Wireless testing: SMH edition

[azazel1024]

Man, every single time I think something is going to be easy, I am reminded how hard wireless is.

I've been playing with some 7dBi antennas on a router I have, replacing the 5dBi antennas. Testing is not fun. Granted, I think it has 100% to do with the location of the router (in AP mode) as it is in a "cube" shelf (only realistic place I can put it, which is better than behind the TV up against my 4ft thick masonry fireplace).

Moving the darned router just a couple of inches with EITHER antennas has dramatic shifts in 5GHz performance. With the 5dBi I can sometimes get 61MB/sec downstream performance with my laptop sitting on a desk 12ft away. Shift it a few inches lower (by removing some books sitting under it) and performance can drop to 54MB/sec. Move to the next room with my laptop on the kitchen table and moving the router a few inches further from the wall and performance can jump from 38MB/sec to 48MB/sec. A little higher or lower can also have large swings in performance. Upstream isn't nearly so effected, but that can still have a +/-10% swing in performance just by shifting the router a few inches.

With the 7dBi antennas, shifting the router around can also have pretty big swings in performance. However, same room it is almost always 5-10% SLOWER with the bigger antennas, but in the next room at the kitchen table it is almost always 10% faster. SMH

I think I am going to have to put in a lot more testing to see which set of antennas is better for average performance and also see if I can actually shift where the router sits (maybe on top of the bookcase/cube unit, but that may put it too high to perform well with either set of antennas as it places it about 6ft off the floor).

I also still need to test 2.4GHz performance.

I wish wireless were easier. It also doesn't help that I can get several tests in a row that have the same performance within 5%, and then my 4th test without changing ANYTHING performance will drop 25% easily for a test or two, and then jump back up. Means I have to keep doing 5-7 runs with each tweak/antenna swap/location, drop the one or two total outliers and average the rest. This router is proving to be a bit more of a pain for testing than my old WDR3600 was. I'll grant, it is faster than my WDR3600, but still a bit of a pain.

 

[azazel1024]

Well Friday I had to head home early because of a "we'll be there between noon and 4pm" deals.

So I had extra testing time. I setup a RAM disk this time to avoid any disk issues (though with an SSD and the speeds involved, there shouldn't be).

After about 2hrs of testing I found that it is EXTREMELY picky when in the enclosed cube shelf. Adjusting just an inch higher or lower or an inch or two forward or backward can easily result in a +/-25% speed difference in one location and 15% in one but minimal +/-5% in the 3rd location (no impact at the 4th location).

If I sit it on top of the entire thing, placement has minimal impact (other than "it is on top". Is it still on top?) when moving it around a bit (+/- several inches in any direction).

The interesting thing is, with either the 5dBi or 7dBi omnis, the performance was FASTER sitting in the enclosed shelf/cube than it was sitting on top. There is a vertical difference of about 18 inches, which may be what made the difference (IE too high). With the 5dBi, it is only marginally faster sitting in the enclosed shelf than sitting on top, but it was/is marginally faster (about 5% faster at 2 locations, no difference at the final location. Location within the shelf DID have pronounced differences in speed though. Only looking at the best performance for both router locations).

The 7dBi antennas were slightly faster than the 5dBi antennas, but only at 3 of the 4 locations. At the furthest location, the speed difference on 2.4GHz and 5dBi was more or less within the margin of error between the two, though if I sat the router on top of the whole shelf contraption, the 7dBi WERE slightly faster than the 5dBi (about 8% faster 2.4GHz and 5% faster 5GHz).

At the other locations, in the same room, 10ft away Line of Sight, in the next room about 10ft away obliquely through the wall on my kitchen table and in the next room sitting on the couch acutely through the wall 15ft away, the 7dBi were 5%, 10% and 5% faster respectively over the 5dBi antennas for downlink performance. Uplink the performance difference was 37%, 10% and 5% faster respectively.

Yes, that is correct, 37% faster same room uplink with the 7dBi over the 5dBi. The 5dBi best uplink performance was 35MB/sec averaged over 5 file transfers (2GB in size each), this was achieved in the same room. With the 7dBi it jumped to 48MB/sec!!! Again, this is an average of 5 transfers.

I did notice at all locations that the 7dBi produced slightly more consitent results.

For example, downlink same room with the 7dBi produced 56MB/sec +/-2MB/sec over 10 transfers. The 5dBi produced 54MB/sec +/-3MB/sec. With both antennas there were occasionaly times (about 1 in 6 transfers) that would be oddly slow, generally in downlink. So I could get around 56MB/sec downlink with the 7dBi with 5 transfers in a row and then the 6th would only hit 42MB/sec for the entire 2GB. Then the next one would be back up around 56MB/sec.

The 5dBi had the same behavior, though again, slightly less consistent even among the "regular speed".

That said I stuck the 7dBi on there in the end, both because of the huge uplink gains in the same room, but most locations either seemed to lose nothing, within the margin of error (I considered nothing less than a 5% difference as being meaningful) or just that transfers were slightly more consistent AND averaged a little faster in most locations.

The location of the router isn't actually the ideal. With it elevated about an inch higher the same room downlink goes from 56MB/sec average to 61MB/sec average and uplink goes from 48MB/sec average to 50MB/sec average. Its just that at all other locations this resulted in a loss of 15-30% performance in downlink and uplink. So I'd rather be running ~10% slower in the same room with much higher performance in the other locations.

I did spot check a few other locations that weren't the standardized 4 locations and performance seemed slightly better or the same with the 7dBi over the 5dBi in those as well, but it was only a few quick spot checks swapping between antennas, not the 5-10 downlink and then uplink file transfers at each location, with each antenna configuration and then testing on top of the shelving unit, versus enclosed in the shelving unit with each antenna configuration and then testing moving the router around at each location with each antenna configuration. I just picked the best location for the router and then tried each antenna configuration for a couple of uplink and downlink transfers. Everything seemed to run the same to about 10% faster with the 7dBi, except same room uplink which ran about 20-30% faster.

Interestingly if I held the laptop at router height at all locations, the general performance was about 10-15% faster with the 7dBi, but only about 5-10% faster with the 5dBi.

I cannot move the router lower than it is now, nor will I be holding the laptop 5ft off the ground in normal use.

I also played with my outdoor AP testing as I swapped my 841nd for a WDR3600. 5GHz performance drops off pretty fast with distance. I see 25MB/sec 5GHz 40MHz at 15ft, but at 30ft it drops to about 15MB/sec. About 10MB/sec at 60ft. This is with my laptop. 2.4GHz 20MHz performance is still about 13MB/sec at 30ft (same as at 15ft) and has dropped to 10MB/sec at 60ft. I didn't test further than that on my laptop. On my phone (iPhone 5) I get about 50Mbps down and 30Mbps up at 15ft. At 80ft I got 30Mbps down and 13Mbps up. On 2.4GHz (I only tested one spot) I got 30Mbps down and 16Mbps up on 2.4GHz, but that was at about 130-150ft.

That is with 7dBi omnis run through my garage wall with RG58 coax 1 meter extensions (I think it is RG58). Looking at charts, that should be 1/1.6dB of additional attenuation for 2.4/5.8GHz respectively. I am looking to get RG213/LMR400 1 meter cable to replace them at some point. Very quick tests with 5dBi vs 7dBi outside, the 7dBi does produces the same performance as the 5dBi, but at about 25% longer range. That is obviously eyeballing the range.

I am tempted to go with 9dBi, but I don't know that I really need better range/performance for outside. My backyard is only about 200ft from the anntenas to the furthest location and it looks like I can probably get >20Mbps at that distance on my phone. Though lower loss wiring is something I'll invest in at some point.

 

[sfx2000]

You know what they say about WiFi - location is everything

The wavelengths are short enough, esp. in 5GHz, that thru Rician/Rayleigh fading, you can quite literally move a few inches and see a significant difference, up to 3 dB in some cases...

Also, note that higher gain omni antennes - more signal (both good and noise), and tighter/flatter beam patterns.

Also, with mobile phones - primary design consideration for most in the 2.4GHz is Bluetooth and near field, WiFi uses the same antenna, but they tend not to be high gain - again, it's about managing the antenna pattern with Bluetooth Headset/Headphone usage, along with BT Hearing Aids.

 

[azazel1024]

Yeah. Just annoying in the testing.

That said, the outdoor testing was nice and easy with predictable results.

I do need a front yard access point eventually. I've got signal over the entire yard, but its only a couple of Mbps down near the street. As in litterally about 1-2Mbps and I can get complete drop outs if I am behind the trunk of one of my pine trees.

It isn't exactly a requirement, but it would be nice to both not burden the inside router/AP that end up serving the front yard by default when trying to connect, especially with a client running that slow and hogging so much airtime. That and, in the mowing season, I wouldn't mind if I could occasionally listen to internet "radio" instead of whatever I had loaded up on my phone when mowing, but I don't know that it would be possible right now with the drop-outs behind those trees (its 4 of them and on a mower, it would probably only be a half a second total drop out).

Anyway, looking at an Engenius EN500 maybe. I feel like even with 5GHz that would probably cover the front yard (half acre, about 200ftx100ft with the AP located vaguely centrally located) with 400mw radio and 5GHz omnis (can always upgrade them to something bigger on there). I'd also prefer 5GHz to keep penetration down, as I'd rather not be pumping yet another AP "in to the house". The backyard AP is located such that the signal has to penetrate as an extreme oblique through the exterior wall of the garage and also the wall between the house and the garage as well At the closest spot inside of the house proper, the signal strength on 2.4GHz is about -72dBm. So inside, nothing is going to try to connect to that AP (because the AP on the main level is in the room nearest the garage, so the weakest signal is about -45dBm and on a different channel.

HOWEVER, the best location for a front yard AP would not be inside the garage with antennas run outside, it would be located on a fence post a third of the way down my drive way, about 35ft from my house, so the signal would be a bit further away, but it would also be a straight shot through a single exterior wall.

I'll probably test using my spare WDR3600 (realizing that the radio is about 4-5dB weaker) and see how 2.4GHz and 5GHz behaves at that location. See if 5GHz would work fine, or if I'd need 2.4GHz and if so, if 2.4GHz will behave okay to everything inside my house (again, I could use a non-overlapping channel for 2.4GHz, but I don't want things even considering connecting to that AP, or if I walk outside to inside, I don't want them sticking with the outdoor AP when the indoor one is going to be stronger. 2.4GHz would, in all likelihood be fine, but if 5GHz can get the job done, then great (and I can run it in 40MHz mode too, instead of 2.4GHz).

Requires burying some outdoor rated cat5e/6 though, and it is about a 100ft run around my driveway to get to the location. At least these suckers are POE and come with the injector.

 

[L&LD]

azazel1024,

Can you give us a diagram or two of your setup, with 3D indicated?

If you are broadcasting from one end of your home to the rest of the areas (on a single floor), then simply angle (aim) the antennae downwards slightly to hit the farthest point of your devices when you're sitting.

The cube shelf is definitely focusing the antennae pattern for your environment beneficially. When the router is placed there, what is the distance from the ground level? I'm guessing it is already ~10 or more.

I would be tempted to test not only 9dBi antennae, but also 14dBi or higher too at your location. While good is great, great will be Awesome!

Thanks for keeping us updated.

 

[azazel1024]

I'll have to do up a diagram. It is a single story rancher with a basement. The Access point is on one side of the house on the main floor on this http://www.ikea.com/us/en/catalog/products/80275887/, top left shelf. So it is roughly 5ft off the floor. Its on the "inner" wall of the room, so located about 12ft from the garage towards the center of the house and roughly centrally located from the front and rear of the house.

The house is roughly 65ft wide and 25ft deep, though the basement is only 50ft wide and 25ft deep (the room that the AP is in only has a shallow crawl space under it, as it was an addition added later to the house). I mostly care about the AP covering the room it is in (my kid's playroom, 2nd living room) and the "next room over", which is the main living room/kitchen area. Combined it is roughly 1/2 of the main level. That and I do care to greater or less degree about coverage in my garage and (for now) the front of my house. Which is does okay, though not great (see mention about adding a front yard outdoor AP at some point).

The basement router (Archer C8), which is high up in the basement (about 7ft off the floor on a shelf) on the opposite side of the house covers the bedrooms above it as well as the entire basement as well as part of the front yard and the bit of the side yard near it that is "shadowed" from the house.

I have a garage access point with antennas outside that covers the entire backyard and one side of the house (the AP in the playroom in the adjoining interior room actually covers the inside of the house...because the aluminum siding attenuates the signal about 25dB, giver or take a little, where as the distance and garage wall of the house only attenuates the playroom's AP by about 18dB, so it is a bit of a toss up what connects to what, but generally stuff connects to the AP inside).

If I could mount the access point on a lower shelf (probably not, it is used to hold bins of my kid's toys. My wife grudingly let me use one of the top spots for the AP) then I might be tempted to try longer antennas, but 7dBi is as big as can fit on the shelf. Larger and they'd need to stick behind the shelves, which I found short of pulling the whole unit 5-6" from the wall (no way the wife will sign off on that) lead to a performance drop of about 20-30% at all locations.

BTW, I did try relocating the router to two completely different locations. One by it's "old haunt" on the entertainment center that holds the "little" 32" TV. It works GREAT in the same room, but horrible elsewhere because it puts the 4ft thick masonry fireplace in the way. I also tried a new location that isn't shadowing it from any location I care about and can be lower (I just added a LAN drop there) and it works pretty well, but I'd deffinitely need different furniture there (and likely won't happen).

Seems like the best location is roughly 3.5ft off the floor, at least with 5 and 7dBi antennas, for coverage on my main floor. That 3rd location had the best performance for all locations by at least a little bit, but isn't terribly viable to keep it there anytime soon (see needing new piece of furniture there, which isn't likely to happen, since "kid's room").

Oh, as an add in, I did try some different antenna orientations and even mixing a 5 and a 7dBi antenna, but straight up and down and all 7dBi worked the best. The basement router DOES have the antennas canted back at about a 10 degree angle (pointed up at the bedrooms above it slightly). That provides the best overall basement coverage as well as coverage above the router. I swapped the 2/3dBi antennas on the Archer C8 for 5dBi antennas off of my WDR3600s, as they are either currently antenna less and not in use, or have 7dBi antennas on them. With the C8, I got a very healthy 10-20% boost in performance everywhere I tested in 5GHz with them and a nice 15% or so boost to 2.4GHz a long/extreme range, though nothing at close and medium ranges on 2.4GHz. BTW, with my current AP 2.4GHz seems to benefit not at all. That said, I am getting ~28MB/sec at all three locations I care about, at the across the house location there might be a slight boost in speed, but it is hard to tell as I don't have a good way to test there (just sitting it on my lap doesn't seem a great standardized way to test, as the angle it is sitting on my lap, not sitting in exactly the same location, etc. doesn't seem like a good way to go. As an example, sitting on the bed across the house, tilting the laptop just 5 degrees is enough to take 2.4GHz downlink from 10.5MB/sec down to 8MB/sec. I tapped off a location at my son's desk, on my kitchen table and on a coffee table to place it in the exact same location each time and didn't mess with the screen tilt between any test).

So that is partly why this testing was so frustrating is that it was so hard to verify gains, where as with the C8, slapping on the 5dBi antennas and some quick tests and it was nothing but gains. Slapping on the 7dBi from the 5dBi on this AP and I saw a bunch of losses. I had to put about 3hrs of total testing between the two antennas and many fine tuned locations (putting one extra book under the AP, then two, then three, scooting forward a few inches, then a few more, then all the way backward, etc.) to finally verify the best location as well as the fact that the 7dBi antennas were at least slightly better than the 5dBi (and a HUGE amount better on uplink within the same room). In some positions and locations the 5dBi were actually faster than the 7dBi. Its just that the best location with 7dBi beats out the best location with the 5dBi, but only by a modest amount.

Oh well, I'll take it. Outside I do think at some point I might try some 9dBi antennas, but I just don't know. It might take getting an 11ac AP out there at some point for me to care enough about attenuation/range/gain and 5GHz performance to think about it.