So, regardless of the quality, those antennas are just plastic that goes to pollute the world.
Just for a commercial reason that the antennas seem bigger…
It would appear that way wouldn’t it!
In my opinion, I’d prefer less plastic, as a smaller antenna is easier on the eyes. Although I expect the rightmost antenna to perform the best, there might have been an engineering oversight. It’s likely they designed it using CAD software and didn’t test it under real-world conditions.
If you have a VNA, it would be interesting to see the VSWR and log mag curves. Engineers often forget that plastic can alter an antenna’s characteristics. They probably designed and tested the antenna without considering its behavior inside the plastic casing. With a VNA, you can see the characteristics of an antenna change in real-time as objects are placed near it, even things thought to be RF-transparent, like a simple piece of paper. Being 1 meter from the antenna, I could see a repercussion on the VNA just by moving my body a few centimeters. During antenna testing, it’s very hard to remove all the variables. It’s the reason why if an AP is performing badly, moving it just one centimeter, or changing the angle of an antenna by 5 degrees, can change things drastically. These things can’t be simulated and must be performed in the field by trial and error.
Could you please re-run the test without the plastic? For an accurate test, please choose 5 locations and run the test from each. Given 3 antennas, that’s 15 tests. Without doing that, you won’t be able to convince me that the rightmost antenna is not the best. If you are willing, also test with plastic, for a total of 30 tests.
Thanks for your thoughts, yes will do, but it will be tomorrow.
Yeah I bought that bigger pair a good while ago thinking the same as you and didn’t see any improvement for my original project so I put them aside.
I think they were advertised as 6db but couldn’t be 100% sure now. Happy to test for you though!
Gosh what a lot of replies. For some reason, phpBB isn’t sending me notification emails anymore so I missed the replies. I’ll have a read when I’ve got more time.
In the 2.4 GHz band? Fair. In the 5 GHz band? Depending on the type of building you live in and the channel you're using, it might be the case that if your neighbors aren't actively transmitting at the moment, there is actually little to no interference to speak of.
By the way, does anyone actually know how big of a factor that is? It makes sense theoretically, and my experience kinda reflects that, but I haven’t actually made any measurements to back it up. Typical wavelengths of radiation used in WiFi are, say it with me, 5-13 centimeters (actually 4-13 if you count WiFi 6E & 7). Which means that the size of a PCB is typically on the same order of magnitude as the wavelength of radiation in question. This means that a metric ton of diffraction must be happening. I remember trying to block some of the signal radiating upwards from an AP by gluing some masterfully disguised aluminum foil to the ceiling (the footprint was the same as the footprint of the case of the AP), as an experiment. And while it wasn’t completely useless, the signal was only getting partially blocked directly above the AP. As soon as I stepped half a meter in any direction, the signal’d get boosted by 20 dBm or something like that. So it’d be interesting to know how much diffraction actually factors into things.
Engineering of any antenna is a pretty complex task. The additional complexity of engineering PCB antenna is that the whole board is then engineered by various experts and inevitably there will be some compromises. But then it’s up to individual vendor whether they actually engineer the particular PCB antenna on the particular board (inside the particular case) or not (I expect some vendors to simply take a design from somewhere without fine tuning it to particular board design and in this case end results can be less than optimal). It does seem that MT does pretty good job at it though.
As to reflective surfaces: in principle they should be grounded for best reflectivity (if they are not grounded, then they will scatter the signal in all directions and due to the size poorly so. So essentially they become a large source of self-correlated interference … which may even become constructive in certain directions). I guess your experiment did not include this part … at least not by design (it may get grounded to some level due to touching actual ceiling). Next it’s important to place it centrally over the antennas to make its effect “omnidirectional” … and at least in my hAP ac2 antennas are not central on the PCB but rather on edges. Which means that for best results, the reflecting surface (aluminum foil) should be offset from the center of case … but gluing it to case doesn’t allow this. And any irregularity of the reflective surface (even if its dimension is only a fraction of wavelength) will cause irregularities to the resulting radiation pattern. The size of reflective surface matters as well, ideally it should “obstruct” signal in whole area where one doesn’t need/want signal to travel (which may be larger than case size if one wants to block half of sphere due to distance between antenna and case) - think of preventing LOS and if reflective surface is very close to antenna, it’s almost the same as blocking Fresnel zone.
So it’s back to fine tuning the antenna design (reflective surfaces are actually part of antenna).
So (I just want to make sure my understanding is correct) the PCB antenna is engineered together with the actual board, so that together they act in a way the engineers want them to act in? I.e. have a certain radiation pattern, gain, etc. Is this process limited by some constraints, like the Ethernet ports kinda blocking the way, or are the results typically roughly the same as having an external antenna and forget the PCB was ever there?
About reflective surfaces, it was a nice read, thank you. There was no grounding, except through the material of the ceiling, which obviously isn’t particularly conductive. I don’t think I observed any weirdness, like constructive or destructive interference, but, to be honest, I wasn’t really looking for it. The foil was glued to the ceiling itself, and not the AP, and was definitely not centered on the antenna. Come to think of it, the location of the foil was really quite suboptimal…
There are a few designs of antennas that can be implemented on a PCB. Mostly they involve the top layer of PCB … but in case of multi-layer PCB something else might be routed via same area using inner layers (and that might affect antenna performance). Elements on PCB in close vicinity affect the radiation pattern as well. And the case affects radiation pattern as well. So in order to produce a well performing device, all the engineering teams should work together to either minimize effects of antenna vicinity … or to adjust antenna design to work around the problem. I fear that some vendors don’t have enough capable engineering teams (or will or means) to do the design verification and/or proper prototype testing.
You mentioned “the signal’d get boosted by 20 dBm” … I guess this indicates some kind of constructive interference, don’t you?
Clearly the best post is Anskys, great pics, gave me a good marketing chuckle.
The bit about speaker size, if I posted that “generalization”. I wanted to use far less kind words, in a hifi forum I would be ripped to shreds.
Not really, no. Please forgive me, I think I just misphrased my findings.
I would be directly on top of the AP, and the signal would be at a level of X dBm. Then I’d move approximately 1.5 meters in any direction, and the signal level would change, to Y dBm. The value Z=Y-X was approximately +20 dBm. The signal was boosted when I moved away from a position directly above the AP, but that doesn’t necessarily indicate constructive interference, since it was getting blocked in the first place.
I didn’t measure a before and after, since I was only interested in the effectiveness of reflective surfaces proof-of-concept-wise, so I don’t know for sure whether the signal was getting boosted relative to the baseline. I had a vague idea of the baseline, though, and even if it was getting boosted where it shouldn’t have been, it didn’t feel like it was by much.
Somewhat OT but remember that MIMO also takes advantage of interference through multipath propagation. So in most cases the quality of the DSP is more important than the antennas, especially for indoor devices.
You know, I think that antenna No3 is all smoke and mirrors, what has surprised me is the spherical antenna no2 is the only antenna to work in my gym on 2.4G, of which I only did 1 test across all 3 antennas. Anyway it’s quick and dirty, I didn’t get time to pull the sheathing off to test, and nor did I want to tbh.
So all tests are uploading to my cAP ax on 5g which is the furtherest AP away and the signal is taken from the router from 5 locations on the ground floor of my house. Apart from the 2.4G test inside my gym of which all but 1 antenna stayed connected, antenna 2 the small spherical one.
Position 1 antenna 1 signal -66 398Mbps
Position 2 antenna 1 signal -66 498Mbps
Position 3 antenna 1 signal -83 154Mbps
Position 4 antenna 1 signal -68 356Mbps
Position 5 antenna 1 signal -65 542Mbps
Position 6 2.4g antenna 1 signal N/C
Position 1 antenna 2 signal -67 502Mbps
Position 2 antenna 2 signal -67 503Mbps
Position 3 antenna 2 signal -83 156Mbps
Position 4 antenna 2 signal -61 540Mbps
Position 5 antenna 2 signal -68 385Mbps
Position 6 2.4g antenna 2 signal -75 39Mbps
Position1 Antenna 3 signal -65 320Mbps
Position2 Antenna 3 signal -72 344 Mbps
Position3 Antenna 3 signal -81 113 Mbps
Position4 Antenna 3 signal -65 505Mbps
Position 5 antenna 3 signal -71 401Mbps
Position 6 2.4g antenna 3 signal N/C
Edit: Just for giggles i removed the sheath off the large antenna No3 before I remove it from the gym, 2.4g now connects -75 20.6Mbps
That was really bad on the eyes. I reordered it in a table.
| Position | Ant. 1 | Ant. 2 | Ant. 3 |
|------------|--------|--------|--------|--------|--------|--------|
| | Sig. | Thr. | Sig. | Thr. | Sig. | Thr. |
| 1 | -66 | 398 | -67 | 502 | -65 | 320 |
| 2 | -66 | 498 | -67 | 503 | -72 | 344 |
| 3 | -83 | 154 | -83 | 156 | -81 | 113 |
| 4 | -68 | 356 | -61 | 540 | -65 | 505 |
| 5 | -65 | 542 | -68 | 385 | -71 | 401 |
| 6 (2.4g) | N/C | N/C | -75 | 39 | -75 | 21 |
Since we have throughput we can safely drop RSSI as it’s less meaningful.
| Position | Ant 1 | Ant 2 | Ant 3 |
|--------------|-------|-------|--------------|
| 1 | 398 | 502 | 320 |
| 2 | 498 | 503 | 344 |
| 3 | 154 | 156 | 113 |
| 4 | 356 | 540 | 505 |
| 5 | 542 | 385 | 401 |
| 6 (2.4g) | N/C | 39 | 21 (no case) |
Some may prefer an horizontal view…
| Position | 1 | 2 | 3 | 4 | 5 | 6 (2.4g) |
|----------|-----|-----|-----|-----|-----|-----------|
| Ant 1 | 398 | 498 | 154 | 356 | 542 | N/C |
| Ant 2 | 502 | 503 | 156 | 540 | 385 | 39 |
| Ant 3 | 320 | 344 | 113 | 505 | 401 | 21 (no case) |
Since we are doing a comparison we don’t need absolute values.
| Position | 1 | 2 | 3 | 4 | 5 | 6 (2.4g) |
|----------|------|--------|--------|--------|------|----------|
| Ant 1 | -21% | -0.99% | -1.3% | -34% | | N/C |
| Ant 2 | | | | | -29% | |
| Ant 3 | -36% | -32% | -28% | -6.5% | -26% | -46% |
Overall this data is rather surprising.
I guess that doing field tests is the only way to tell if an antenna is good or not and to doubt any claims by the manufacturer.
Yup, I told you that engineers don’t test antennas inside their casings. They test them without, Q/A gives the go ahead, they manufacture them in mass, put on the case and ship them without further testing to make sure the case did not alter the antenna.
Thank you for doing the tests and for sharing this data.
thanks for that, lol
Yeah I did multiple tests and took the best figure for each so it was as fair as I could make it, agreed removing the sheath allowed an antenna to connect for an otherwise none functioning device. So for me it was worth taking a view, so thanks for pointing me in the right direction on that.
i still use antenna 2 on my stationary laptop/desktop AX210 150mm pigtails. Antenna 1 is what came with an Asus PCE-AX3000/AX200 card.
Good observation. I have a client with accommodation where they use Kingspan in the walls. Great for insulation but awful for wireless propagation. You can have full signal inside the bedroom. Step onto the veranda and almost nothing. We had to install a couple of outdoor access points.
To be fair, in a residential setting, you’re usually restricted. They’re not going to want you digging plaster out the walls, drilling holes or running unsightly conduit all over. I usually try a Powerline if cost is a consideration and these days, it often is.
Thanks everyone for their input here. Fascinating reading. Not sure I’m that much clearer or have the definitive answer 
Maybe there isn’t one. Certainly get the point that it’s also about the number of antennae for MIMO. That clearly makes sense.
I annoyingly dropped a brand new cAP ac yesterday and then kicked it across the car park whilst it rolled around on it’s edge. More haste, less speed! It stopped working so happened to open it up. Are these the antennae?
Plus in Mikrotik parlance, does number of chains = number of antennae = spatial streams? The cAP ac specified is two chains so I assume a device with four chains would be generally “better”?