Unexpected antenna results

[azazel1024]

So the 11ac router I am working with comes with what appears to be 2dBi, or possibly 3dBi external antennas. My laptop has an Intel 7260ac adapter in it and I was hitting around 52MB/sec LoS and fairly close (10-12ft) away. I figured that was pretty respectable. Well I now have a spare WDR3600 and replaced the 5dBi antennas on my other WDR3600 with 7dBi antennas. So I tried swapping on the 3 5dBi antennas from those routers to my 11ac router.

HOLY HECK BATMAN! I expected no LoS/close benefit. In my basement with the router my max speed close went from 48MB/sec slowly ramping over 3-5s to 52MB/sec to an immediate 58MB/sec topping out around 59.5-60.5MB/sec over 4-5s. I didn't test as much transmit to the router, but anecdotally that also seems to be up 10-15% (not as much of a gain as the receive from router).

Most other basement locations also seem to be up 10-15% in performance on 5GHz. 2.4GHz performance is much more minor, maybe a 5% gain. With the limited testing, within statistical margin of error. Upstairs ABOVE the router, what should be outside of the HPBW for 5dBi antennas and more in to the sweet spot of 2/3dBi omnis, I am seeing no difference in 2.4GHz close in and 5GHz I am seeing from a 3-5% gain very close up to about a 30% gain (max, average is more like 10% gain) on my tablet with its 11n 1:1 adapter if I move it a little further away so it is closer to the HPBW sweet spot.

Medium far upstairs I see a similar 10-15% 5GHz gain, but I am also finally seeing about a 5-10% 2.4GHz gain.

Very far/high attenuation I am seeing pretty much no difference between the 5dBi antennas and the others (2.4 or 5GHz). My guess is it is a beam forming thing.

A extreme attenuation, I am seeing a slight gain again on 2.4GHz (too far to connect on 5GHz. With the old antennas it was around -84dBm, I am seeing -82dBm with the 5dBis at that location. At the high attenuation location I was seeing the exact same dBm between the two antennas, at medium/low attenuation spots I am seeing a 2-5dBm increase in signal). At extreme attenuation with 20MHz 2.4GHz, it was about 2.5MB/sec with the old antennas, about 2.8MB/sec with the new ones. Barely anything, but it was very repeatable.

Very surprised by the results. Especially the close in stuff. I really didn't expect any increase in max/close range performance, but there was a huge boost.

At some point I might test 40MHz 2.4GHz to see if there is any difference, but no time and I don't run the router in 40MHz 2.4GHz mode.

I am interested in testing the 7dBi antennas on my WDR3600 now, but no time to do methodical testing (5GHz performance on it tanks as soon as I go outside the room the AP is in, but one room over it has to go through 4ft of masonry fireplace, though 2.4GHz performance barely nudges, maybe 10% loss on 2.4GHz, but a 40% performance hit on 5GHz right now).

 

[TonyH]

Hi,
With gain antenna, keep in mind radiation angle changes. Prolly you hit the sweet spot. Will be interesting to see the result of trying 7dbi sticks or mixture of two.

 

[stevech]

rather than radiation angle, gain omni antennas change from a flattened sphere to a doughnut to a bagel as gain increases.

2-3 dB of gain increase is small compared to the typical path loss from end to end - which is 60dB or more.

 

[TonyH]

Hi,
Keeping in mind vertical antenna is half of dipole(dbd) where half of the radiation element is projected into ground. Dipole pattern cut in half. Gain value is relresentation of peak lobe value. Can be measured by erp too.

 

[azazel1024]

As mentioned, 2-3dBi is a minor change, so I am especially surpised that there is such a huge difference in performance even relatively close with LoS. Its well beyond the maximum I could possibly squeeze out of it before, which was at about 5ft straight at the antennas, I could hit around 54MiB/sec absolute max. Here I was hitting a bit over 60MiB/sec at more than twice the distance, and lower than the optimal location in relation to the router (sitting on the floor 10ft away with the router up by the ceiling on a shelf, antenna radiation not point directly at the laptop, but straight out to the sides).

Not surprised that the longer and longest distance only saw tiny gains, but again still very impressed at the big gains both in absolute max, but also in low/medium attenuation performance gain. With a little more testing last night I am seeing between 5-20% gain in performance up till about -80dBm on 5GHz, at which point I don't see any performance gain. On 2.4GHz, at least 20MHz, I am seeing no gain or at most 5% gain up until -75dBm, at which point I see around a 10-25% gain up to about -82dBm (furthest point I could realistically test). This is compared to the 2/3dBi antennas the router shipped with. There might be a more noticable difference with 40MHz and 2.4GHz, at some point I do plan on testing it.

I didn't do much signal spot checking, but the couple of places I checked showed from 1-4dBm increase on 5GHz and a -1 to 2dBm increase on 2.4GHz. Beamforming might be one of the things messing with outright gain, but also might be why the closer locations saw such big gains with 5GHz (IE explicit beamforming might be working much better with the new antennas than with the old ones).

Unfortunately I won't be able to test with the 7dBi omnis I am getting. Where the router is located, I only have about 1 or maybe 1.5" of clearance (built in shelf right by the ceiling in my basement office). I MIGHT be able to fit 6dBi omnis depending on the exact antenna style, but 7dBi are deffinitely out.

The 7dBi are going on my WDR3600 to replace the 5dBi on there, and of course I'll do plenty of before and after testing at difference locations to see if there is any real improvement or not. I do not expect to really see anymore range, but there are a couple of locations close by that I am hoping will see a big enough 5GHz signal strength bump to see a nice performance gain. For example, my dinning room table sees a lot of laptop use and it is only 10ft from the AP...but it is also through 1 2x4 wall at about a 30 degree shallow angle, and also through about 2-3ft of my masonry fireplace. So the ~25MiB/sec speeds I see line of sight in the same room with the AP on 5GHz, drops to 12-16MiB/sec. 2.4GHz (40MHz) is 23MiB/sec same room and only nudges down to 20MiB/sec sitting at the table there.

Really some of it is that I need to find a better way to relocate it so that the fireplace isn't in the way, but without the fireplace then being in the way of connecting in the same room. Though recently we moved stuff around so the room with the AP is now full time my kid's playroom, instead of it being dual purpose playroom and TV watching room, where my wife and I would sit most evenings and want high speed wireless device access. Now that is the basement with nice clean (or only mildly obstructed) sight lines to the router.

It is making me wonder if I should change things up so that the basement router and the outdoor AP should be running 40MHz 2.4GHz and the playroom AP running 20MHz instead of the other way around. Since the basement and outdoor APs have effectively no location where their signals overlap (well, the tinniest smidgen of a location outside), I've been having them co-channel on 20MHz and the playroom AP covering the core of the ouse on 40MHz. Maybe have to play around with it and see what works best. Though what would really work best is a nice AC1750 router in the playroom and outside for some awesome high speed 5GHz coverage.

 

[stevech]

when visualizing the doughnut shaped antenna pattern of 7dBi and higher gains, take note that from a basement, the upper floors, esp. in areas nearing the equivalent antenna location below, the signal strength will be lower.

You can tilt the antennas to "aim" this doughnut somewhat.

And fiddling with 1, 2, 3, 4 dB of change in signal strength is kind of insignificant in a total path of 60, 70dB or more.

The practical way to improve speeds vs. coverage is to add an Access Point.

 

[azazel1024]

Already have a 2nd access point. Just filing in coverage a little.

I think one issue here is explicit beamforming. Also the antennas are tilted back at about a 10 degree angle to provide better coverage for the area above the router, but still cover the basement okay.

I had been anticipating that the switch from 2/3dBi to 5dBi would result in coverage too low upstairs and I'd switch back, but in ALL locations, even significantly above the router, the speeds increased by a resonable amount, or at worst, saw no change. Laying in bed my tablet is probably 4ft above and about 6ft to the side of the router. With 5dBi omni, the HPBW should be around 32 degress or so...and they are only tilted back at about 10 degrees, since that 32 degrees is TOTAL width, not width off of center line, that means anything above 25 degrees off the router should be outside the HPBW and losing significant strength, especially compared to 2/3dBi omni, which should still place my tablet within the HPBW (should).

Yet I see about a 10% increase in speed on 5GHz laying there with my tablet on my lap in bed.

It makes me curious about explicit beamforming that much more, if it is possible that the higher gain in the one direction allows the router to setup constructive interference on the streams that can result in even higher gain in a direction that the antennas are not optimized for.

No clue, but I do know that the impirical results are not lining up with what I thought they would be based on positioning of devices and test results.

I have done almost NO Rx testing with the router though, it is possible that Rx has dropped off considerably in some locations (especially the locations above the router). Something I need to get around experimenting with at some point.

PS and on top of that, 5GHz gain should be higher than 2.4GHz gain...which also means that the 5GHz HPBW should be even narrower than 32 degrees. Though I guess that said, if the antennas see a 3dBi increase over the old ones, the half power gain, would be identical to the old antennas at their maximum gain, just the extremely off angle gain would be much lower, like directly above the antennas. So maybe that is a bit of it, its off axis, but not so far off axis to not be the same or better than the old antennas.

 

[stevech]

Beamforming is a term today used to mean electronically steered antenna beam patterns. It has to do with altering the directionality of a small or large array of antennas - by controlling the signal phase differences among antennas.

Physically pointing antenna with directionality - I wouldn't call beamforming, but rather, "aiming" !

A gain omni's flat-sphere or bagel radiatin pattern (more gain, flatter), would be affected very much by 10 degrees of tilt change.

I've used 14dBi omnis - about 3-4 ft. long - a stacked array of dipoles inside a PVC pipe - sold by many vendors. They are hard to use as their flatness is about 7-10 degrees. If you put one on a roof of a multistory building, there's a real dead spot close to the building, on out for 100's of feet. These high gain antennas are also a problem in that they really increase the interference and non-WiFi noise received due to their gain, in urban areas.

 

[azazel1024]

I can deffinitely see that. Changing the direction of the antennas could/can impact beamforming though as the spatial location is then different, changing how the spatial streams and constructively/destructively interfere with each other. Changing the gain of the antennas would also possibly do the same thing to, but have better gain to start with, but it might provide other benefits for matching the phases of the spatial streams to increase signal strength.

At least in this case it is about the only thing I can conclude from how positive the difference was in going from 3 to 5dBi anntenas on the AC1750 router. Though, that said, reading the Wall-O-text that is my 5 to 7dBi antennas swap story on my WDR3600 and the testing involved, even on that, once I had the WDR3600 in a proper location, there were neutral to rather modest gains in performance between 5 and 7dBi antennas too, though no increase in maximum performance of the router (still capped at about 25MiB/sec in either band with the bigger antennas, but performance was generally 5% better in 2.4GHz and 10% better in 5GHz in most test locations compared to 5dBi antennas and around 10% better 2.4GHz and 18% better 5Ghz at the furthest location). Arguably its only a 2dBi difference and it lead to those kind of gains...but at the same time, without knowing the exact radiation pattern from the antennas, there might be even more gains due to changes in reflective interference, better "donuting" or whatever (because even the best omnis don't produce a perfect donut, they are almost always a wrinkly donut so even in open space moving around the anntena at the same elevation and exact same distance the beam gain might vary by a decible or two and some produce more even donuts than others).

I do suspect my front yard wifi signal will go kaput when I move the WDR3600 permanently, as right now it has to only go through a few feet of open space (~10ft) and an exterior wall. Now it'll have to go through about 14ft of open space...and around 6ft of masonry chimney and possibly the flue and also with the 7dBi up at cheast level inside, close to the house you are pretty far below the antennas (maybe 6ft below them standing outside with a phone held at cheast level).

Might need to actually do an outdoor AP in front of the house somewhere at some point. I'll have to test that.

 

[azazel1024]

Oh to add, as you've mentioned, the path loss is so large indoors, higher gain antennas generally don't add much or anything to range (unless you are using high gain direcitonal antennas). In my testing between the 3 and 5dBi on the AC1750 router and the 5 and 7dBi on my N600, in neither case could I get a usable connection in a place I couldn't before (not that I've spent much time testing that at all), though I could get a 5Ghz connection laying in bed, where as it was too weak to do that before. That said, it was still not a usable connection (IE pages would not load and file transfers would work, but windows would at least see it and connect to it, where as before InSSIDer would show it at -91dBm, but Windows wouldn't even display it).

Maybe an increase from 5dBi to 9dBi would net an actual usable, if very slow, 5GHz connection. Doesn't matter to me as it wasn't my goal. In both cases though, the small 2dBi increase between the antennas on the two WAPs does generally seem to range from neutral to a resonably significant gain in performance in locations where I could already get a connection (around a 20-25% gain at best, and in most locations the two WAPs see a 5-10% gain, all of this is very repeatable*).

All-in-all a nice boost in performance by simply stepping a "size up" in anntenas.

*I am refering to the N600/WDR3600 when it was physically moved to what will eventually be its final location, when swapping between the 5 and 7dBi antennas, as its current location seems to have, what I can only assume, are some severe signal reflection issues going on.

 

[stevech]

RF signal attenuation in line of sight (free space) is governed by the laws of physics. It's a logarithmic world. An inverse square law world. Good thing it is, as we're able to communicate with the Voyager satellites that are, as I write, some 15,726,392,959,000 meters' distance from us. With a signal round-trip time of 29+ hours.

I really suggest that you forget about 2, 3, 5 dB more antenna gain in WiFi. It's a nit. Antenna gain in point to point links matters more.

 

[TonyH]

Hi,
Keep in mind hi gain means focused beam width to a direction instead of spread RF power omni-pattern. Total radio TX output is constant. The higher gain, the narrower beam width and vert. horiz. radiation angle. So it will help overcome path loss in one pointed direction. Always give some, take some situation. Antenna engineering is still much empirical.
Designed one when tested in open test chamber or field measurements come out quite different needing lots of tweaking by trial and error. Experienced playing with antenna as a life time HAM, professional from HF to MW range to a point of going crazy running out of patience, LOL! 2.4 and 5 GHz basic antenna size is so small,
it is quite easy to make one engraving copper clad blank PCB, wanna get your feet wet on antennas? I had enough fun. now your turn.....? ARRL antenna handbook
will be a good reading material if interested.

 

[azazel1024]

I guess. I'll grant I don't have gobs and gobs of experience with it, but my limited experience with swapping and playing with antennas is that it is worth while. In an indoor environment 2, 3, 4 dBi difference isn't likely to make any difference in extending coverage, but so far in both cases where I have increase gain by 2dBi with an AC1750 router and an N600 router running in AP mode, both times has lead to either no change at worst (in locations I was looking to improve performance), or a modest boost in throughput, with average gain up in the 10+% range in both cases, more so in 5GHz than in 2.4GHz.

It might be a nit...but it was $10 for a pair of 7dBi antennas, which freed up a pair of 5dBi antennas and I had another set of 2x5dBi antennas so I could swap out the 3x3dBi antennas on my AC1750 router.

For a total cost of $10, I've got roughly a 10% average increase in net yield on my wireless network. Granted I've probably now invested >>>6hrs of work in testing and moving things around to verify and optimize, but if you include the optimization finding the better location for the N600 AP, that lead to something in the 50-80% range increase in average performance between locations...which granted is WAYYYY more than the antenna swap resulted in, but that is also what took most of the 6+ hrs of testing.

Compared to the probably $150 +/- for the two devices, a 10% gain for $10 is pretty cheap.

It does also make me think at some point going from 5dBi to either 7 or 9dBi for my outdoor AP might make sense. I'll have to do a bit more site surveying and maybe play with swapping the 7dBi antennas from the indoor AP on it to try out.