Waveguides for Hobbyists!

[bingowashisnamo]

Waveguides are expensive. A quick look at ebay shows used waveguide to coax transitions for well over $100. These prices are a road block to most hobbyists who might be interested in playing with waveguides. Can dirt cheap waveguide components be made so that hobbyists can build and learn from them? With a 3D printer, some copper tape and hardware I've managed to build a functioning coax to WR159 for under $5 worth of materials (including the cheap chinese sma adapter). I recently bought a LiteVNA that works up to 6.3GHz and so selected WR159 and designed for the wifi band around 5.7GHz.

For the transition I used a two-step end launcher design and was surprised to measure decent performance. I think with some more careful craftsmanship the insertion losses can be reduced. Since my setup is just the LiteVNA and what came with it I had to measure two transitions in back-to-back configuration. For the return loss measurement I just added a load at one end. Once the transition worked I added a couple of Ebends to make a very simple circuit.

With this cheap and feasible method I want to start building more complicated components. If anyone else has tried this or other cheap methods of building waveguide components I'd love to see your work.

[joeqsmith]

Welcome to the forum.   I started out with using PCB materials, hobby brass and printed plastics.   You can read about my attempts here:

https://www.eevblog.com/forum/rf-microwave/experimenting-with-waveguides-using-the-litevna/

[mr ed]

Me too.  Brass, tape and el cheapo  diode. It's addictive.

https://groups.io/g/TekScopes/photo/270729/3525291

[bingowashisnamo]

Thanks Joe, I saw your series on waveguides. In fact, it was after the viewing video with the cool frequency extender that I decided to buy a LiteVNA.

Mr. Ed you are right that it is addictive. It's fun to have your creations come to life. I'm going to see if it's possible to get a septum polarizer working with the plastic + copper tape method.

[DaJMasta]

Shahriar of The Signal Path has done something similar too, using conductive shielding paint on printed parts:

Probably a little more uniform application than the foil tape (and foil tape is probably more conductive), but still with the limitations of layer height and that roughness.  I wonder if using a conductive PLA as the base improves the performance at all, and I'll bet using an SLA printer to get much finer finish parts would be even better for performance.

[joeqsmith]

Thanks Joe, I saw your series on waveguides. In fact, it was after the viewing video with the cool frequency extender that I decided to buy a LiteVNA.

Mr. Ed you are right that it is addictive. It's fun to have your creations come to life. I'm going to see if it's possible to get a septum polarizer working with the plastic + copper tape method.

Your printed foil parts look much nicer than what I was doing.   

The latest firmware for the LiteVNA allows up to 9GHz but the dynamic range was poor.   The only option I could come up with to use the LiteVNA was to run it below 4GHz and extend the frequency.   If you decide to try it, my Solver software supports external mixers.  It also supports waveguide calibration.   

I wonder if using a conductive PLA as the base improves the performance at all, and I'll bet using an SLA printer to get much finer finish parts would be even better for performance.

We tried various plastics and attempts to coat/plate them.  Link to video showing several horns using different techniques.
https://www.eevblog.com/forum/rf-microwave/experimenting-with-waveguides-using-the-litevna/msg4698596/#msg4698596

[bingowashisnamo]

That's interesting. For waveguide calibration do you mean the TRL method? I was thinking of trying to print some PLA (or perhaps buy a different plastic) pyramid type loads and see if they can half decently absorb RF energy. I saw a paper somewhere on the web where they did such research, I'll have to dig it up.

I thought about applying the conductive paint although I was skeptical about its conductivity and it's cheaper to just add the copper tape. True, the surface roughness play a large role in determining conductivity. At these high frequencies getting an RF design is half the battle, considerable care needs to be taken when building the circuit. At some point a hand crafted design becomes hopeless and the other options are expensive: EDM, machining or even 3D printing metallic structures.

[joeqsmith]

I had provided details on the number of coats of that coating were required before I started to replicate the results of the metal parts.   

I gave up fairly early on the home made parts and started to hunt for used.  Cost wasn't too bad but it's not readily available.   I had posted a paper on where all the surplus waveguides ended up when the mom and pop salvage yards began to cash out. 

I started to talk about calibration here.     
https://www.eevblog.com/forum/rf-microwave/experimenting-with-waveguides-using-the-litevna/msg4792883/#msg4792883

[ArgyllGargoyle]

I was thinking of trying this technique on a horn antenna. Or, how about 3d printing a parabolic dish and copper taping it? You could print it in segments to yield a bigger dish than your printer stage.

I also saw a few papers claiming that the loss tangent of PLA wasn’t horrible. So you could possibly use it for a dielectric and not just a mechanical element. The dielectric constant could be tuned from around 1.5 to 2.5 by adjusting the infill percentage.

This is a 2x2 patch antenna array for 2.4ghz using PLA as a substrate. I print the patch just 0.2 or 0.3 mm raised above the substrate- this makes it easy to put copper tape down then trim it with a hobby knife. You can build other microstrip circuits this way.

[coppercone2]

quality parts require deep understanding of throw chemistry for plating baths. Their getting quality because they know exactly what they are doing, in my opinion.


The biggest problem I have is that I find the electroless plating baths needed for starting a plate are a bit spotty, and I think the paint is work of the devil.... I followed up my work with calipers and it never seems quite right.

and I am not to happy with what plating actually is, I tried to make some very proper electroplate solutions, and the result feels... off. It's not like normal copper sheet. I think it has high internal strain or something, it tears differently. Well I noticed it on PCB's too, if remove the copper its not exactly the same as sheet you would buy.

The plated plastic is not like plated metal, I think the shell is basically ready to 'pop off' the object its on, the substances are too dissimilar.


What I did not rule out is that it was the paint that is pissing me off, because I only did electroplating of painted objects, and experiments with electroless plating that were never followed up by electroplating on plastic or ceramic objects. I wonder if it would be less likely to peel.

I found that i find thick film on ceramic to be a more interesting use of my time compared to trying to metalize plastic. However, its still appealing for advanced structures like meta materials, but I don't know enough yet to get into that

tape really does work, but I hate tape


I am thinking that you need a special paint, like a plating primer, not a normal 'sleek' paint. Something that leaves a sand paper finish. The EMC products have a surface finish that I think is non condusive to good plating between such shoddy conductors  I think its a difficult problem if you want it good.

[joeqsmith]

Link to Maury Microwave WR90 calibration standards user guide:
https://maurymw.com/wp-content/uploads/2023/12/WR90-31.pdf

If you read all the posts in that thread about calibration, while there was some suggestion that my math was incorrect,  Solver uses the same format as Agilent and I attempted to use their math.   As of today, I am not aware of any problems with these calculations.   The newer released versions of Solver support waveguide calibration by selecting the waveguide connector type.   

[bingowashisnamo]

Hi Joe, yes now I remember that video. It was a short, offset short and load calibration. That's cool that you implemented that in your software. I will try it out as I plan to build other types of waveguide components and being able to de-embed past the coax-to-waveguide transformer will allow for better characterization of the component.

Coppercone, I understand why you don't like tape. It's not at all professional. It is strictly for amateurs or if used in a commercial setting just for temporary debugging. What tape does offer the hobbyist is a combination of acceptable conductivity, its inexpensive and easy to use. For the hobbyist who is playing around with prototypes that's a good combination.

Argyll, I like your patch antenna! Did you feed it with some resistive or reactive dividers?  I think your idea about the horn and reflector would work. A horn has very small loss to begin with so the added losses from the application of copper tape would be minimal to the overall system. How would feed the horn and would this also be for the 2.4GHz band like you patches?

[mr ed]

I tried soldering and taping together brass to make horns, copper tape etc. for waveguides but was never happy with the results. As I wanted to join the 100GHz club I decided to go more or less pro and make proper brass cavities and the like.  I also decided to avoid any commercial guides entirely. That's why I decided to learn enough FreeCAD to make my own hardware. Freecad can be daunting as there are unwritten design rules and drawing approaches you need to learn. Bugs Too. But, simple rectangular bits and pieces are easy enough. Also, in the mm range the parts are small and cheap to mill. So, for 24GHz and above (wr-42) it's not so expensive. Those of you with a machine shop are all set.  Brass is easily polished smooth. I figure 20 hours on Freecad (or fusion) would be enough for basic waveguide work.  This brass work (https://groups.io/g/TekScopes/photo/270729/3849179)  was under $100 offshore. Its 0.1mm accurate by spec but possibly better so it would be good in theory to 250GHz! This one has a built in horn  exsentially for free. Have you priced out commercial horns? Eye watering $ for the ultimate performance which is not needed to experiment with. So, go modern go 3d, go brass milling and go crazy.

[coppercone2]

I figure the reason why people are interested IS the irregular structures though, I guess I never thought about it from a "make a QSO" prospective that you have. I figured most end game would be some thing real weird, which is why I put alot of effort into these obnoxious manufacturing techniques that may have some unique capabilities if they are mastered, but its alot of stress for slow results lol

but on the other hand if you can't make standard parts its like not having good test leads for a experiment which makes it ultimately fail.


But there is another aspect to this, that is simple parts with bad aspect ratios for machining, but I guess that is not a problem unless its a serious installation

Especially with high frequencies, the connected at the end seems not achievable with machining techniques, I think there is a giant wall there, but I am seeing that there does not seem to be a good way around it, you need both machining and the 'other' techniques to be practical/high quality. The machining part is difficult to learn and time consuming but somewhat standardized and approachable, the other part is a total crap shoot

perhaps you are right that machining is the way to get some of the system running, but the other part is necessary. I am not convinced that you can get away from machining by only using the other part, but it seems like the way most people would try to do it because in theory its alot easier to plate something or what have you, but I am seeing nothing but obstacles trying it that way.

machining everything that can be machined is going to make the errors go way down, unless you truely master some exotic manufacturing methods. But, it depends on the experiment, some experiments are so insensitive (or sensitive if you think about it conversely) that it almost does not matter if your hookup is crap, and some experiments are the opposite where you won't see jack shit unless the rest of the system is NIST grade. However, given that the MMWAVE test equipment (calibration? whats that) people are likely to have are already going to be dodge city, having the best hookup possible probobly will help lol

[joeqsmith]

I don't have a lot of interest in experimenting with communications.   The closest I got was measuring the radiation patterns from those printed horns.   Making a working EPR spectrometer  required a load, tuning cavity, iris, hybrid Tee, isolator and two coaxial transitions.  There are some benefits to working beyond the X-band like I have demonstrated but I can't see custom making these parts.   

[ArgyllGargoyle]

Argyll, I like your patch antenna! Did you feed it with some resistive or reactive dividers?  I think your idea about the horn and reflector would work. A horn has very small loss to begin with so the added losses from the application of copper tape would be minimal to the overall system. How would feed the horn and would this also be for the 2.4GHz band like you patches?

I am planning to feed the 2x2 patch array with a microstrip feed network. Initially, that is going to just be a 1:4 power divider using 3 Wilkinson dividers. I also plan to experiment with a monopulse network to see if I can do basic tracking. This is all at 2.4GHz

Here are some images of the process I'm experimenting with for rapid prototyping of microstrip circuits. I start by generating a layout with a CAD tool. Then, I export the artwork and import it using OpenSCAD. I print it in PLA and apply copper tape. Since the microstrip traces are slightly raised, it is easier to trim the copper tape using a hobby knife. Then, I solder on connectors (quickly). I find this much easier than just measuring and cutting the tape since the 3d printer maintains decent dimensional accuracy over the whole circuit.

You have to be careful not to melt the PLA too much when doing this. In the images below, you can see that I need a second rev on this - the connectors are too close together, and the 100 ohm resistors for the Wilkinson dividers were too small at 0402 size, leading to some melting of the PLA as I tried to solder them. But the whole process only took about an hour from start to finish, so I can try again quickly.

[coppercone2]

I thought the dielectric constant would be inconsistent with 3d printing

resin seems more solid

[ArgyllGargoyle]

It's a good question about the dielectric properties (relative dielectric constant and the loss tangent). The papers I saw suggested that the loss tangent was on par with or maybe even better than FR4 -- but not as good as a 'real' RF substrate. How much does it vary between brands? different colors? Humidity?
I have been printing with white PLA with 100% infill, but I also have a spool of unpigmented PLA that I want to try.
A lot of the loss with FR4 microstrip is due to the copper roughness from the adhesion treatment. The copper tape approach should be better in that regard.
Maybe I'll print up some resonators and try to make a measurement of Er and tanD.