Ethernet breaker

Yes, that is why I thought it might be better to short the line.

A double pole relay with the armature of the two poles shorted, and then the two wires of the appropriate pair to the fixed contacts. The fixed contacts are much shorter in most relays, and so should have less effect on the signal. Still, it might be that 100baseTX is as far as I might want to go with it. The cable can be arranged so that the twisted pair comes right to the contacts without a long stub, other than the contacts themselves.

At least you don't need a very expensive coaxial relay!

-- glen

Reply to
glen herrmannsfeldt
Loading thread data ...

Thanks all for the input. For those concerned with my electrical skills, I assure you I have almost none! I'm a software/systems eng. at a fairly large company with access to top notch EE and ME folks. However, like many these days our management is all about the buy and very down on the build solutions. However, if we do go the build direction you've given me some great bits of data to start the ball rolling.

So, I did end up finding something fairly close to our needs. The big question will be pricing, but it looks to be a cool device for our type of application.

formatting link
a discussion of this whole problem I was asked about embedded, data inspection type security in the equipment and this Siemens device claims to address some of that. However, most importantly to me is that it has 2 x 10/100B-TX ports and a standard industrial type package.

But if that doesn't work out you've got me started with some design points to pass along for an in-house switch. Thanks again!

-Rich

Reply to
Rich Noonan

Yes, it's safety critical and when the equipment is installed in the US I'm sure it's covered by OSHA. But it's covered by much more stringent industry regs which is what I'm helping respond to.

As an example, OSHA defines high-voltage as >50V. When fully operational, areas of our equipment have >3KV running through. It can arc almost 3 inches and you can be injured just pointing in the wrong direction. OSHA regs, if any relevant ones exist, would not adequately cover this.

So, that's why when a tech has defeated interlocks on the safety doors to troubleshoot a problem in a high-voltage area it's critical that they be solely in control of the machine. With that much danger at play you might think they would just go through the trouble of unplugging the ethernet as needed, but this can be challenging both physically and procedurally. A blend of automation, direct human interface, and careful procedure is the standard solution.

-Rich

Reply to
Rich Noonan

No ready made part recommendation. Basicly you could build what you need by combining an Ethernet switch and an relay connected to PC serial/parallel port (in case you have PC nearby that runs the application).

Just connect the switch between the power source and the switch. The computer control power to switch on/off. If your switch uses low voltage power supply (like those small "wall wart" powered switches), then just adding a relay to one of the power wires is preaty easy and safe. If you know how to safely build mains powered things (and you have necessary qualifications to do that) then circuit for controlling mains power to device is another option.

You can find some information on interfaceing relay to serial port at

formatting link

Another interresting RS-232 port hack: Marko Mäkelä's electronics projects: RS-232 Solid State Relay

formatting link

Parallel port interfacing made easy

formatting link

There are also commercial products and electronics kits for PC interfacing.

And there are companies that make remotely manageable power switchers that can take controllign signal from almost everywhere (serial port, Ethernet, telephone line, cellular phone SMS text message etc..)

besides using switch and turn it's power on/off you can use the relay directly to cut the Ethernet signals.. A relay that cuts the connection on all the four signal carrying wires (applied to 10Base-T and 100Base-TX) shoudl do the job. Keep in mind in the circuit design not to cause too much impedance mismatches and crosstalk for the signals. Many normal signal relay contacts should carry Ethernet signals acceptably. Other way to cut the connection woudl be to short the signal wire pair (one relay contact on TX pair and other on RX pair). Short circuit should stop the Ethernet signals from getting to other end, so should look same to cards that line is cut.

Those are just few of my ideas for this.

tomi031cv

Reply to
Tomi Holger Engdahl

I'll take mil-spec electronics any day over a bunch of Radio Shack relays and wallwart powersupplies and the odd cables and crappy connecters and ducttape used to hold it together.

I bet if you *do* find a "breaker" it's not going to be a anything like a mechanical relay. It will be electronic and have no moving parts.

I asked before, You aren't doing anything safety-critical or covered by OSHA, are you?

Reply to
Al Dykes

No ready made part recommendation. Basicly you could build what you need by combining an Ethernet switch and an relay connected to PC serial/parallel port (in case you have PC nearby that runs the application).

Just connect the switch between the power source and the switch. The computer control power to switch on/off. If your switch uses low voltage power supply (like those small "wall wart" powered switches), then just adding a relay to one of the power wires is preaty easy and safe. If you know how to safely build mains powered things (and you have necessary qualifications to do that) then circuit for controlling mains power to device is another option.

You can find some information on interfaceing relay to serial port at

formatting link

Another interresting RS-232 port hack: Marko Mäkelä's electronics projects: RS-232 Solid State Relay

formatting link

Parallel port interfacing made easy

formatting link

There are also commercial products and electronics kits for PC interfacing.

And there are companies that make remotely manageable power switchers that can take controllign signal from almost everywhere (serial port, Ethernet, telephone line, cellular phone SMS text message etc..)

besides using switch and turn it's power on/off you can use the relay directly to cut the Ethernet signals.. A relay that cuts the connection on all the four signal carrying wires (applied to 10Base-T and 100Base-TX) shoudl do the job. Keep in mind in the circuit design not to cause too much impedance mismatches and crosstalk for the signals. Many normal signal relay contacts should carry Ethernet signals acceptably. Other way to cut the connection would be to short the signal wire pair (one relay contact on TX pair and other on RX pair). Short circuit should stop the Ethernet signals from getting to other end, so should look same to cards that line is cut.

Reply to
Tomi Holger Engdahl

If you want to make a windows computer turn off it's *own* ethernet interface on and off it's easy with the "netsh" command which can be scripted. No hardware required.

Do a netsh /? for help..

For a *inx box do a similar thing with the ifconfig command.

Reply to
Al Dykes

Given the situation you described in another note, particularly the high voltage and safety issues, you'd best convince your boss to hire a qualified electrical engineer. This is not a job for someone with no electrical skills.

Reply to
James Knott

Right. I'm often amazed how much of the former is tolerated in a multi-million dollar machine while absolutely zero of the later can be accepted.

You speak our language. Seeking a physical LOTO similiar to a padlocked box.

No question, this would be the best approach and we're looking at how it could be accomplished, but unfortunately physical constraints are numerous. Let's just say the ethernet switch and PCs are in an unfortunate location relative to the rest of the equipment. This is why the design proposal (and this is very preliminary stuff at the moment) gaining traction is a multi-position switch in a lock box, the switch controling "breakers" on the ethernet lines.

Yes, that's correct. My ill use of language b/c I don't deal directly with that stuff. I find the tool owner and have him check state before I touch it!

-Rich

Reply to
Rich Noonan

The one I remember was 10kV/cm or so, but that would be for DC. RF can go much farther than that. In any case, yes, be careful.

(snip)

How about a mechanical interlock that won't open the box until the cable is removed? Remember in the days of TV tubes that were user replaceable, the power cord went through the back in such a way that it was disconnected when you removed the back. Not that you couldn't cheat, but they could have done better if it was required.

-- glen

Reply to
glen herrmannsfeldt

Ah, much better. When looking at critical systems, it's important to specify the preferred failure modes. Which is worse: something not working when it's supposed to? or something working when it's not?

Most of the posters have assumed you want a reliable network and of course that doesn't include trash like wall-warts. But I get the impression network reliability _isn't_ critical (for once), but the disconnect _is_!

This is a LOTO (lock-out, tag-out) situation. A strapped down (security Torx-head) wall-wart on a lockable breaker is OK. But for more reliability (uptime) and less bypassibility, I'd simply put jack and patchcord inside a lockable box.

When the tech needs isolation, he follows the written procedure including a signoff step to open box, pull RJ45, leave air gap, and lock box with LOTO hasp.

You will need to make sure the switch (or any live network port) and the machines ethernet are not accessible. Otherwise, the yahoos will run a cable when the tech is on break.

This doesn't sound good. Defeat nothing. Disarm with signoff. Sometimes double.

-- Robert

Reply to
Robert Redelmeier

Okay. Safety control of an industrial machine for servicing. Proper procedure is for a lockout switch to be used. Turn your ethernet topology around. Have a small three port ethernet switch that connects to the machine, the priority control PC, and the other ethernet switch. The other ethernet switch connects to all the other PCs. Depower the second ethernet switch with an approved lockout switch and you are good to go.

--Dale

Reply to
Dale Farmer

Are you worried about the conduction of HV or the tcp bits? If the former you could use fibre optic ethernet to eliminate any DC conductivity to the equipment.

Reply to
Al Dykes

Well, I don't think any passive switch exists that is Cat5 rated. You might be able to find an ethernet A-B box, but it will be light-duty rated. You could try putting the Rx pair through a DPST explosion proof box. and the Tx pair through another. This will probably work so long as the runs are well under 100m. It sure won't meet the untwist requirements, but the explosion-proof aluminum housing should cut stray fields.

-- Robert

Reply to
Robert Redelmeier

There are a couple of things that were designed for exactly this purpose. i.e. disconnect a network conenction when required (usually when a device was not working).

  1. This one, fortunately (or fortunately - I don't know), is for Fiber Optic networking.

It is called an Optical Bypass Relay. (Optical Bypass Switch too apparently)

The intention was to cut off the two fibers that connect to a port and re-connect the external pair to maintain a FDDI ring. The FO spec for GE and 100M Ethernet is I think the same as for FDDI. Maybe you can still get them.

Google for [optical bypass relay] There you go:-

formatting link
"Failsafe return to bypass mode with loss of power"

You could put more than one in series if one was not reliable enough. (in a disconnecting sense).

The consequence of looping the ethernet would have to be considered (spanning tree protocol, split IP networks, others?), however this could be avoided with two OBRs back to back.

Maybe you don't want to change to FO networking and is does sound as if using convertors would be unaceptable.

I would guess that these devices were made to be VERY reliable.

Oh, maybe they are used today by Telcos for SONET.

  1. Using relays (mechanical and solid state) to switch networking signals was standard practise with token ring hubs. T/R used signalling that was very similar to ethernet and Cat5 cables.

A T/R hub is formally called a MAU.

Multi-station Access Unit.

I am sure that you could probably find someone to dig out their old T/R drawings and design something for you.

You could hear the relays clicking as stations came on to and left the ring.

Maybe you could stick to lower data rates if 100M was problematic.

Reply to
anybody43

Found the mechanical switchbox:

formatting link
Down to $12.60 from $33.10 last time I looked at them.

You're welcome. 8*)

Reply to
William P. N. Smith

It says "CAT5" but I'm sceptical. It may work at 100Mb with short cables but that's way different from CAT5 compliance.

I'm sure it works fine at 10Mb.

Reply to
Al Dykes

OK, I'd imagine the OP would be able to slap a Cat5 certification tool on it, but what's so hard about Cat5 in a giant box like that one? An {four,eight}-gang rotary switch and a bit of care with your twists and it should be just fine, IMHO...

Reply to
William P. N. Smith

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.