Crazy Download Speed Variances

[Curmudgeon10]

Correct. No AFCIs on the circuits involved with PLA "A" or the PLA I am testing at "C." I don't know about the type of breaker through which power arrives at "B," but that's the best performing AP by far so the point is moot.

I think the Ethernet cable test you suggest could tell me the electrical wiring in my old house is not particularly attractive to PLAs in some spots. This is already what I suspect rather than believing there is anything wrong with the PLAs I am using. Unfortunately, as I've mentioned, there's little I can do about the locations, geometry, or house electrical wiring. It is what it is.

When I think internet traffic is back to "normal," I'm going to do some more speed testing at the modem and router to insure the ISP is delivering the bandwidth I am paying for. As far as improving the performance of my PC, it does not appear that it will be achieved through a PLA/Direct Ethernet connection or MERLIN Firmware.

I discovered another Securifi ALMOND in my stash yesterday, so I may try another AP in a bedroom just down the hall from my office. Perhaps that bedroom will have the uncluttered circuitry that makes the operation at B the best in the house.

 

[Curmudgeon10]

Yes, I've verified that there are no AFCI circuit breakers involved in this problem.

My comment about not knowing what to do with the information has to do with my inability to change anything but the firmware involved in the network. Can't move anything, can't run new wires, etc.

The current plan is to acquire an Ethernet USB adapter and take my Chromebook around to a number of outlets in the house and see if I can develop a "map" of which ones are best for PLAs, which ones are worse. (even if not knowing precisely why, but suspecting it is just poor wiring practices behind the walls).

 

[Curmudgeon10]

I'm back. I acquired a USB Ethernet Adapter and took my wife's Chromebook around the house testing various electrical outlets for download speed using either a "direct" wireless connection to the primary ASUS Router, a wireless connection to a signal produced by a Securifi Almond configured as an AP and driven via Ethernet froma PLA, or connecting the Chromebook directly to a PLA. The included table lists the results:

My goal was to see if there were some electrical outlets that were better than others in delivering download speeds. The results were ..... blah. I can draw no conclusions other than ........ in my particular house, PLAs are probably not worth the money. I couldn't find any that provided an advantage over simply connecting to my router.

In fact, as part of the test, I sat next to the PLA/Almond at C where before I installed the Merlin firmware and put everything on Channel 11, I had measured 75 MBps. Now it was 43. On a whim, I decided to connect to the Router's 5 GHz network, although the signal strength was showing "Medium" or intermittently "Low" on the Chromebook. Download speed was 150 MBps. I went to the PC -- again connecting it directly to the 5GHHz network, and measured 125MBps.

I had always confined the use of the 5GHz network to devices in the same room as the router, noting all the limitations on line of sight, and I guess (incorrectly?) believing there was a direct correlation between signal strength and download performance. I always connect to the strongest signal, and outside of the room the router is in, that has always been the 2.4GHz signal. My Securif Almonds, used as APs, only broadcast on 2.4 GHz.

I still don't understand the dramatic differences between the two networks:

At the PC, connected wirelessly to the 2.4 GHz network with Strong signal: 38 MBps
At the PC, connected wirelessly to the 5 GHz network with Medium to Weak Signal: 125 MBps.

If I could offer any advice, it would be if you are contemplating the use of PLAs, run some tests as I did before your free return period expires.

 

[L&LD]

Seems like you can get rid of some PLA's and have a more responsive network at the same time.

The correlation you saw between signal strength and throughput I've seen many times and that is why I nominally discount any reports of a new/different router having a weaker signal. Wi-Fi isn't about signal strength (although it is important, of course), it is about the throughput. And stronger (steady, reliable) throughput gives a better online experience, always. (And to be 100% clear; as opposed to the fastest possible throughput, but one that is sketchy, inconsistent, and variable).

With the results of your spreadsheet you've attached, I would disable any PLA's you don't need (at least their Wi-Fi, which is mudding up your wireless environment for little to no reason).

You may want to test what Control Channels work best now. Since you know that connecting directly to the router via 5G is your best bet.

Control Channel Set up https://www.snbforums.com/threads/a...details-in-the-description.55582/#post-472051

Control Channel Setup 2 https://www.snbforums.com/threads/very-slow-on-2-4ghz.26942/#post-205456


Above and more can be found in the link below.

L&LD | SmallNetBuilder Forums

 

[Curmudgeon10]

Yes, I'm going to do some control channel experimentation, but first I've decided to upgrade the network to Wifi 6 Mesh using ASUS equipment. My existing ASUS RT 68 will replace the AP at "C" in the diagram. Except for that position --- which supplies WiFi signals to a separate building about 50 yards away --- it appears from testing I should forget 2.4GHz and just go with 5.

Just to clarify one statement you made: "..the correlation I saw between signal strength and throughput..." Actually, what I observed was an "inverse correlation." I recognize this doesn't mean there is a universal principle: weaker signals mean better throughput. But what it does mean in my application is that even though a signal might be weaker, as long as it remains reliable, you actually need to test its throughput without making assumptions that "weaker = less."