What's new

Wireless is finicky, head desk version

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

azazel1024

Very Senior Member
I am continuing to do lots and lots of wireless testing, mostly for fun, partly to try to stamp some wifi driver bugs I have on my tablet as well as optimize my setup the best I can in my house.

Last couple of days I tracked down a "wireless bug" related to inconsistent performance. It didn't seem to have a reason as to why sometimes I'd test and I'd get 48-62MB/sec (depending on exact positioning) on my laptop for a 867Mbps 11ac transfer and then other times I'd be lucky to see 42MB/sec and maybe more like 36-38MB/sec depending on where I was sitting with my laptop in my basement. Exact same location could easily have a 10-20MB/sec variation (or 20%+).

Then last night I realized why. It wasn't run to run variation, it was test session to test session variation. The window in my basement (hopper style, so it swings in from the top when opened) was open during the tests where performance was poor and it is located only 3ft from the router (about 16 inches when opened). Close the window sitting on my couch, 50MB/sec. Open the window, 41MB/sec. Rinse and repeat.

10 minutes of playing with router location on the shelf it sits on (only viable location) and moving the router 8 inches further from the window, instead of being 6 inches from the end of the shelf) and the open vs. closed difference is 2MB/sec. 52MB/sec with the window closed (baseline performance also went up a tiny bit with the new location, at least from sitting on the couch, 20ft line of sight from the router), 48MB/sec with it open.

Much better. My tablet sees no difference now (about 9.4MB/sec average 5GHz 40MHz transfers, instead of bouncing all over from 7.5-9MB/sec with it open).

What a weird, weird thing.

To add, I also got a chance to really play with channels. Looks like the router is using the old radiation standards. Using the 147+ channels the signal strength jumps 8dBm compared to 48 and below. Line of sight performance jumps about 10% when using the higher channels. Interestingly, in a couple of quick tests with obstructions in the way (such as one floor up and a few feet over, as well as one other basement location that is 30ft away and 1 wall in the way), signal strength is only ~2dBm higher than the lower channels and performance is within 2-3% (only 1-2MB/sec faster with the higher channels, instead of ~5MB/sec faster).

Now I just have to combine that channel performance information with another test location or two (such as a far location, as all have been close/medium distance) and then try out my N600 AP with the lower 5GHz channels (currently set on the higher UNI-III band channels) to see if there is a performance impact there. Its already slower than my AC1750 router on 5GHz (obviously), so I'd rather the extra performance when I need to be connected to the N600 router than I would the, generally, small performance benefit the AC1750 router seems to have with the higher channels.

Lesson here, come on manufacturers, use the higher power limits for UNI-I!!!

PS Of course other than the anecodotal signal strengths and performance figures, I don't have a proof-in-the-pudding measurement or answer from the manufacturer that the lower 5GHz channels are using reduce power output compared to the higher 5GHz band. It does seem like it though. It could maybe be that the higher band frequencies work a lot better for EBF, or reflections off the materials in my home or something similar and nothing to do with actual higher broadcast power with the higher part of the 5GHz band.
 
Oh and to add on, I also relocated my N600 AP (TP-Link WDR3600) after my earlier testing of a couple of weeks ago, as I found a longer network cable to run from the jack to it's new position (not obstructed by the 4ft thick fireplace for most of the house that matters). At least until I can install a jack below where it's, likely permenant, new home is (as the current jack is 6ft away and around the corner).

Signal strength is so much higher around the 1st floor of the house, especially with the 7dBi antennas that I decided to bump the 2.4GHz broadcast power to min from max on the WDR3600. I tried medium and my phone seemed a little faster and more willing to roam, but it was still taking awhile (1-2 minutes) when I walked in to my bedroom on the far side of the 1st floor. Laptop and tablet would roam over to the router below them pretty readily, though slightly slower (10-12s instead of 4-5s). InSSIDer on my tablet was saying 5GHz was unseeable before and 2.4GHz was -78dBm before. Now it says 5GHz is -86dBm (still no bueno, but I am fine with that, I don't need or want stuff connecting to the AP on the side of the house really, I want it connecting to the router) and 2.4GHz is -61dBm...a little stronger than I want (because it is making things reluctant to roam).

Bumping it to medium broadcast power didn't make the RSSI twitch in InSSIDer, it still claimed the same, but stuff seemed to roam a little faster (tablet and laptop were deffinitely under 10s, but not by much, phone and iPad 2 were taking 30-40s, instead of over a minute). Bumping 2.4GHz broadcast power to low dropped RSSI from -61dBm to -63/64dBm. Phone and laptop are back to roaming in around 5 seconds and phone and iPad 2 are roaming in around 15s, which isn't far off from how they were behaving before.

I haven't had a chance to run performance testing, other than a VERY quick spot check, but turning down broadcast power on 2.4GHz doesn't seem to have made any significant changes to performance anywhere where I care about it.

I need to do a little more thorough testing to see though, as the 7dBi antennas showed a fair amount of gain over the 5dBi antennas in performance at that location, but that was at full broadcast power (which 5GHz is left at and also the band I care a little more about). The two quick tests seemed to be within the margin of error (5%) and also still faster than the 5dBi anntena results for 2.4GHz.

PS this is all by way of saying I am crazy and probably don't spend my limited free time as a father of 3 well.
 
If the window is Low E it can have quite an effect on RF.

It is and after I realized that the window being open was causing the issues, that was the assumption I jumpted to (darn you sputtered aluminum!!! Though I thank you in the winter)
 
Fun read...

Reminded me of an old friend that converted a cold-war era Bomb Shelter into a home office... at least he didn't have to worry about AC :cool:
 

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top