.
You're confusing USB with audio (XLR). Different problems, different solutions.
XLR cables are usually grounded only at one end. It's common to unground the shield on RCA cables, too (particularly at low-level sources, like turn tables).
Is BNC differential? What are you trying to accomplish. A solution isn't very useful without a state problem.
That's a good idea, unless it causes more serious problems. It often does. One size does *not* fit all. I generally ground chassis through 1206 sorts of devices so I have "knobs" to turn if testing doesn't go well.
ElectronDepot 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.