Sometimes, when solving field trouble, a network technician needs a quick, expendable way for testing the life of Ethernet switch ports or patch cables, or identifying which wall socket leads to which port.
Using a laptop or other active device is possible but tends to be cumbersome and locating one which the user is willing to part with for a while is often more difficult than it should be.
A loopback device can be used, which mirrors the keepalive packets from the Tx to the Rx pair of the cable, making the port talk with itself; the port then reacts with lighting up the corresponding LNK indicator.
The device is as simple as a connector with two and two of the wires connected together. A field-improvised male one can be made from a cut-off RJ45 with a piece of cable attached.
It is however easier and more expedient to already have one. For testing of cables it is also beneficial to have a female one in addition to a male one. A keychain form factor was chosen for easy portability.
A standard RJ45
connector was chosen on the basis of its ubiquity.
A male RJ45 was crimped to orange and green wire pairs. (Two pairs only were chosen to keep the cable thickness down.) A transparent heat-shrink tube was placed over the wires. The cable length was trimmed to just allow the male to plug into the female.
A female RJ45 was soldered to the end of the cable. The orange wire was connected to green-white and the orange-white to green on the female's pins, forming a Rx-Tx loopback. The pins were covered with hot glue, an expedient insulation.
As the male's plexiglass tab is sensitive and annoyingly easy to break off, some sort of mechanical protection was desired. A piece of 0.6mm copper sheet was bent to a rectangular tube fitting just over the female. A shallow V-cut was made over the place where the male's tab will be. The tube was soldered along the seam. A loop of a thick copper wire was soldered to its side, for secure attachment to the keys without directly transferring mechanical load to more sensitive electronics parts. The shield was secured in place with suitable amount of guess what? Yes, hot glue.
The male and the cable were wrapped in sticky tape, the female end was stuffed with tissue paper. A spray-painted coat of matt black acrylic paint was applied over the shield and the outer side of the female.
The device was tested on a couple switches/hubs and a random embedded printserver and shown to be working. (Oddly, it does not seem to work on a Thinkpad R60 Ethernet port. More diagnostics has to be performed.)
 Assembly |  Assembly |  Assembly |  RJ45-M side |
 RJ45-F side |  Attached shield |  Attached shield |  Attached shield |
 Attached shield |  Attached shield |  Attached shield |  Before painting |
 Painted |  Finished device |  Finished device |  Finished device |
 Finished device |  Finished device, looped |  Finished device, looped |