IC selection input for a 1000BASE-T disconnect circuit

[barfdogg]

I am looking at creating a stand-alone device that will sit between a switch and a few PCs. My goal is to handle multiple 1000BASE-T port disconnects on a timer. Instead of unplugging a cable from the patch panel or doing a port shutdown via a switch command like a sane person, I want the ports to be disconnected by this device.

 I have found single cable A/B switch/cutoffs that are physical, but nothing solid state. I was browsing sparkfun recently looking at Arduino projects to try out and saw some relay/timer guides and thought of this idea. I know you are scratching your head because this is redundant but I want to give it a shot as I will be going into a few areas that I have little/no skills currently.

While my electronics knowledge is beyond the average dummy, I am 100% a beginner in this space. My experience normally includes retro/arcade repair type things, (burning EEPROMS, replacing caps, fixing traces, etc.) I want to try making something from scratch and try PCB design, etc. I built a CNC and figured I’d try routing a PCB design out if suitable for this project.

Here is where I am at:

The actual timer/trigger:

This will be the easy part and is more in line with the sort of stuff I normally play with. I have an Arduino Nano and am currently working on the code for a simple countdown timer and LCD output. At zero, the Arduino will either stop triggering a pin to cut power to the ICs handling the Ethernet via a relay, or to trigger the SEL pin if I go with an Ethernet switching IC. I think no matter what the solution is configuration will be similar.

Actually handling the network traffic without destroying the signal:

I am aware that there are complexities handling 1000BASE-T on a PCB, and see there are many guidelines to follow if you don’t want to cause signal degradation, cross talk, etc. I have found half a dozen manuals/guides on the subject and am currently reading through best practices for PCB design regarding Ethernet. Im sure Ethernet on PCB is common enough of a task that I can gather enough data to make good choices. (I am certainly open to input though)

That said, the part where I am struggling the most is on deciding what components can handle these signals without degradation AND be able to stop them on demand.

After a few hours browsing through several TI and MAXIM products, I have a few ideas, but they don’t feel right.

The first component I considered basing the design on is a MAXIM MAX4890E 1000 Base-T LAN Switch. I started reading up and my thought was to route the Ethernet in and then only connect one of the outputs. I would then use the SEL pin to switch to the connected output pins when I want the port to be up. I am now reading into it further and if I am understanding correctly I would need to use Ethernet Transformers on both sides of each chip so it could handle the signals. In trying to keep cost semi- reasonable (I will probably have 8 ports or so), and coupled with wanting the traces as short as possible for signal quality, this solution seems flawed.

I later started looking into signal conditioners and redrivers. I found some TI chips that seemed to be promising, however just slapping 8 of these in again seems less than ideal. I have looked at several different ICs like CB3Q3245 and SN74CB3T3245 as  I just need to cut off the connection. I see skew/bandwidth limitations on most devices that I would think may impede the signal. I feel like I am either over complicating this idea or reading too deep into some way to do is in a more simple manor.

Someone please tell me how off course I am! Thanks!

[radiolistener]

you can connect cable through switch and power off switch with relay for example.

Attempt to build your own circuit will cost you much more and needs a lot of knowledge and experience in RF circuit development. 1000BASE-T works with 125 MHz  signals, so this is not easy to deal with it.

[barfdogg]

you can connect cable through switch and power off switch with relay for example.

Attempt to build your own circuit will cost you much more and needs a lot of knowledge and experience in RF circuit development. 1000BASE-T works with 125 MHz  signals, so this is not easy to deal with it.

Thanks for your response. I know it's a fool's errand, but figured it could be an interesting challenge. Your suggestion would work, but having an unmanaged switch exposed would not be a preferable solution. I don't need this, but do like making stupid goals to reach.

I'm hoping to figure out an ideal direction to go and then continue to read up on the subject, which i'm sure will lead me to a dead end when the project becomes too far out of reach. Either way I would enjoy learning more about the subject, or which direction someone who knows what they are doing would go. I am currently working on getting my CCDP for work and the subject in general has been consuming my life lately.

Thanks again!