@David-TP Thanks for your reply.

Yes, I've observed this on both my work and personal laptops (both MacBook Pros), and also on various streaming devices (eg. Apple TV).

I tried changing the DNS servers as you suggested; this didn't seem to make any difference.

I also tried swapping the main and satellite units.  I left them in the same physical locations, which necessitates using wireless backhaul (due to the available cabling).

So my "A" unit, which was the main, is now the satellite, connected by wireless backhaul to the "B" unit, which was the satellite but is now the main.  I've left it this way for a couple days now, just to see.

Interestingly, I still seem to get a far better connection to A than to B.

• I tested by downloading a 2G test file using `curl` on my MacBook.
  • While doing this, I was sitting ~10' from B with direct line of sight, while A is about two rooms away.
  • When connected to A, I'm getting a normal-ish download speed.
  • When connected to B I'm only getting 10-20% of that.
  • Anecdotally, I was also not able to download the latest MacOS update while connected to B, despite several attempts.  Once I switched to A, it downloaded and installed fine.
• Observed similar behaviour while streaming video on the Apple TV.
  • This is <2' from B but also two rooms away from A.
  • Initially it connected to B, due to proximity.  Video would routinely freeze as it tried to buffer more.
  • I switched it to prefer connecting to A.  Since then video has been streaming flawlessly.

Both of these were using 5GHz.  I tried switching my MacBook to prefer 2.4 but weirdly then it was no longer able to connect to the network.  Not sure why, haven't investigated this specifically yet.

I'm going to try swapping their physical locations, but keeping them configured the same (main vs. satellite), just to eliminate environmental factors.

Update

OK, this morning I've now swapped their physical locations while keeping their configuration the same.  So A is still the satellite, connected to B (main) via wireless backhaul, except that A is now located where B was, and vice versa.  I still observe the same thing, ie. way better connection to A than to B.  You can see this below.

(FYI I'm testing using a 2G test zip file from a website, but had to scrub the URL for it to accept the post.)

andrew@Andrews-MacBook-Pro Downloads % curl -o test.zip <test_file>
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
 24 2047M   24  504M    0     0  7306k      0  0:04:47  0:01:10  0:03:37 8217k^C
andrew@Andrews-MacBook-Pro Downloads % curl -o test.zip <test_file>
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
  1 2047M    1 33.5M    0     0   521k      0  1:07:05  0:01:06  1:05:59  297k^C
andrew@Andrews-MacBook-Pro Downloads % curl -o test.zip <test_file>
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
 18 2047M   18  379M    0     0  5462k      0  0:06:23  0:01:11  0:05:12 9051k^C

These are similar numbers to what I was seeing last night.

• First download was connected to A on 5GHz.
• Second download was connected to B on 5GHz.
• Third download was connected to A on 2.4GHz.
• I then tried to reconnect to B on 2.4GHz...but strangely whenever I try this it prompts me for my network password again, and then fails to connect.
  • I think this might be what I observed once or twice last night but didn't go down that trail at the time.

So now I'm kinda wondering if the simplest explanation here is just that my "B" unit is a dud...?