I had different prices quoted to both you and @jmarcelino, and was told the adapter board wasn't for sale outside of the IoT discovery pack!
What about the new sx1308? was, wasn't the press-release, etc, suggesting that we'd be having sub US$100 concentrator boards?
You can already have sub $100 concentrator boards, see the pricing of RisingHF . But for a gateway you also need a controller board (e.g. RasPi), some wiring, a power supply, an antenna, an enclosure. These quickly add up to another $100. And then we're not even talking industrial grade material.
The SX1308 allows some cost shavings (around €25 actually), but it's rated for indoor use only.
A nontransparent/opaque pricing policy enables distribution channels to charge (distributor-specific) hi profit margins. This is new technology and booming business..
At minimum suggested retail prices (SRP) should be known. If lacking then that gives room for thoughts.
How about some $75 gateway that can be bought via AliExpress?
(industrial grade for this price range not required)
There will probably exist (much) lower cost alternatives than a €40 Raspberry Pi for basing a gateway on.
Not with the current TTN packet forwarder. The TTN gateway will use a PIC32 afaik, allowing to shave some €15 of the BOM. The microcontroller is cheaper, but your (custom) board adds some cost again, especially since you are only making a few thousands at most. The RaspPi on the other hand is manufactured at very high quantities (millions) and close to component cost (it's a non-profit), adding only the margins of RS or Farnell (typically 20%).
I think a price level of €200 for a consumer grade gateway seems natural. It's not that far off from a decent wifi router that is produced in quantities a thousand times higher and already has a well established market.
I think what @bluejedi was reffering to are the cheaper Pi boards like the NanoPi or Orange Pi (can be had for as little as $7.99)
I can confirm these work very well for gateways - maybe even better than the Raspberry Pi has they have native real Ethernet and don't use an internal USB-Ethernet adapter
I think for an average decent consumer WiFi router €100 to €120 should suffice. In my opinion a consumer LoRa gateway should be available for similar price range.
I was in fact targeting everything that is much cheaper than a Raspberry Pi 3.
Could even be a Raspberry Pi Zero, but that is single core/single threaded. I'm not sure if that will be sufficient for a gateway. But could also be PIC or ARM based, maybe even an ESP32 when sufficient.
Good point about native Ethernet instead of Ethernet over USB.
I never looked in-depth at requirements for a LoRa gateway. Not sure what the max concurrent network traffic is and what requirements are needed regarding processor speed, multi-threading, memory (incl. cache) etc.
I have little experience with the Allwinner SoC's you mentioned but I have seen many articles that their support is quite poor and their community (compared to Raspberry Pi and Arduino) is quite small.
and the (pcb) design, prototyping , writing and testing the firmware ect is for free ? interesting
I haven't seen a (working) 'DIY hobby' gateway with semtech chips , maybe that's because, until a couple of months ago you couldn't buy them in small quantities.
It starts with all the semtech tech info is under NDA
Where did I mention "for free"?
you only seem to add up the price of the basic components, that's ok ... for a hobbyist
Where did you see me adding up the price of the basic components?
The Raspberry Pi has been mentioned several times in this thread. It's price is around €40,- which is a rather substantial contribution to a hobbyist/consumer DIY gateway's price.
For a more dedicated gateway application there will very probably be (much) cheaper alternatives around than a €40 Raspberry Pi.
No adding up price of basic components here. Only subtracting savings for one of the components.
The AllWinner H3 and specifically boards like the NanoPi NEO are already supported in the mainline Linux kernel, by the community at http://linux-sunxi.org and the top team at Armbian.com
I actually don't recommend the official Raspbian distribution for gateways for stability reasons and find ResinOS or Ubuntu on my Raspberry Pi gateways far more stable so, for my purposes, the support of the Foundation doesn't really matter much.
In this case having native Ethernet isn't about performance, it's just that on the Raspberry Pi any USB glitch will also kill networking.
The pricing of a final product is seldom based on the cost-plus approach (BOM + R&D + marketing etc), there are many more parameters in play such as market size and market competition. That's why I don't see the pricing of consumer grade Lorawan gateways ever coming down to price points of a wifi router. Compared to the wifi market, Lorawan will always be a very small niche market with relatively few competitors and manufacturers that will demand higher margins because of relatively bigger R&D costs and risks.
You can get lower if you do like Honewell and others: sell gateways below cost but capitalise on the services you provide through them.
I can't see why not, technology wise LoRaWAN gateways are a piece of cake compared to modern WiFi routers that do MIMO, beamforming, dual bands and far more complex modulations at far higher data rates.
In fact I suspect someone will eventually do multichannel LoRa gateway functionality using SDR and publish that so anyone with a $10 RTL-SDR and Pi board will be able to join in the fun.
Of course I expect Semtech will deploy their crack team of lawyers but as we've seen in many attempts of thwarting technology via legal means good luck with that.
Just my opinion of course but to me the best - if not the only - option for Semtech is to allow gateways at a price point where doing this just isn't worth anyone's time.
ok, maybe I should have said individually priced sub US$100 concentrator boards!
The economies of scale will hopefully kick in with commercial networks deploying gateways.
Has anyone announced a sx1308 product?
I fail to understand why people are being so entitled about having dirt-cheap gateways? And, please, let's not trivialize the job of people designing and manufacturing gateways.
- Gateways are worthless on their own. Even being extremely conservative, and assuming a lot of redundant coverage, a gateway will serve >1000 devices, each costing a bare minimum of 10 to 30€. So, with a 200€ gateway, you serve 10 to 50k€ worth of sensors, but somehow, the gateway is overpriced? Isn't that missing the forest for the trees?
- Let's compare with Wifi. Sure, access points are cheap (and not getting cheaper), but how many of them do you need to cover a single building in the city? 10? 30? So, in a typical urban building, everyone has their own router (or several). For the same building, a single LoRaWAN gateway provide sufficient service. Divide the price by more than a dozen of neighbors each giving 10-15€, and one end up with a different kind of network but for way less money per user.
- When was it decided that a technology enabling the deployment of large scale, power efficient, sensor networks must be available for an arbitrary low price? Have anyone checked the price of a NB-IoT or Sigfox gateways recently? Get ready for a heart attack
If gateway cost is the primary driver for an application, then maybe the alternatives should be considered. Maybe build a dirt-cheap Wireless M-bus gateway with a RTL-SDR stick. Maybe build a gateway around a Zigbee USB sticks, they're less than 50€.
Very few of the gateways here will ever serve 1000 devices - especially with indoor units such as TTN's - so if you're only serving 100 or so devices (that's being very optimistic) gateway cost becomes a significant part of the total.
I can understand expensive gateways for commercial operators, very much like enterprise/carrier grade WiFi equipment costs big bucks, but does that represent the TTN community?