The problem isn't a mater of openness, it is Could new transmitters be built? Yes, but it would be at a price no one is willing to spend. I imagine that it would be feasible to obtain the proper documentation, but there is the matter of building the hardware. More significantly, I suspect that duplicating DSN is a rather monumental task.
The transmitters and receivers that were designed and built in the '70s required lots of discrete components. The up-converters, amplifiers and matching networks will still need to be discrete components, but I'm estimating that the other 80% of the radio equipment can be implemented as a software-defined-radio (SDR - see http://spectrum.ieee.org/geek-life/hands-on/a-40-softwaredef... for an example).
I'll be reaching out to a friend and former colleague who specializes in this type of hardware/software.
I have no expectation that we can replicate even a tiny portion of the DSN, but I'm talking with people that already have big dishes and tracking capabilities. Remember that we don't have to talk to this spacecraft continuously. I think even downloading the last of its collected data would be a win but positioning it to do more science, then collecting data from it again in several months/years would be really cool.
"The Very Large Array, one of the world's premier astronomical radio observatories, consists of 27 radio antennas in a Y-shaped configuration on the Plains of San Agustin fifty miles west of Socorro, New Mexico. Each antenna is 25 meters (82 feet) in diameter. The data from the antennas is combined electronically to give the resolution of an antenna 36km (22 miles) across, with the sensitivity of a dish 130 meters (422 feet) in diameter. For more information, see our overview of the VLA, and the configuration schedule."
So, yes, with multiple IP-connected software-defined radios, you could provide the required signal strength without a Deep Space Network antenna.
At best, your signal will improve linearly with the number of antennas (N), while the noise will be completely incoherent and scale like sqrt(N). Thus your signal-to-noise ratio will improve like N/sqrt(N)=sqrt(N). You're going to need an awful lot of radios...
