Sticks,
I know only a little about this, but hopefully it can help.
The lower in frequency you want to measure to, the longer the impedance tube must be. The impedance tube at IBM's acoustics lab I use is 14 feet long (I think) and it can get down to 70 Hz accurately. The construction itself needs to be massive and rigid, though that's probably not as difficult as coming up with the hardware and software to create the signals and analyze the results. I have no idea how that part is done, but Dr. Matt Nobile, head guy at IBM, says it's fairly straightforward. Yeah, sure, for him maybe.
I'll be quite interested to see if Fran or anyone else can explain more.
--Ethan