I have 4 AP903s in a mesh, the root and 2 nodes being wired to the router and a wireless link between the root (in the kitchen) and the fourth node in the garage. The straight line distance between the root and the garage node is about 5 metres, with a brick wall and a wooden door in the path.

When I set these up a few weeks ago the signal strength between the root and the garage node was about 30%. I noticed over time that this appeared to reduce, such that it was around 20% a week ago and this morning it has dropped to 7%. (When I set them up I could get 60% if I put the root in the kitchen window but this was inconvenient domestically)

No APs have been moved and nothing extra has been introduced into the path. Is this some undocumented "power saving" feature of the AP903 or what else could be happening? As a radio amateur in my dim and distant youth (still have the licence though!) I have some familiarity with EM wave propagation but I find this puzzling. Is the 5GHz signal taking some sort of multipath route, such that moving small items around in the garage could affect the reflections off various objects?

Piste Basher wrote: Is the 5GHz signal taking some sort of multipath route, such that moving small items around in the garage could affect the reflections off various objects?

I have no expertise in the subject ... but, thinking about it, 5GHz must reflect off objects - given that they use the band for Radar. You've not converted your garage into an aircraft storage facility, during the lockdown, I take it

Have the AP903's aerials been moved? (I have no idea how they're supposed to be orientated :? )

Do you see the same relative changes in signal strength on the various Wifi 'phone apps? - and have any new Wifi networks appeared on the same channels?

Thanks for your interest - I should have made it clear that not moving the APs included not moving their aerials. I have often pondered orientation of the aerials as cross-polarisation can certainly reduce signal strength (assuming that the orientation of the Draytek aerials corresponds to the direction of polarisation, which may be a rash assumption), but I have ended up just leaving them all vertical.

I don't check for other signals regularly but looking just now I can see only one Sky box somewhere on the same channel as me, but that's down at -96dB (1%).

Interestingly, the signal now (1730) is at 20% again, from 7% this morning, and nothing has been moved inside the garage or kitchen, although it is somewhat warmer. Puzzling.

Just wondering if its possible our garage node has connected itself to one of the other nodes. I'm sure you'd have noticed the extra hop but it is one possibility.

In my work-in-progress setup, I've noticed that one node - an AP802 - needs very little excuse to jump on to and stick the other node (AP903) in preference to the 2862 root with its a stronger signal.

I'm using the dashboard on the root to see what's connected and the connection between the garage and root remains stable, apart from the strength variation.

You may be able to persuade the node to be more "sticky" by juggling the roaming parameters.

(The fact that we can even have 2.4GHz and 5GHz transceivers in a cheap plastic box is a modern miracle - in my day to use the 23cm band (1.2GHz) we had to buy highly specialised valves and construct elaborate metal cavities as the anode tuning circuits....)

Piste Basher wrote:
I'm using the dashboard on the root to see what's connected and the connection between the garage and root remains stable, apart from the strength variation.

You may be able to persuade the node to be more "sticky" by juggling the roaming parameters.

I didn't think it was likely...

Switching off roaming has one of the things that settled down my users' experience aka they've stopped shouting at me. AP's seemed to be continually playing 'pass-the-station' resulting in drop outs in Zoom & Teams calls and whatever game . If I get time, I may carefully try roaming again .

Piste Basher wrote:
(The fact that we can even have 2.4GHz and 5GHz transceivers in a cheap plastic box is a modern miracle - in my day to use the 23cm band (1.2GHz) we had to buy highly specialised valves and construct elaborate metal cavities as the anode tuning circuits....)

You're quite right! I remember my Grandpa's loft stacked with valves - he was a radio and TV engineer - all of which couldn't come close to anything like these consumables. Lovely things - I wish'd grabbed a few