Cantenna implementation: just how tricky is it?

[MarkMLl]

A few days ago I needed to check for the presence of a WiFi AP (possibly not actively advertising its availability). I knocked together a cantenna in a hurry from a somewhat-undersized drinking chocolate can, its off-axis rejection was good but its rear-lobe was disappointingly strong: I suspect because the can was too short so the probe was too close to the centre.

I've come across https://web.archive.org/web/20210512030616/https://www.wifi-antennas.com/topic/52-dual-band-cantenna/ which is interesting, but it appears to be an old discussion which went quiet.

* Can anybody comment on the computed dimensions?

* Does anybody recognise the software used for the gain simulations? Is there a half-decent free alternative?

* If having to use "rules of thumb", how does one allow for the horn which various people suggest adds substantial gain?

* If fabricating from "found" components (aluminium ducting, stainless cans etc.) how paranoid does one have to be about good contact? Should everything be TIG welded to penetrate passivation layers?

* How disruptive are internal steps or ridges of a couple of mm at joins likely to be?

Hope that lot makes sense and sparks somebody's interest. There seems to be an enormous disparity between the way people say "this stuff is non-critical" vs the amount of care that goes into professionally-manufactured waveguides.

[joeqsmith]

I tried to make one from a couple of bean cans soldered together. 

https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg4166587/#msg4166587

We tried to 3D print some microwave horns and compare them with a production part. 

[MarkMLl]

I was in a bit of a hurry... can I used was plated steel (not stainless since it appears to be magnetic), about 84mm diameter 125mm long with 30.6mm probe on an SMA about 56mm from back. NanoVNA suggests it's broadband centred on the the 2.43 GHz region with an SWR of 1.4 and doesn't care whether the plastic lid is fitted (to add a bit of strength).

Definitely better than a WiFi whip (I tried both on location since nobody was actually shooting at me). However now that I've got this far I'm interested in how to "do things properly" without having to learn thin-wall TIG welding etc. :-)

[Marsupilami]

Software is CST. Not cheap.
OpenEMS might be a solution for you but it has a learning curve with scripting and stuff.
There were threads here too in the past about affordable/ free and open source em sim software.

[vk6zgo]

I was in a bit of a hurry... can I used was plated steel (not stainless since it's magnetic), about 84mm diameter 125mm long with 30.6mm probe on an SMA about 56mm from back. NanoVNA suggests it's broadband in the 2428 MHz region and doesn't care whether the plastic lid is fitted (to add a bit of strength).

Definitely better than a WiFi whip (I tried both on location since nobody was actually shooting at me). However now that I've got this far I'm interested in how to "do things properly" without having to learn thin-wall TIG welding etc. :-)

I very much doubt that the magnetic properties of stainless or plated steel have any effect at 2.4 GHz.
Certainly, UHF Yagis are quite commonly made of stainless steel for durability.

Interestingly, the term "Cantenna" has a quite different meaning, as an RF test load consisting of a large tin can full of transformer oil with a biggish 50 resistor suspended in it

[MarkMLl]

Software is CST. Not cheap.
OpenEMS might be a solution for you but it has a learning curve with scripting and stuff.
There were threads here too in the past about affordable/ free and open source em sim software.

Thanks for that. I spent a fair while yesterday Googling images etc. and noted OpenEMS: obviously my RF is rusty but learning curves don't generally bother me... I am, after all, a FreeCAD user :-)

[MarkMLl]

I very much doubt that the magnetic properties of stainless or plated steel have any effect at 2.4 GHz.
Certainly, UHF Yagis are quite commonly made of stainless steel for durability.

Sorry, that was a "too much detail" moment: I'd originally written stainless but checked and changed it.

Even if the material doesn't have a bearing on the RF properties, it does on how easy the stuff is to fabricate. Copper, of just about any size, can be intimately joined by soft-soldering, while "found" items made of stainless or aluminium are far more tricky: as alluded to in my OP.

Interestingly, the term "Cantenna" has a quite different meaning, as an RF test load consisting of a large tin can full of transformer oil with a biggish 50 resistor suspended in it

Yes, as well as various Unenlightened who try to use the term for absolutely any antenna made out of a can https://hackaday.com/2018/01/23/an-ads-b-antenna-built-from-actual-garbage/

Unfortunately, we live in a world in which vocabulary is in increasingly short supply, with competing specialties in vicious territorial disputes in their attempts to find pronounceable neologisms. However even the disruption of "peak vocabulary" doesn't absolve people of some minimal duty of care when recycling: one really is supposed to check that nobody's using stuff before appropriating it.

[MarkMLl]

I've just soldered a second (tinned mild-steel?) can onto the first, and am now getting somewhere around 15dB (+-5) off-axis rejection. Centre still at 2.430GHz, SWR looks like it's gone up to about 1.75 (subject to my cheap gear and undoubtedly my poor technique and understanding :-)

For something that's likely to take abuse it's not very robust, but I'd certainly call it an at-least adequate proof of concept.

MarkMLl

[EggertEnjoyer123]

I made a cantenna with about 14dBi gain. I used 4 inch dryer pipe and a 4 to 6 inch duct reducer. (Try to get a thinner pipe because I'm pretty sure 2.5GHz is above the cutoff for another mode in a 4 inch circular waveguide)

It's important to make the length of the dryer tube 3/4 + 1/2 * N wavelengths. I'm pretty sure 3/4 wavelength would work the best but I made my tube longer and it didn't really affect things (though I'm pretty sure it made my bandwidth worse). Mine was 4.25 wavelengths because I didn't have the tools necessary to cut the tube to size, and I ended up trimming a little from the top. Measure from the end of the pipe covered with metal to the base of the slanted part for the duct reducer. The only thing that I found to be very critical was the length of the feed element. It took multiple attempts to get the dip in S11 centered at 2.45GHz. My construction was pretty bad (everything was held together with copper tape, and I just copper taped one end to make the reflector) but it seems to work acceptably well.

Of course my application (which is making an antenna for a radar project) is probably different from yours so you might need to spend some more effort figuring it out. But I did get a higher signal strength when I tested it out with a wifi dongle.

[MarkMLl]

My 84mm is a bit below your 4", and of course mine are straight, but it's interesting that the results appear close. I'm eyeing up some stainless storage containers (one straight, one conical) in case I decide to tidy things up a bit: a cantenna prominently labelled "Tesco Drinking Chocolate" takes "nothing to see here, keep moving please" a bit far :-) The interesting thing there being whether I could solder stainless- using the correct flux- as easily as I soldered mild steel... plus of course how careful one has to be about a small step at the join etc.

Assuming that there's a bit if strength in the tube one could probably drill a spiral or zigzag so that the axial position could be adjusted in (say) 3mm (1/8") steps, and the probe could be made out of something like a threaded brass standoff to make it adjustable.

I've thought about using things like copper tape for joining before. However the thing that concerns me there is that there's going to be an adhesive layer between anything that isn't also mechanically fitted, and that's effectively going to be somewhere between a capacitor and an open circuit.

MarkMLl

[EggertEnjoyer123]

Here's a picture of my antenna (as well as the rest of my project). You can see that the S11 is under -13dB over 2.4 to 2.5GHz. I could not solder onto the steel so I ended up just wrapping copper tape around everything, and then soldering to that. As long as you apply pressure to the copper tape while sticking it down, it makes a good enough electrical connection. I assume the adhesive layer is thin and easily penetrable.

I measured the gain without an anechoic chamber, so take the 14dB number with a grain of salt. I went outside and pointed the antenna at another antenna with known gain, then used the Friis transmission formula to determine the gain of the antenna.

[Marsupilami]

If you guys post the dimensions I'll throw it in the simulator.
The geometry is simple enough for a lunchbreak run and I'm always interested to see comparisons with real life results.

[EggertEnjoyer123]

My cantenna is more than half a wavelength wide and 3/4 of a wavelength long. However, all the other dimensions are the same.

Can you simulate my design with 0.75 wavelength instead of 4.25 wavelengths for the width? I'm pretty sure it would increase the bandwidth of the antenna without losing gain, but I'm not sure about that statement. Would be nice if you could verify that.

[Marsupilami]

It certainly is doing some wavy stuff.

[EggertEnjoyer123]

Looks reasonable.
You can improve the matching by messing with the length of the feed. I think it took me multiple attempts to get the exact length right.

Also I'm wondering if making the tube longer has any effect. Could you try increasing the length of the tube by a few wavelengths?

[Marsupilami]

The feed length change didn't help much, but I realized that my ratios were way off. The tube length and the feed position should use the guide wavelength while the diameter and the feed length is based on the free space wavelength.

I got much better match, but it needs to be adjusted to bring it lower.


I just have to redo my model to be able to properly parametrize it in the sim software for sweeping.

[MarkMLl]

In my case the frequency was quite a lot down, about 2.3Ghz rather than the intended 2.45ish. However it's definitely directional and reasonable amount of gain at WiFi frequencies, which is a win for something knocked together in a hurry for a specific job. Sorry about the crappy photo, I really don't have time to mess around with a tripod etc. and for some reason a flash didn't seem to improve things... ;-)

Must organise some screen capture software, which a bit of Googling suggests should be an easy job.

To my surprise, I found soldering the two cans that I used (as a rush job) surprisingly easy. I sanded off the paint on the outside of one and the lacquer on the inside of the other, put in a small kink so I could nest them, applied SMD-style flux and soldered with 60/40 (lower MP than unleaded) and a PCB rework hot-air gun.

The key point, I found as I was working, was to start at the kink where there's a bit more gap: from that point onwards I was able to go round the joint fairly fast as the solder wicked into it.

@Marsupilami do you have any thoughts on the dual-band cantenna at https://web.archive.org/web/20210512030616/https://www.wifi-antennas.com/topic/52-dual-band-cantenna/ which I posted at the start of the thread? The discussion there appeared to go quiet, and it would be interesting to know how well his probe/loop arrangement might work.

MarkMLl

[Marsupilami]

Marsupilami[/member] do you have any thoughts on the dual-band cantenna at https://web.archive.org/web/20210512030616/https://www.wifi-antennas.com/topic/52-dual-band-cantenna/ which I posted at the start of the thread? The discussion there appeared to go quiet, and it would be interesting to know how well his probe/loop arrangement might work.

I'll run it when I have time to finish the parametrized model.

[Marsupilami]

and it would be interesting to know how well his probe/loop arrangement might work.

Do you care about the 5GHz loop feed thing or the combination of both?
Also there are two sets of dimensions, red and black, is it explained which one is what? Which one did you build?

[MarkMLl]

and it would be interesting to know how well his probe/loop arrangement might work.

Do you care about the 5GHz loop feed thing or the combination of both?
Also there are two sets of dimensions, red and black, is it explained which one is what? Which one did you build?

That's not one I've built, I'm just interested in what somebody who obviously knows what he's doing thinks about that design since discussion stopped abruptly. I think a dual-band cantenna is of general interest if trying to sniff WiFi leakage, even if it were suboptimal for a fixed link. I think- from the discussion- that the black numbers in the drawing are the original (2.4GHz) dimensions and that he's penned in amended red ones after adding the 5GHz loop.

I was digging around, and found that somebody has stainless spaghetti containers which (from elsewhere) appear to be sized approximately 9.5cm diameter and 29cm high, and also has conical biscuit tins of a similar diameter https://www.ebay.co.uk/itm/262943791596 Obviously I'm fully aware of the difficulty of joining stainless without using TIG, but /if/ it were possible to do so the combination might be interesting.

I experimentally snipped 1/2mm off the probe in my existing cantenna yesterday (since I don't immediately need it) which moved the frequency up by a few MHz at the cost of a slightly higher SWR, then snipped another mm off it... with interesting results which might or might not be usable. From that it's clear that while the overall design might be fairly forgiving, getting it optimal takes the same sort of careful design needed for any other antenna type.

Must learn OpenEMS...

MarkMLl

[EggertEnjoyer123]

A S11 of -10dB is decent, and you should be able to use it without any problems.

[Marsupilami]

I'm just interested in what somebody who obviously knows what he's doing thinks about that design

Well, in that case I'm not your guy.
But also I still feel there is no real point to the dual band antenna this way since it still uses two separate feeds. Like if you only had one coax to hook up to you SA then fine, but if you have to have two separate connections you might as well ducttape the two tubes together and then have two properly tuned antennas instead of one that's kinda crappy at both bands. 

Tuning these is not trivial. A single simulation run takes 15+ minutes on my potato computer, exponentially more if I up the accuracy and then there are 4-5 dimensions that can change to sweep over.
I'm sorry, I don't mean to complain. I offered to simulate these as I am interested too, I'm just trying to explain why is it not easy to do anything else then a single,  precisely dimensioned drawing. Being able to tell whether a very particular configuration is good is still somewhat tedious, but to come up with "something like that" just better is an iterative time sync even with proper tools.
At least you can snip the feed wire in real life and see the result immediately on the analyzer. I wish the simulator could do that.

I'll have more time over the weekend I'll play around tho.
Can you post the dimensions of your build as is now? We could start from trying to get a match on the S11 plot between your VNA reading and the sim and work from there on.

[Marsupilami]

@Eggert,

Same for you, can you clarify the dimensions as you built them? The second drawing has wavelengths and stuff. I'd love to see an S11 plot from you and my sim to line up.

[joeqsmith]

If you like, I can provide you with the dimensions for the horn I tested along with details for the transition.   The nice thing is you could start with just the coaxial to waveguide transition.  Once that model is correct, you could then add on the horn.    I also recently replaced the directional coupler I am using to measure S11 with one with a higher directivity (and wider BW) along with some decent home made standards.  I should be able to measure S11 fairly accurately.    I can also measure the radiation patterns as well.   

Keep in mind this was all done in the X-band.  I think I ran those horns at 10GHz.   

***
Comparison of S11 using my original coupler and the new part.  The original coupler was only spec'ed to 8.3 or so GHz and performed much better than I had expected. New coupler allows measurements above 12GHz.   
https://www.eevblog.com/forum/rf-microwave/experimenting-with-waveguides-using-the-litevna/msg4903016/#msg4903016

[joeqsmith]

Looks like the STL file for the large horn exceeds the forum's 4MB limit but I could place it on Github.  Assuming you can work with an STL file.