Just ran across this. Slightly cheaper than SX1308. Anyone have a comparative analysis? They don’t even seem to be on the Semtech site.
Semtech SX1302 LoRa® Gateway Baseband Transceivers reduce current consumption and simplify the thermal design of gateways while decreasing BOM costs. The SX1302 is a new generation of baseband LoRa chip for gateways that is capable of handling a higher amount of traffic than preceding devices.
The high-speed baseband digital engines are clocked from a single 32MHz clock source. The chip embeds the capability to support SF5 and SF6. Eight modems have been added to specifically handle the high-speed SF5 and SF6 packets, while the eight others are currently taking care of the SF7-SF12 traffic.
The SX1302 LoRa Gateway Baseband Transceivers provide an architecture that reduces power consumption significantly, allowing easy embedding in highly-integrated environments where power dissipation might be a challenge. The SX1302 Transceivers are also serialized in production, with a globally unique 64-bit number. The SX1302 has an operating temperature range of -40°C to +85°C.
Semtech SX1250 Multi-band Sub-GHz RF Front End is designed to work with the SX1302 baseband engine, to design a high-performing LoRa® or LoRaWAN gateway. The SX1250 Front End covers any frequency band below 1GHz, making the device ideal for license-free band support.
The SX1250 supports up to +22dBm output power on-chip and can be leveraged to reduce the bill of materials cost when designing a gateway. The need for an external high-end and power-consuming Low Noise amplifier is eliminated due to the extremely low noise figure of the system.
I would expect most users are downstream of the point where this is available on a price-competitive concentrator module card (if not a finished gateway); not that there aren’t things I’d like to include if doing my own board design at that level, but efficiency of effort points to designing the hard part once. And preferably by someone who has already had the experience of doing a board with the '1301.
(FWIW: what I’d like to see is a transmit pulse signal as typically used to drive an LED broken out on the docking connector as when trying to tighten node receive window timing it is hugely useful to trigger a scope on a GPIO of the node and put another channel on the gateway transmit and see how the pulses line up. And to share signals that could be used for solid T/R switching when either combining multiple concentrators to make 16 channel gateways, or when adding an additional node-class radio to be able to send overlapped but non-conflicting downlinks, granted that’s more useful in a custom network than in TTN)
Interesting, the SF5 and SF6 combination was part of the SX1262 silicon. And with that silicon SF6 is not compatible with the SF6 that SX1276/8 devices can use.
I see the SX1302 and this (indoor) ref design as an enabler for high density node deployments indoors with short range hence the benefit of SF5/6/7 shorter on air time - good for battery life/low maintenance as well as reducing collisions between devices with similar GW rcvd pwr levels. Guess we need L-A & TTN to allow quick adoption of these other SF’s for community to benefit…risk is 8 channel 6SF nodes & GW’s become as unfashionable (and frankly undesireable) c/w 8/16 channel/demod 8SF GW’s (note 8 of the 16 demods only allow for up to SF10 I guess for die size/power limits) as old single channel GW’s & nodes from a couple of years back! (IMHO Single channel GW’s are the spawn of the devil and should not be called GW’s! nor allowed on the network given problems they cause for users )
and everyone buys new gateways (and nodes) ? … this will take a while imho
No. Because it’s backwards compatible at least between SF7-10 (and 11 and 12 where legal).
If the node can do SF5 or SF6 and its within range of a gateway that can and ADR’s up to that, great! If not, or if connectivity via only that gateway is flaky, it will only ramp to SF7 (or the wideband or FSK channel) where it can use other gateways.
Supporting SF5 and SF6 in network software is fairly straightforward, for the most part you just give the node back what it used, via the same gateway that heard it. And the bits for ADR to command datarates would have to include a suitable definition.
If the handling of those only gets sufficiently specified in some future version of the LoRaWAN spec beyond what TTN is using, it could still be possible to “backport” a custom variation of the old version, and nodes or gateways that didn’t try to use them would never notice.
However, it’s probably not worthwhile until this supposedly cheaper hardware ends up cheaply deployable, and not just as reference designs costing twice what an SX1301 concentrator card does.
Edit: it does seem like a gateway would have to include metatdata to tell the server that it has an SX1302 so that ADR to those datarates could be considered. Otherwise an SX1301/1308 should be assumed.
Hi. I am using the reference design customizing to my application with a Raspberry Pi, and I ask to the deserve to use the SX1261. I didn’t find in the SX1302 HAL repository the library for SX1261. Even in the reference design the external of the SX1261 are labeled as Non Connected (NC).
(IMHO Single channel GW’s are the spawn of the devil and should not be called GW’s! nor allowed on the network given problems they cause for users 
)