What better time than Christmas to finally get around to fixing all your parents’ technology problems, am I right? Speaking of which: https://decentsecurity.com/holiday-tasks/.
This year’s key task was fixing a problem that has vexed my parents for years: mediocre WiFi performance. WiFi can be a real pit of despair for laypeople because RF technology in general is just so opaque: in most cases, you’ve got some generic box with only two cables plugged into it and no physical indication of what it does or how it works. Everyone knows WiFi goes through walls, but very few people know that what the walls are made of and how they’re oriented makes a great deal of difference. Mesh WiFi kits have gotten extremely popular in recent years, addressing the problem using a shotgun approach with greater chance of success than band-aids like generic range extenders. Even still the nuances are sometimes too tricky for people without a good functional sense of how RF works.
Making matters even worse, cause and effect of performance can be obscured by confounding variables. For instance – if performance is really bad in one room, you’re likely to notice. But if performance is subtly worse in the living room, where you’re more likely to hang out on a rainy day, it’s easy to lock on to rain as the cause. Especially if you (rightly) hate your cable company and want to find fault with their network.
My parents have been having trouble with their network, consisting of a single home router; actually a Netgear cable modem/WiFi router combo, the C6300, which by the numbers should have been a very capable device with the recent 802.11AC standard and multiple antennas on both 2.4 and 5GHz bands. Let’s look at their first floor layout.
The WiFi router is located at the blue “desk” on the map. This is one spot a coax cable comes up from the basement, and it’s also where a desktop is located, so it’s the obvious choice of locale for networking hardware. Immediately next to that location and along basically the entire wall is a bunch of brick work for a central chimney running vertically through the center of the house, a wood stove immediately next to the desk, and a fireplace on the other side.
WiFi performance was modest most locations in the house except the family room containing the desk. Interestingly, indicated signal strength was no worse than decent everywhere, but in some locations, latency, latency stability, and throughput dropped off badly. One definite failing point was that 2.4GHz and 5GHz were separate SSIDs, which didn’t allow the smarter devices to de-congest the 2.4GHz band when 5GHz performance supported that. 2.4GHz congestion was therefore a problem across the board, even close to the router.
The biggest performance issue was in the bedroom corner with their TV. They’re all-in on streaming, so poor WiFi performance on the TV they use most often was extremely frustrating. My dad tried to remedy this with a 2.4GHz range extender positioned in the outlet directly below the TV, with predictably lackluster results. However, he thought this SHOULD work because the TV showed full bars on this range extender. Yet another foible of wireless.
In this case, I think the root cause of the performance issue was the brickwork at an extremely oblique angle close to the router, effectively shadowing that corner of the house, along with two roughly perpendicular walls. Subjecting the range extender to that same poor transmission path doomed it to failure, even though the TV reflected high performance along the very short path between the repeater and TV.
- Likely 2.4GHz interference and crowding
- Separated 2.4GHz and 5GHz SSIDs
- Range extenders deployed using yet additional SSIDs
- Range extenders deployed in same RF shadow as the devices they’re expected to serve.
Possible No-Equipment Mitigations
- Put 2.4 and 5GHz on same SSID, letting devices intelligently switch. Devices very strongly prefer to remain on their current SSID even if multiple credentialed networks are available, but will much more quickly switch between frequencies or access points of the same network. This will also un-crowd the 2.4GHz band as more capable devices switch to 5GHz.
- Range extenders broadcasting their own SSID are subject to the same sort of issue – if you move away from the range extender to a different corner of the house, you might remain connected on a suboptimal path when an AP with a much better link exists.
- Range extenders/repeaters should be thought of like a bucket brigade rather than a lamp. Give or take for device factors, a range extender will have the same link performance back to the AP as any other device put in the same physical spot. You should put a repeater in a place with good line-of-sight to both the router AND the area of poor coverage. In this case, that would mean moving the bedroom range extender from under the TV where AP coverage is bad to the other corner of the bedroom with mostly line-of-sight to the AP and also the TV.
Of course, I went straight for the throat and bought a mesh system, since this is Christmas and the name of the game is physical, tangible consumerism. That and range extenders are bad, but I thought a second AP would be necessary for the bedroom scenario. More on that later.
MeshForce M3 Kit
The first product I tried, mostly because I hadn’t before, was one of the new cheap mesh WiFi kits from MeshForce on Amazon. It’s a dual-band 5GHz/2.4GHz kit that comes with one base station AP and two repeaters, all of which get automatically configured for a single SSID with roaming. I set up this kit with the AP at the same desk as the original router, put the original router in bridge mode with WiFi off since it’s also the cable modem, and put the two dots in the bedroom by the chair with LoS to the AP, and on the other side of the house though that wasn’t super necessary. My thoughts are below.
Ease of setup: Excellent. It’s clear that MeshForce prioritizes this, as the whole setup experience was trivial. You do have to download a cell phone app, but I don’t recall how I figured everything out, other than I absolutely did not read any of the paperwork in the box. In other words, when you plug it in and connect to the new network, it will hold your hand through the entire (very easy) setup process. No need to overthink it.
Performance: Bad. Results were initially good and performance improved all over the house, and I thought the issue was taken care of. Unfortunately, this didn’t hold. After a few hours, performance was back to being about as mediocre as before. I did some digging, which I’ll get to below.
Configurability: Awful. Trying to debug performance issues was frustrated by the absolutely spartan options around device configuration or diagnostics. There’s a mobile app, and you must use it. There’s no web interface. The configuration settings are missing basic options like setting separate SSIDs for 2.4GHz and 5GHz (not that I think you should), and there are almost no diagnostic tools like wifi site survey.
MeshForce M3 Failure Description and Analysis
With only the base station plugged in and working, performance measured well using my laptop and iPhone, throughout much of the house. Areas I thought should be covered by the dots were relatively poor, as expected, but anywhere in the main living areas was great. I took note that my phone and laptop both connected on the 5GHz band. I also took note that performance by the bedroom TV was sill as poor as before, defaulting to 2.4GHz and poor throughput and latency.
I plugged in the dot in the bedroom to cover that TV blind spot, and was pleased that it seemed to work well. The TV connected to the dot vs the AP; the app indicates which devices are connected to which mesh points. I went about attaching other devices with the new credentials, like thermostats and other TVs.
A few hours later, performance was back to “not good” on the high-test devices like cell phones and computers. I looked at the app again and noticed that literally every device was reporting a 2.4GHz connection, and no matter how close I was to the router when restarting wifi on my laptop, this remained true. I searched the internet a little, and it turns out this behavior seems common for this system. I also contacted MeshForce support to see what they had to say, and the only thing they could offer was “restart the AP” (which, notably, didn’t change anything. I had to disconnect the dots, reverting to pre-MeshForce levels of performance anyway).
My theory, unconfirmed, is that the MeshForce network topology, being only dual-band, uses 5GHz for the backhaul between the dots and router. My guess is that, due to this resource contention, devices primarily connect to the AP with 2.4GHz, which is subject to the same interference and throughput limitations, maybe specific to this house, that the network experienced before making any changes. Unfortunately, in this day and age, 5GHz performance really is much much better than 2.4GHz if the RF line of sight and obstacle environment allows, so this reversion to 2.4GHz for the final hop really is noticeable. It could also be that backhaul performance was ALSO bad – I didn’t test too rigorously i.e. by plugging a hard wire into one of the dots. Long story short, the best my MeshForce setup ever performed was no better than the C6300 baseline.
Netgear Orbi RBK30
My second attempt was with the Orbi RBK30 kit, with a router and single plug-mount “satellite.” I have experience with Orbi, using a similar system in my own house. While I have my misgivings about and annoyances with Orbi, the raw performance is pretty good in my own house. As far as I’m aware, Netgear has phased out all the older dual-band Orbi products, leaving only tri-band systems. These have two separate 5GHz radios, one of which is dedicated to backhaul between satellites. Orbi also allows a daisy-chain topology, where the router is at one end of a chain of satellites (though this is sub-optimal).
Here are some bullet point comparisons with the MeshForce system:
- Setup: MeshForce was a bit easier and hand-holdier, but Orbi was hardly difficult. A seriously non-technical layperson would find MeshForce easier, but would probably succeed just fine with either. There were one or two hiccups where I took knowledgable-user sorts of shortcuts with the Orbi setup flow, but I think it’s a safe bet a letter-of-the-prompt user would have succeeded too.
- Configuration options: Orbi blows MeshForce out of the water here, with configuration options MeshForce could only dream of. For the most part, this shouldn’t be too relevant: if the defaults result in an optimal experience, as they should, the options maybe SHOULD be hidden from the user. But in the scenario that default performance is bad, options become very useful. On the one hand, Orbi has configuration settings that confuse even me (like, “Enable Implicit Beamforming” and “Enable MU-MIMO,” both of which confusingly default to “off”). But then, default performance is excellent, and much better than the M3 system. And Orbi also offers features like a relatively easy-to-configure OpenVPN host for logging in to your home network from afar, which M3 does not.
- Product Value: This is a little tricky. I ended up buying this Orbi kit for about the same price as the MeshForce, clearly suggesting the Orbi has much better value (since it works well). On the other hand, MOST Orbi products are in a higher price tier. It’s possibly that I just got lucky.
- Product Line Clarity: There’s something to be said for the fact that MeshForce only has a couple products: There’s M1, which is a cheaper product with no gigabit ethernet ports. There’s M3, where only the base station has gigabit ethernet (and only one port). Then there’s M3s, where all mesh points also have gigabit. My single biggest annoyance with Orbi is that there are a ZILLION products in the lineup, some that look identical with huge performance/feature differences, some that look totally different with extremely subtle performance/feature differences. Two otherwise-identical one-satellite kits might only differ in the number of ethernet ports on the satellite, but thanks to the complexities of the marketplace, might cost $100 different on Amazon. A cheaper kit with the extra ports might land right in the middle price-wise, but have 2x faster WiFi hardware. And then the entire mess is repeated with the WiFi 6 product range (at least they’ve all got shiny metallic bits for visual differentiation). Seriously Netgear (and nVidia), cool it with the ridiculously overzealous product segmentation.
I set up the Orbi kit about the same as the M3 kit, with the AP positioned at the desk next to the cable modem, and the one satellite in the master bedroom by the chair where it has approximate line of sight to both TV and AP.
- Performance is excellent everywhere in the house
- Every device with 5GHz capability is almost always connected with 5GHz, as determined by occasionally looking at the “connected devices” page to see.
- Oddly, VERY rarely are devices connected to the bedroom satellite, even in scenarios where you’d expect them to be. The Roku stick that started this whole charade stays connected to the central Orbi router via 5GHz every time I’ve checked, which is especially surprising since its performance is now totally acceptable.
- I notice one behavior that’s especially interesting in tandem with the sudden full-home coverage from the AP alone: signal strength shows excellent or at least good anywhere I go with my laptop, but if I run a speed test, it chugs slowly for the first couple seconds then quickly speeds up to full-tilt. This leads me to theorize that the beam forming performance of the Orbi base station is MUCH better than either the Meshforce (if it has such?) or the C6300 (it does). The couple second lag before high throughput is suggestive of the Orbi deciding to reprioritize beam pattern to cover the suddenly-high-throughput device.
If you’re buying a new WiFi system right now, I’d steer clear of the MeshForce kit, and I’d probably double-down and steer clear of the Amazon (eero) and Google dual-band kits, and go straight for the tri-band. Unfortunately, tri-band kits from Eero/Asus/Linksys/etc are available, but quite a bit more expensive: Eero Pro seems to start at $250 for a AP+Repeater kit, and is more like $500 or $600 for a 3-base-station kit. I’m looking at a Linksys Velop 3-node kit for $370. Asus runs $600. Fortunately, Netgear’s been making tri-band kits for longer than the others, and some of their older hardware is quite affordable, like the $160 kit I scored (refurbished). If you want to spend $600 on Orbi, you’ll be getting VERY good state-of-the-art WiFi 6 gear. If you spend more like $200, you’ll get a very functional 802.11ac kit that probably performs better than the flashy gear from Amazon or Google.
Note: I don’t have experience with this, but looking at some 802.11ax documentation and product specs, it’s POSSIBLE that, due to extensions to MU-MIMO in WiFi6, “dual-band” mesh products will perform as well as tri-band 802.11ac products. That said, all the “expensive” Orbi WiFi 6 products are tri-band.