For gigabit ethernet you are probably S.O.L. As 10BASE-T and 100BASE-TX only use two pairs in the cable you only need two DPST reed relays - the other pairs can be left open circuit. Add a grounded foil shield over the reed relay and a ground plane under it and it should be reasonably trouble free for 100BASE-TX Ethernet, as long as the total cable length is small compared to the specified 100m max cable length. If a true physical disconnect is required, a miniature DIL DPST reed relay per pair would minimise the impedance discontinuity. Simple, usually scriptable via SNMP, etc.
Personally, I'd agree with others, I'd rather just use the management in an ethernet switch or similar to do this, or turn power off to a switch, etc. I haven't used these so I can't discuss how well they work, or what is involved in using them. If you want to do this in a nicer way, there are also several vendors which have signal switches. Actually now I think about this I might have only switched the 4 which are used in the 10/100 ethernet standard, but the premise is the same. Like I said, it works well enough, but again we're only taking a few feet of ethernet cable here, so the signal has a lot of room for the problems introduced by running these through relays. I wrote a chunk of software for the arduino that switches the relays. In my case, it switches between the test equipment and an ethernet port. For each of the 8 wires in the CAT5 cable, each gets switched by a single relay. What I have done is take a USB attached arduino and hook it up to one of these: They also make an 8 relay version, which might be better suited for your needs (I needed a few other things switched, which is why I used the 16 instead of an. With that in mind, it works well enough for me to do 100Mb/s traffic tests through the device. So this method can add severe flakiness to your ethernet connection. In particular ethernet can be more fussy and also less fussy than you'd expect. This SHOULDN'T BE TRUSTED to always work. BEFORE I DO, SOME DISCLAIMERS: This does NOT preserve the balanced signal on the ethernet correctly.
What I am about to describe is how I do this.
So, I switch the ethernet to the rest of the test gear. However, before I hook the unit under test to ethernet, I want to do some electrical testing to make sure all the connections are electrically ok. Part of this is automated test is to test connectivity to an ethernet port. Some background: So, I have a chunk of automated test gear.
0 kommentar(er)
