I'd like to build a device using a 2.4GHz radio with a directional antenna to connect to a reasonably distant AP, then the 5GHz radio to provide local hotspot for devices.
Assuming I can use the radios like that...
Am I correct thinking I'd be best with a device that has a single SMA mounted external antenna for 2.4GHz, rather than an array of them and / or mixed band ones used for beam steering, etc?
And if so is there an easy way to identify them from the antenna specs like 1x1 (no idea what that means ) on devices? Or something like the radio chips used?
Likely best is to buy 2.4ghz ap for long range link and other for LAN.
1x1 is not important. Go for AX for double speed You can attach antenna to 1st antenna port (type is not important, try to get around without converters) and config wifi capabilities to match. You need lightning protector etc, this may accumulate.
Other option is to buy complete point to point pair of directional access point with builtin antenna. Check likes of tplink, mikrotik etc.
x86_64 with Alfa Networks AWUS036ACHM USB-based WiFi adapter. Or, really, anything else that can provide sufficient power to this adapter over USB.
P.S. There is no need to limit yourself to a single antenna; cross-polarized antennas are a thing, and they double the possible bitrate, but you need Alfa Networks AWUS036ACM (which has less range than AWUS036ACHM but still much better than commonly found elsewhere) to use them.
After trouble getting the original router to recognise the USB device I came up with the idea proposed here of using a suitable router without USB . However, I switched to a different router and now have an aged 802.11b/g single antenna USB adapter giving a mid-range link.
The consensus seems to be that I can use a multi-antenna device, so I may experiment with putting a homemade bi-quad, etc in place of one on something.
As you said, relatively expensive compared with making something. I already have the one I made previously somewhere, so will try it out before deciding if the idea warrants further spending.
Main difference I see is that item has two N-connectors. Does it replace two antenna and hence keep the MIMO type stuff working? With one bi-quad and one standard antenna I'd expect the diversity algorithms to be somewhere between passively confused and struggling to understand - with possible performance risk. Which gets back to using a single antenna device, or more generally replacing all antenna not a subset.
This item has two antennas inside, one with vertical polarization and one with horizontal. So yes, it does replace two antennas that point in the same direction, and it keeps a good separation between the two signals, which is exactly what is required for MIMO. And it has "MIMO" in its name