Antenna design

svenbieg
Posts: 39
Joined: Tue Feb 11, 2020 5:48 pm

Re: Antenna design

Postby svenbieg » Fri Jun 17, 2022 1:35 pm

You are right. The pins could make problems, but only in the layout. We don't have any signals on the WiFi-frequency about 2.4 GHz, You would have problems on the upper side already. These waves don't have any effect, except in the antenna.
For me every antenna looks linear, because it's magnetic field only has one direction. It has a north- and a south-pole, You can't generate two fields with only one antenna. The inverted F-antenna simply is a misconception and hurts me in the eyes.
The distance is relevant for speed, the length and direction of the antennas, something between, noise caused by reflections, and Your self-made interference with phase-drift. ;)
Please don't get me wrong, i'm trying to help.
With my formula my design looks better. I just connect the two end-points and get my effective antenna, a simple vector-addition. The inverted F-antenna makes the calculation a little more complicated. I have two diagonals, the effective antenna should lay somewhere between with an interference. Two points seem to be connected to the transceiver, because interferences are bad for signal-quality.
It is just not possible to get a signal, having two effective antennas in a plane.
Again, please don't get me wrong! The WiFi on the ESP32 is working fine, i even get a signal through thick walls.
Last edited by svenbieg on Sun Aug 06, 2023 3:54 am, edited 49 times in total.

svenbieg
Posts: 39
Joined: Tue Feb 11, 2020 5:48 pm

Re: Antenna design

Postby svenbieg » Sat Jun 18, 2022 6:41 pm

What You need to get the expected connectivity is two orthogonal antennas.

Image
Last edited by svenbieg on Sat Jul 15, 2023 11:40 pm, edited 4 times in total.

ESP_Sprite
Posts: 8921
Joined: Thu Nov 26, 2015 4:08 am

Re: Antenna design

Postby ESP_Sprite » Sun Jun 19, 2022 1:48 am

Note that those 'straight' antennas are a quarter lambda... meandering those gives you 3/4th lambda, which leads to better performance (ref here: "Replacing conventional PCB line in PIFA by the meandering line and meandering shorting strip improves the efficiency of the PIFA as well as the bandwidth."). If you do that, plus maybe move the antennas around a bit so you aren't forced to put the module at the edge of your PCB, you end up with something like this.

svenbieg
Posts: 39
Joined: Tue Feb 11, 2020 5:48 pm

Re: Antenna design

Postby svenbieg » Sun Jun 19, 2022 1:51 am

This looks better, but i like my design the most. If You use Your design, You need to cut off the T to eliminate the interference in Your antenna.
To understand this interference You have to draw a sinus-curve starting at the end of the antenna. Then You draw another one from the top of the T and You can see that the curves are out of phase at the receiver. You have to put the short parts upside down to the end of the antennas.

Image

The bandwith from 2.4GHz to 2.5GHz is given by the WiFi-standard and cannot be increased, those people obviously don't know what they are talking about.
An antenna captures all frequencies, the receiver filters out the interesting ones. The effeciency is proportional to the length of the effective antenna, not to the length of the antenna-line.
You just have to prevent tree-structures and put the end of the antenna to a maximum distance, then You get the maximum amplitude.
Last edited by svenbieg on Wed Jul 19, 2023 9:30 pm, edited 33 times in total.

ESP_Sprite
Posts: 8921
Joined: Thu Nov 26, 2015 4:08 am

Re: Antenna design

Postby ESP_Sprite » Sun Jun 19, 2022 7:42 am

svenbieg wrote:
Sun Jun 19, 2022 1:51 am
This looks better, but i like my design the most. If You use Your design, You need to cut off the T to eliminate the inteference in Your antenna.
Note that the image in that PDF is an artists impression.
The bandwith from 2.4GHz to 2.5GHz is given by the WiFi-standard and cannot be increased. An antenna captures all frequencies, the receiver filters out the interesting ones. This people obviously don't know what they are talking about. The effeciency is propotional to the length of the antenna, and not to the length of the line.

... I give up.

svenbieg
Posts: 39
Joined: Tue Feb 11, 2020 5:48 pm

Re: Antenna design

Postby svenbieg » Sun Jun 19, 2022 8:10 am

I know what You mean. The length of the line has to be a multiple of the wave-length lambda to get a standing wave. The phase-drift is caused by the reflection of the wave at the end of the antenna. To understand this You have to draw a sinus-curve starting at the transceiver. At the end of the antenna You invert the phase and go back to the transceiver. My teacher once showed me with an analog oscilloscope, and this is the picture the display has shown:

Image

You should avoid inverting the phase at half lambda without studying, just by trying. The energy in Your system looks the same with a half curve, but the resonant curcuit in the transceiver works best with a full sinus. In Germany we call it a "Schwingkreis", a swinging circuit.

The wave-length lambda is about 12cm. Electro-magnetic waves move at the speed of light, we simply divide the speed by the frequency. The lines should have about this length:

Image

Meandered lines, high-school-physics. You may think these lines are interferring, but this is not true. This are antennas in a plane. You never tried one of these antennas, right? This solution is almost omnidirectional in 3d-space, and it is cheap. Maybe You can have two transceivers and use only the one with the stronger signal. Even one of these antennas should perform better than Yours.
So now, our first working design is the best in the world. Nothing more, nothing less.

Image

Thank You for Your investigation!

Happy greetings to Your marketing-department, they should have written WiFi with small letters! ;)
Last edited by svenbieg on Tue Aug 01, 2023 4:50 am, edited 175 times in total.

svenbieg
Posts: 39
Joined: Tue Feb 11, 2020 5:48 pm

Re: Antenna design

Postby svenbieg » Fri Jun 24, 2022 12:54 pm

Last but not least, Your photo!

Voyager 1 by NASA:

Image

Thank You very much for giving me this platform!

Best regards,

Sven Bieg

ESP_Sprite
Posts: 8921
Joined: Thu Nov 26, 2015 4:08 am

Re: Antenna design

Postby ESP_Sprite » Sun Jul 23, 2023 4:51 am

Okay, kicking an old topic as you nerd-sniped the shit out of me when you started this topic. I knew for a fact that your antenna configuration probably wouldn't perform well as any antenna designs currently in use, because of the logic that your antenna is so dead simple: if it really was that easy to get optimal performance, everyone else would have stumbled on the idea and we'd all use the antenna.

The issue is that that is only circumstancial evidence, though: perhaps humanity ignored the trivial solution all this time? To prove that the meandered printed F-antenna we have on our current designs works better, I'd either have to be able to analytically explain why or to simulate or measure one. I'm decently sure learning why antenna designs work is not something I can pull off in a few days, so I spent the time instead learning a simulation package, in this case OpenEMS.

As a reference, here's our mPIFA antenna as seen on the WROOM modules.
ant-mpifa.png
ant-mpifa.png (202.37 KiB) Viewed 3197 times
I needed to convert the ever-loving shit out of the thing (Gerber -> PDF -> Inkscape SVG -> png -> Inkscape SVG -> Freecad -> STL -> OpenEMS) to get it to import in OpenEMS, and I think some error creeped in the scaling, as seen in the results:
ant-mpifa-res.png
ant-mpifa-res.png (139.42 KiB) Viewed 3197 times
Of major interest is the large S11 graph on the right. As described here, the S11 ratio is the amount of signal *NOT* radiated by the antenna. If this is very low at your target frequency, you have a good antenna. In this case, there's a really nice low -20dB minimum around 2.7GHz. This should be at 2.4GHz obviously, but as I said it's likely the antenna got scaled down somewhere along the line, increasing its frequency.

Now your antenna design. As your specs aren't much more than 'meander a line for about 12CM', that is what I did. I took a fairly random trace width, trying to keep your drawing in mind but also the size it occupies on a module. For the substrate, I use the same 0.8mm FR4 as I did for the mPIFA.
ant-sven.png
ant-sven.png (204.96 KiB) Viewed 3197 times
And here's the result:
ant-sven-res.png
ant-sven-res.png (77.43 KiB) Viewed 3197 times
Note the antenna does actually work: there's a minimum nicely around 2.4GHz in the S11-graph. Unfortunately, this is only at -5dB, meaning that for WiFi, this antenna will emit about 32x less energy than the mPIFA we use right now (ignoring the scaling error in the previous simulation, in other words assuming the actual minimum is at 2.4GHz).

All in all, it looks like the 'meandered whip' you made is actually functional, but it has a very low Q-factor; it trades a high gain for a high bandwidth. Unfortunately, that is not very efficient for WiFi which uses a narrow range around 2.4GHz and doesn't need the rest of the band where this thing works well. If this were for e.g. an UWB signal, it would be the other way around: the mPIFA would not be ideal as it has a narrow bandwitdh and your meandered whip would work a lot better. (But also note that UWB modules tend not to use either of these designs: they likely use something that given the bandwidth they do need has the highest gain possible.)

I'll include the OpenEMS scripts I used/wrote for this as an attachment as well; I'm done playing with this but if you think I did something wrong or you want to mess around to get a better result, feel free to use those as a starting point.
Attachments
antenna-sim-sven.tgz
(38.5 KiB) Downloaded 207 times

svenbieg
Posts: 39
Joined: Tue Feb 11, 2020 5:48 pm

Re: Antenna design

Postby svenbieg » Sun Jul 23, 2023 4:42 pm

Thank's a lot! I gave it a try, and the result of my antenna looks much the same as Your current design:

Image

The S11-graph on the left has a positive share, looks like the negative half sinus.
Both graphs on the right side have a higher scale. Unfortunately i don't hit the 2.4GHz, maybe i have to play around with the length.

Looks like my graphical solution from high-school isn't that bad, You just forgot to put the end of the antenna to a maximum distance.

Image

You can find the STL and the Python-script in the attachment.
Attachments
highschool.zip
(119.89 KiB) Downloaded 190 times
Last edited by svenbieg on Wed Jul 26, 2023 7:42 pm, edited 10 times in total.

ESP_Sprite
Posts: 8921
Joined: Thu Nov 26, 2015 4:08 am

Re: Antenna design

Postby ESP_Sprite » Mon Jul 24, 2023 1:49 am

That is interesting. I took the feedpoint more-or-less from the existing PIFA I modified. Didn't imagine there would be such a difference in performance, honestly. Do note, however, that you demonstrated at best parity with the PIFA rather than superiority, but I didn't expect that to be possible either, I'll admit.

One thing I do have to note is that your antenna is a lot more directional, with minimums of -7.5dB in the far field; the PIFA only drops to -4.5dB. That means the orientation vs the access point matters a fair bit. This is around the 90 degree angle, though, which means that the dual-antenna setup (the 'T-shirt ESP32 module' design) would compensate for this. It does matter for a single antenna though.

Another thing is that my hacked-together design seemingly is not perfect; the literature suggests that a 'real' PIFA should get an extra 4dB or so if tuned properly. As I said, I don't really want to invest more time simulating as I've got better things to do, so I'm not gonna try to eke out that extra 4dB in the sim.

Who is online

Users browsing this forum: No registered users and 99 guests