Hi, I’d like to connect two RAK2287 concentrators to a Raspberry Pi 3+ CM.
I’m wondering if it is better to connect both concentrators to a single SPI bus on the RPI or whether I should put them on separate SPI busses since the RPI has multiple SPI busses available?
Has anyone else got multiple concentrators running on a RPI?
cheers
Really it’s better to use two pi’s.
If you really, really want to go forward with this, using distinct busses might make more sense than using distinct chip selects on the same bus. But I’d suggest you prototype it with jumper wires and breakouts before committing to a design.
What benefit do you hope to achieve with dual concentrators anyway?
Thanks for your reply.
Can you elaborate on why you think two pi’s would be better? Our current gateways are based on a Pi Zero, so I’d thought that a Pi 3B+ would have had more than enough CPU to support two concentrators.
My concern was that two concentrators might saturate the SPI bus itself, I would have thought putting them on separate busses would address that fully.
We have a network of around 250 gateways covering both urban and rural areas. We’ve had some issues with coverage not being as good as expected. More gateways helps, but negotiating to install them can by time consuming and expensive. As such I’m looking into how having two concentrators on a gateway site can help. I see a number of possible use cases for dual concentrators:
Some of our gateway sites in the country are right on the side of hills, so rather than using an omni directional antenna I’d like to try using sector antennas. A single sector antenna might not have a wide enough beam, so I envisaged having two sector antenna connected to a single gateway, or a sector antenna and a yagi.
At some rural sites we using 10db omni directional antennas to maximise range, but the narrow beam angle combined with the height of the tower means we get poor coverage close to the gateway, so I’d considered fitting both a 3db and 10db antenna to the gateway.
In urban areas where we are struggling to get reliable readings from LoRa units that are installed underground (water meters) I was interested in trying a sector antenna with vertical and horizontal polarisation to see if the horizontal polarisation helped with receiving multipath signals.
Also when trying to provide coverage to large apartment buildings a 3dp whip on the roof usually covers the top floors, but sector antenna pointing down would be better for the lower floors.
And finally, if the coverage issues prove to be interference related, then a 2nd concentrator would allow us to use additional receive frequencies.
Looking forward to your reply.
In the grand scheme of things, there are minimal cost savings for a £30 Pi and potentially hours of testing and having a single point of failure by trying to get one Pi to run two concentrators.
It’s probably not going to actually help.
Can you elaborate on why you think two pi’s would be better?
Mostly simplicity and the ability to use stock software and setups unmodified; also the likelihood that you’re going to find the dual concentrator gateway accomplishes nothing, so it will be easier to separate two already distinct gateways to two locations.
My concern was that two concentrators might saturate the SPI bus itself, I would have thought putting them on separate busses would address that fully.
I think that’s unlikely to be an issue, but I don’t disagree with using two busses if you have them in the sense of both hardware and understanding the needed software configuration support.
My main argument against the whole idea is complexity and lack of a likely benefit. At least, I’d try your ideas with two single-concentrator gateways before I worried about creating a dual-concentrator one.
a 2nd concentrator would allow us to use additional receive frequencies.
That’s complicated in that you’d either need the dual concentator everywhere or figure out how to command some but not all nodes to use the additional set of channels.