150 Mbps connection drops to 72 Mbps

Sounds to me like your monitoring the reported speeds by the adapter, and not the actual speeds as *measured*, say by NetCPS ( http://www.netchain.com/netcps/ ) or Iperf ( https://publishing.ucf.edu/sites/itr/cst/Pages/IPerf.as... ). Those adapter readings are less a measure of speed, and more a measure of signal strength/quality, with only ESTIMATES regarding performance. Therefore, they can be wildly inaccurate! The ONLY thing that matters is measured performance. Until you do that, the reports coming from the adapter are pretty much meaningless.

I know it's a measurement of signal strength rather than speed, but there's still a correlation between strength and speed. I get noticeably slower speeds when it drops down and occasionally it will plummet to 5 Mbps or even 1 Mbps, which would kick me from games like Battlefield 3.

I can't quite figure out how to use those programs you linked, but I get these results from Speedtest.net using a local server.

At ~108-150 Mbps:


At ~54-72 Mbps:



I'm not saying that 21.40 Mbps is terrible, but I'm just trying to optimize and get what I'm paying for.


I guess the real issue here is the times when it drops to literally 1 Mbps and it feels like I'm running on dial-up. It tends to fix itself after a while, but is there anything I can do to try to prevent it from happening?


Funnily enough, a few minutes after posting this my network dropped to 5.5 Mbps:

The reason I prefer tools like NetCPS and Iperf is that when testing *local* connectivity problems (in this case, wireless), it's best to eliminate as many variables as possible. The Internet has an inherently high degree of latency, and so tends to distort results. It gets hard to determine if the differences from run to run are due to local issues, or just variances in your Internet connection.

To use either is incredibly simple. You just run the executable on the server side with the -s option, then run the same executable from the client side and specify the IP address of that server. And in this case, because it's wireless and wireless is half-duplex (i.e., you can't transmit and receive at the same time), it's best to have the server side a WIRED device.

Anyway, if you believe these results w/ Speedtest.net are at least consistent wrt the reported changes in wireless strength/speed by the adapter, it may not matter. But I would like to know if you get more consistent results over a WIRED connection. If so, at least that would suggest the problem really is local, and wireless specifically.

Like all wireless, it’s subject to numerous environmental factors, including interference from other wireless routers, other 2.4GHz devices (microwave ovens, cordless phones, even wireless keyboard/mice). The 2.4GHz is not reserved for routers, it’s open and available to 1000’s of other devices. How and where your desktop is located/situated can lead to variations too. All this makes it really difficult to diagnose wireless problems.

Have you at least tried using tools like inSSIDer or perhaps NetStumbler to monitor the 2.4GHz freq? That’s the first place I’d start, see what’s happening around you.

A wired connection will yield a good 100-115 Mbps on Speedtest.net. The wireless connection is certainly not efficient, I'm aware of that.

The modem and router are a floor below me maybe about 30 feet away. I almost never experienced this issue before when it was just the router broadcasting alone, but even when both the router and modem were broadcasting it worked fine for a few days before this started happening.

Let me just be clear that, like I mentioned at the end of my previous post, my primary concern is now the tendency for the connection to drop to 1 Mbps.

I am using Meraki's WiFi Stumbler to ensure I'm operating on channel 11. What's peculiar is that sometimes it will decide to mosey on over to channel 1 for no apparent reason and I have to manually hit apply again to send it back.

The only devices nearby are my phone and another computer in the next room, both of which are connected to the router rather than the modem.

One thing I've been observing is, when the network is under load, the speed/strength will drop accordingly. Sometimes it returns to the original value (72 Mbps usually), but most of the time it will get stuck at some lower value like 39 Mbps or 26 Mbps. I just tested this by running a Speedtest.net test while it was at 72 Mbps. After the test, it stayed at 26 Mbps. I ran a second test and it then stayed at 6.5 Mbps. A third test brought it down to 1 Mbps, it won't go any lower than that and it definitely isn't going back up.

Could there be some glitch that's causing this cascading degradation of the connection? Why wouldn't the signal strength go back up?

While it doesn't happen often, these wireless products do fail from time to time, and deteriorate over time (heat is the enemy of wireless). If you're seeing the wireless signal "wander" from time to time, that's not normal (at least I've never seen that happen before). It could be an indication of a hardware problem, or maybe a poorly tuned wireless device (adapter/router). Perhaps this "wandering" explains the loss of signal from time to time.

Any chance you can at least try another wireless adapter, even if you have to borrow one from a friend?

I’d also like to know what happens if you disable one of the wireless radios. I wonder if there’s something “funky” going on between them that could explain it. Or perhaps try changing channels (use 1, 6, or 11 only, the rest overlap).

I don't really have the time right now to fully explore all those options, but I did do something that seems to have improved my situation, if only temporarily. Even though it's completely contradictory to wireless networking optimization, I set both the modem and router to channel 11. Since then I haven't hit 1 Mbps. I got down to 5.5 Mbps, but when I ran a test like I did before, it shot right back up to 52 Mbps. I even managed to play a whole match of Battlefield without any issues.

Right now it's fluctuating between 39 Mbps and 52 Mbps, but I'm getting great results, basically the same as the first test I linked for 108-150 Mbps.

It's not a perfect nor pretty solution, but it's working, so I'm not going to try to break it.