Network/wifi experts: explain this mysterious discrepancy?

Factual Questions
1

I live in a three-level house, with a basement, first floor, and second floor, total area of about 3,000 sq. ft. (280 sq. m). We have Xfinity (Comcast) cable service with a nominal Internet speed of 150 Mbps. The house is wired with RG-6 cable to most rooms, and I use Actiontec ECB 2500 MOCA converters to get wired Ethernet connections to the PCs in my office and the Roku and Tivo in the home theater setup.

When we first moved in, the basement wasn’t finished, and I installed the cable modem (Netgear CM-500V) and Wifi router (TP-Link AC1750 v.2) in my home office on the second floor. The router has a very good range, and it covered the first and second floors nicely.

But after we had the basement finished last year, we moved the home theater setup, and now spend most evenings down there. The our seats were about as far in the house as you could get from the router, and the signal was not great.

At first I thought about getting a wifi extender, but I decided before spending any money to do the first thing everyone recommends when placing a router: put it as close to the center of the house as possible.

So last weekend I did that, moving the modem and router to the first floor family room. Before doing so, I used my Samsung Galaxy S5e tablet running Ookla’s speed test app to measure download speeds in the rooms where we use our devices. I ran at least five tests in each location, averaging the results, then repeated the process after the move. My hope was that download speeds would increase on average, even if they dropped slightly in one or two rooms that were now farther from the router.

I accomplished my primary goal, which was to increase signal strength in the basement and out on the deck outside the family room. I now have good coverage everywhere in the house. So I’m happy about that.

However, something strange happened. In the old set up, I could get download speeds of up to 237 Mbps over the 5 Ghz band when I was in the same room as the router, and between 100 and 165 Mbps in other rooms nearby. In the new setup, I can’t get more than 100 Mbps anywhere, even standing right next to the router! (The speeds over the 2.4 Ghz band are essentially identical to the old setup, maxing out around 40 Mbps.)

Now, I know that when using my tablet, cell phone, or laptops, there is almost no situation in which I could tell the difference between 100 and 237 Mbps. Streaming video from Netflix or using Zoom video, which are probably the highest bandwidth uses we’re likely to do, work fine at 25 Mbps, which is the lowest we get anywhere since moving the router. So it shouldn’t matter.

Frankly, I was quite surprised to see that I could get speeds between 150 and 237 over 5 Ghz under the old setup, and I’m puzzled and a little annoyed that I can’t get them now.

The only thing I could think of is that previously I was only using two taps off the incoming cable: one for the modem/router in the office, and one for the home theater setup. Now I’m taking three: router, home theater, and office. So I removed the line to the home theater and checked the speed. No change.

Any ideas?

2

Have you tested the speed without WiFi? (I.E. Wired directly into the router)

I assume the router in the center of the house is hard-wired to the home’s cable wiring.

3

My bet would be signal loss from the extra tap or from too long an ethernet cable if WiFi connectivity is rock solid. I’m an RF/Microwave engineer who designed the RF half of a number of wireless ethernet products and I did 1/4 of Sysco’s network admin course, and that’s what I would look at first.

Physical connections can be finicky at high frequencies and if all the network settings on all your gear are the same as before, there’s not much left.

4

Could it be your testing methodology? Instead of testing 5 times back-to-back in each room, what if you sampled the speeds throughout the day, for a week at least, and see how they perform?

Two reasons for this:

  1. Coronavirus may be affecting download speeds intermittently
  2. Xfinity used to burst downloads speeds up initially, for a few seconds, but then slows it back down for long sustained downloads. The rationale is arguably to let webpages load faster while still saving total bandwidth for large downloads, but this would also have the side effect of making speed tests seem faster than expected.

Overall, it might be better to separate out your tests from internet speed (router to internet) from your wifi speed/signal (using a wifi tester tool on your phone). The two are related, sure, but ultimately they are separate things with separate fixes.

If this remains an ongoing issue, you can pick up a mesh network router system, rather than a wifi extender.

5

I am no electronics engineer but did you have to cut, extend, or otherwise re-terminate any coax when you moved the modem. You may just have a weak crimp on a coax line and or some other in line connector that may be damaged or functioning poorly…

6

I had this happen once. Bad cable but not so bad as to be obvious. Just bad enough to make things crappy.

Testing speeds while plugged directly in to the router and not via WiFi will help narrow the potential issues.

7

The sudden appearance of a 100 Mbps cap suggests a cat-4 cable sneaked in there somewhere.

8

Ding, ding, ding! We have a winner!

I wasn’t fully aware of the differences between the various Cats, but as soon as I read this, it tickled something in the back of my mind. Sure enough, I had used a pretty blue cable between the modem and the router, and when I checked it just now, it is thinner than my other network cables, and has “Cat 4” printed on it. Into the trash with it!

Download speed via 5.0 Ghz is back to 237 Mbps! Hooray!!! Now all I have to do is re-do all the speed tests to get my real statistics. And scour the house for other Cat 4 cables.

Thanks, The Librarian! You are one smart monk–, I mean, ape! Oook!

Thanks to everyone else for your suggestions, too.

9

As usual, no good deed goes unpunished. Having accomplished my goal of improving wifi reception in most of the house, I’ve uncovered two new problems that I hope you folks can help me with.

The first is that my wired connectivity is nowhere near as fast as the wifi! At each of the three cable drops, my office, the family room, and the basement, the max download speed I see on any PC is about 60 Mbps. The computers are Dells (Precision and Vostro towers, and XPS-13 laptop), all running Win 10 with the latest updates and drivers.

Unfortunately, since my main goal in moving the router downstairs was to improve wifi performance, I didn’t think to measure download speeds on the PCs before the move. So it’s not 100% clear that they have gotten worse because of the move. But in any case, clearly I should be able to get speeds over Ethernet that approach or exceed the wifi speeds, right?

Obviously, the first suspects will again be cables. The one Cat 4 cable that caused the first problem seems to have been the only one of that type in the house, but all my other network cables are Cat 5 or 5e at best. I just got a set of Cat 6 cables, and they’re now installed in the office and family room. (I presume the ones in the basement are 5/5e, but haven’t checked.) I’ve ordered several more sets of Cat 6 in various lengths, which will arrive tomorrow, and once I have installed them I’ll banish all lesser cables from the house.

But since the best I’m getting in the office with Cat 6 wires is well below the nominal 150 Mbps that Xfinity says they’re delivering, clearly I need to look at the coax cables as well. Do you have any advice, beyond looking for obvious loose connectors (which I don’t expect to find) and using trial and error replacement, on how to judge whether a coax cable cable could be faulty or unsuited for purpose?

For instance, would an RG-59 cable slow the signal the way the Cat 4 did? Should I make sure everything is RG-6?

Also, could splitting the signal to three drops instead of two be causing a significant slowdown? Under the old setup, I was using the same 3-way splitter on the incoming line, but only two cables were connected to it. As I mentioned above, a quick test of that theory didn’t seem to affect wifi speeds, but could it affect the wired speeds?

The other problem is that, now that the master bedroom is the farthest room from the router, the Amazon Echo my wife uses in the master bathroom to listen to music while putting on her makeup can no longer get a reliable wifi signal. :smack: :smack: :smack: Our phones and tablets can handle the slightly weaker signal fine, but the damn Echo can’t. :mad: I’ve tried moving it to several different places in the bathroom, but no joy. And there’s no way to move the router significantly closer to the master bedroom, either.

I’m toying with getting a simple wifi extender just for the sake of the Echo. But rather than do that, I could just revert to the old setup and use an extender (or a mesh adapter) to boost coverage in the basement. But I was doing so well without spending any money! :frowning:

Suggestions?

(Unfortunately, because my wife is working from home and using wifi during the day, I’ll have to wait until Saturday to start doing most of this testing.)

10

How old are the PCs? Is it possible that your LAN cards are old enough that they have a max speed less than 200 MB?

11

The Vostro is 11 years old, the laptop is 8, and the other desktop (I was mistaken before, it’s an XPS 8900, not a Precision) is three years old. But the network devices in all of them are rated for gigabit speeds.

An update: I am now getting 120 Mbps on the laptop next to the router, but the desktops in the other two locations are still only getting 30-40 Mbps. The laptop is on the cable connected to the splitter’s -3.5 dB tap, the others are -7 dB. Is that relevant?

12

Good point. Cleaning and tightening the coax connections may help.

13

Most of my coax cables were factory made, but I installed a couple of F connectors myself; I think they’re all right, even though I had to compress them with a channellock instead of a proper compression tool. But I’ve ordered a compression tool, some connectors, and a coax cable tester that should arrive on Saturday. I plan to weed out any RG-59 cables I find, and I hope with all this to eliminate the wired speed problem.

Still pondering the best solution to the Echo situation.

14

Your MoCa adapters have 100Mbps ethernet ports. (Cat4/5 is good enough–Cat5e/6 is needed for 1Gbps)

60Mbps in real life sounds about right, no need to check you coax.

15

If you’re bored, look for InSSIDer - download it to your laptop and it will show what WiFi signal strength it sees.

just because your network is rated 1Gbps does not mean the entire computer can handle that, especially older systems. but it should be able to exceed 60Mbps. Always start by eliminating the extra bits and working your way out.

16

Dammit, you’re right. The specs mention 270 Mbps over the coax, but I didn’t notice the 100M Ethernet port. :smack: Now I’m tempted to buy a set of these updated MOCA adapters, even though my original motivation was to spend as little money as possible. :rolleyes:

But I can do better in the office over wifi. I’ve always assumed that wired connections would be faster than wifi.

Someone please persuade me that there’s no practical advantage to ordinary home and small business computing in having a 200 Mbps connection instead of a 60 Mbps one. I think I need to stop obsessing over making “improvements” that won’t make a real difference.

17

If you are not copying large files, you will not see a practical difference between 60 and 150Mbps. Others metrics (ping, DNS response time etc) have more impact on your browsing experience.

For example Netflix says it needs 5Mbps for HD.

Good, stable WiFi coverage and a stable low latency network have much more impact on your experience than raw bandwidth.

Make sure you can use all the bandwidth your ISP provides, otherwise get a cheaper plan.

18

I would start worrying about bandwidth when you get around to buying an 8K TV and a subscription to some streaming service that can deliver that kind of content.

It might be a while.

19

Note that there’s a company producing MOCA 2.5 adapters. I replaced my Actiontec MOCA 2.0 adapters (that I had to reboot every few days) with these and they are considerably faster and more reliable, and cheaper than the MOCA 2.0 adapters from Actiontec. Plus they have a better management interface, better encryption, etc.

I don’t think you need the additional bandwidth (which isn’t really accessible anyway since the 2.5 adapters have gig ports) but I’m impressed with how much more reliable my network is. I would suggest buying their splitters if you need them. I had some odd network issues initially until I started using their splitters, which are pretty reasonable at $8.50/ea.

I know I sound like a shill for GoCoax but I’m not. I’m just really impressed with how much better the new adapters are working in general.

20

Yes, I don’t know how many sites I’ve seen where the latest 1Gb workstations are connected to the central switch in series through a VoIP phone with only 100Mb ports.

Vonespy is right - what are you doing that NEEDS more than 60Mbps? And how often do you need internet that fast? Some web servers limit the output anyway.