Vintage Computer Festival wrote: I would greatly recommend using a signal ground wire (pin 7 on RS-232). You can get weird, unreproducible problems without it. It can cause really hard problems to trace down. The ground between two different boxes can float about enough to cause signal reliability problems. The float can sometimes vary enough to generate random character streams. The longer the distance between the boxes, the more likely you will have problems. But you normally don't want to connect frame grounds (pin 1 on RS-232 cables iirc). You sometimes get an amazing amount of current through the cables, even enough to melt them.

On 6/21/05, Paul Koning <pkoning at equallogic.com> wrote: The key word is "should"... If you get some ya-hoo who decides they can save a few $$$ and wire up an outlet... well... let's just say that the first homeowner's tool I bought was an outlet tester, and I found two outlets that had hot and neutral reversed... I worked at a place where the owner plugged a serial cable in between his PC and his Mac, and he roached the Mac serial port because of the same problem. He had all the outlets tested and fixed, but he also made a proclaimation - no computers were to be connected via serial cables unless they pulled their power from the same wall outlet. A bit inflexible, but guaranteed not to have a mismatch. -ethan

Bj??rn> On 21 Jun, 2005, at 12:35, Ethan Dicks wrote: Bj??rn> In Norway, you will not get mains-powered devices approved Bj??rn> unless the chassis is isolated. Our AC wiring does not have Bj??rn> any concept of "neutral", since both wires are supposed to be Bj??rn> at the same potential relative to ground. That way, everybody Bj??rn> learns to be careful. That doesn't address the point I was talking about. While US practice does distinguish neutral and hot leads (for 110 volt outlets, that is), equipment doesn't depend on that much anymore. The only place where it is still used is in power switches, which are often single pole and in the hot wire only. In ancient days, designers would connect the supposed neutral to the chassis, but that's no longer done and probably no longer legal. Instead, the grounding wire (green wire) connects to the chassis. That's true worldwide. (In Norway, that would be the connector at the outer edge of the outlet.) All this works fine in correctly wired installations, where no current flows on the ground wire, and all ground wires are connected to the same common ground as required in the USA by the regulations. If the installer gets this wrong, you may end up with a voltage difference between the ground pins on different outlets. That's both illegal and dangerous -- it can fry equipment, but more importantly it can fry people. The same can occur in any other country -- all it requires is that the electrician screws up sufficiently badly. paul

Paul Koning wrote: Grounds are usually local to the building they are in. Grounds are often done using a metal post hammered into the ground (hence the name). The ground is not supplied by the power company. Two buildings can each be properly grounded, but their grounds can be at different voltage levels. Sometimes one building may have two different grounds. It gets fun where the outlets in one wall are grounded to one side of the building, and another wall is grounded to the other side. It's a good way to fry equipment. If you have a cable between the two buildings, you can get a voltage difference between the two grounds. Especially if you have a lightning strike in the area. And yes, I've seen damage caused by such differing potentials, as well as measured the difference between two buildings. I measured 70+ volts in one instance, without any lightning in the area (after one melted terminal). We went with some short-haul modems in that instance (opto-isolated, differential pairs).

Paul Koning wrote: Usually not too hard to get fixed, if you talk to the right person. Sometimes it's hard to find the right person. (ie. phb: "sounds expensive, we'll just work around it") However, you first have to discover the problem. It's rare enough that it's not something you usually check for, unless it's causing problems. Until you can get it fixed, you still have to deal with the problem in the meantime, or suffer more damaged equipment. You unplug it, people will say "we need this" and plug it back in. (Hint: to keep something unplugged, turn it upside down and open a random panel or two, or remove the cover. Scares them off for a while. An oscilloscope nearby helps too, even if it's just plugged into it's own square wave generator.) Cables between two buildings are usually not going to get the building grounds joined. You then need to do something to work around it (opto-isolate, optical fiber, radio/microwave, etc) Yes, you can get an amazing amount of power out of it in some cases, but it's not reliable enough to expect to power anything with it. You usually need an electrical storm nearby to cause a gradient.

I keep hearing about this TU58 simulator, but haven't been able to find it. Can someone point me to it? Preferably one that compiles and runs under Linux. Thanks, Rick Bensene

In a lot of cases -- certainly on DEC machines -- signal ground is connected to frame ground internally. If a frame ground interconenction passes enough current to melt the wire, then you've got significant voltage difference between the 2 machines. That is a major safety problem if the 2 machines are close enough that you can touch them. For %deity's sake sort out the wiring before somebody gets killed. If the 2 machines are far apart -- like in separate buildings, then you should really be using some kind of isolated connection (either transformer isolated or opto isolated) to avoid such problems. -tony