A quick test was 1.5km line of sight. RSSI was about -80dB with a total budget of about -140dB. Over the next weeks I'll do some real testing over longer ranges, it should have no issue in open line of sight conditions reaching 30-40km which is my goal for very low data rate telemetry.
Lora is an impressive technology and you can trade off packet transmit time for SNR/range. At the higher link budget levels, overall data rate gets down to just 10's or 100's of bits per second. At those rates you need to start looking at tighter tolerance crystals and then to TCXO modules. These cheap modules use regular crystals but you can easily remove (hot air) and replace with known quality crystals from a known distributor that are in the 10ppm range. Semtech (the chip maker) has a reasonable link budget calculator where you can see various tradeoffs (snr/rate/crystal tolerance).
Current consumption of the module depends on the power level you are transmitting at (you can adjust output in 1dB steps). Overall consumption depends on transmit and receive duty cycle. The module can also be flipped into sleep mode and then it's down in the uA range. At full output and during the transmit phase it will range up to 120mA (at 3.3V), that's a 20dB output with the power amp turned full on. The tests I were doing were at 13dB and around 30mA transmit current.
The biggest task was porting the radiohead code to a non-arduino and non c++ environment - but that was just a day or so of code slashing to get to the core routines that are actually necessary. A worthwhile process to fully understand the command/control code for the SX1276 (the chip that is used/relabeled) on the module. I've used some of the Digi 900HP style radios before, so already had a decent understanding of the broadcast/unicast/routing schemes that some radio protocols use.
cheers,
george.