@ Mr. John Smith - I think short as possible is recommended. I know the G120E uses the internal Ethernet so it is different than the ENC28. However in regards to the same signal wires this is the quote from the ENC28J60 that is used on other GHI boards:
"Since relatively high currents are necessary to operate
the twisted-pair interface, all wires should be kept as
short as possible. Reasonable wire widths should be
used on power wires to reduce resistive loss. If the
differential data lines cannot be kept short, they should
be routed in such a way as to have a 100ohm characteristic
impedance."
ENC28J60 data sheet, Page 7, Microchip Technologies, Inc.
(This was fresh on my mind since I just read that particular portion Saturday as I am working on a board design with the ENC28.)
@ skeller - 100Ohms so thatās max 100 meters then? Also, if necessary I should ramp up the voltage to 5V and use a regulator to bring it back down to 3.3 at the jack. A simple zener should do.
@ Mr. John Smith - I donāt think that will work, at least not reliably.
Here is the a document from TI on design layout for a PHY:
This, I think is the PHY component on the G120E:
My guess is that even the J11D, while it works, is not a manufacture (Texas Instruments) recommend design. It probably would not stand up to rigorous EMI/RFI testing.
(I do understand the J11D was probably not designed to pass any kind of EMI/RFI testing. It was designed for the quick prototype market. I donāt mean to take away anything from GHI or the design of the G120E/J11D. I use them with good success.)
@ skeller - It just has to be reliable enough. Assuming I used actual ethernet cable with shielding as opposed to ribbon cable. Then for the most part all Iād be doing is joining 2 Ethernet connections with a rj45 jack. Thatās how I see it.
Of course Iām going to have to do it to find out. Based on the spec they arenāt too picky about the connection between the Magnetics and the PHY.