Loop vs Cantenna detector

[kb3puw]

Could someone point me to or explain the differences in characteristics and physical implementation of a loop vs a cantenna detector of the type Eduard showed at today's meeting?

[kb3puw]

I figured I would try asking ChatGPT this question and got an interesting answer.  The AI suggested that the cantenna with a monopole had narrower bandwidth and a few other advantages, including easier construction. 

[Eduard Mol]

Making a loop feed is probably just as easy or maybe even easier to make than a cantenna depending on the materials you have available. I chose for a “cantenna” because I could better match its radiation pattern to my relatively shallow f/0.5 dish. A loop feed or simple cantenna would be sensitive to radiation coming in from over the edges of the dish (spillover) thus leading to more noise. By adding a choke ring or flare section to the cantenna it’s radiation pattern becomes narrower so that jt is picking up less spillover. 

As an aside: in my experience LLMs like chatgpt are very unreliable when it comes to niche subjects such as antennas or radio astronomy. Always keep in mind these “AI” systems are still just “stochastic parrots” that predict the statistically most likely string of tokens (words) given the input prompt and the training data. As far as we know LLMs have no real understanding of the concepts you are asking, it is all correlation. And when the volume of training data on a certain topic is relatively small, as is the case with radio astronomy, the LLM is more likely to “hallucinate” an output that is inaccurate. 

Op ma 7 jul 2025 om 05:18 schreef kb3puw <kb3...@gmail.com>

I figured I would try asking ChatGPT this question and got an interesting answer.  The AI suggested that the cantenna with a monopole had narrower bandwidth and a few other advantages, including easier construction. 

On Sunday, July 6, 2025 at 3:22:28 PM UTC-5 kb3puw wrote:

Could someone point me to or explain the differences in characteristics and physical implementation of a loop vs a cantenna detector of the type Eduard showed at today's meeting?

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sara-list/8c2ad488-631d-4a5e-b83f-33eef44bf040n%40googlegroups.com.

[Alex P]

https://www.w1ghz.org/antbook/chap6-3.pdf

http://www.om6aa.eu/Loop_Feed_with_enhanced_performance.pdf

[kb3puw]

Thanks Eduard.  I do use these AI systems quite a bit -and I never trust them at face value - I am aware how they work.  I would not be surprised if they were getting most of their information on this subject by summarizing SARA material  :)

[kb3puw]

Thanks for the links Alex - I am reading them - they are what I was looking for.

On Monday, July 7, 2025 at 4:17:40 AM UTC-5 Alex P wrote:

https://www.w1ghz.org/antbook/chap6-3.pdf

http://www.om6aa.eu/Loop_Feed_with_enhanced_performance.pdf

[James Abshier]

The cantenna is essentially a waveguide, and as such greatly attenuates signals with frequencies below cut-off. This can be helpful in limiting RFI. The loop feed does not have this advantage.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/CAFcm1Q7jQ_Fgji60up9Ooe7iAB-3qVwxfA%2BO---eWax_qxQh0Q%40mail.gmail.com.

[b alex pettit jr]

Freq marked = CellPhone 750 MHz

 

[Don Latham]

Eduard is absolutely correct re chatgpt!! I've found errors especially in feed design etc. And he has the way it works in a nutshell.

Don

---

From: "Eduard Mol" <eddiem...@gmail.com>
To: "sara" <sara...@googlegroups.com>
Sent: Monday, July 7, 2025 12:15:17 AM
Subject: Re: [SARA] Re: Loop vs Cantenna detector

To view this discussion visit https://groups.google.com/d/msgid/sara-list/CAFcm1Q7jQ_Fgji60up9Ooe7iAB-3qVwxfA%2BO---eWax_qxQh0Q%40mail.gmail.com.

--

------------
Don Latham
PO Box 404,
Frenchtown, MT, 59846
406-626-4304

[ja...@ganssle.com]

An example AI screwup: I asked ChatGPT to draw a voltage-follower op amp with offset and unipolar supply:

 

To view this discussion visit https://groups.google.com/d/msgid/sara-list/1584844919.51132808.1751921219092.JavaMail.zimbra%40blackfoot.net.

[Neil Smith]

it's vital to use properly designed prompts with any of these systems. Insist that the model must not speculate, that it must only use published sources, that it quotes those sources, and that it ignores any content from Reddit or other unverifiable sources.

Unfortunately, that means that it will miss any authoritative material that is in books under copyright or hidden behind academic paywalls, or cannot otherwise be obtained without paying the author or publisher.

However, even under those constraints the models can be extremely useful in pointing out things that us mere humans might not have considered. Today I was trying to find the maximum torque setting for my rather beautiful Italian precision machine vice. It is only specified in terms of maximum jaw force. Google's AI thing came up with an interesting suggestion. Having noted the screw diameter and pitch and material of the vice, it gathered material data from other sources to calculate a likely friction coefficient and came up with a very reasonable estimate of the likely tightening torque to achieve a specific jaw pressure with that specific vice.

Where you really don't know what the answer should look like it's always worth trying multiple models and asking similar questions.

The answer to the original question is that Cantennas and loops are compromises that will work acceptably on small dishes, where the dish surface is scarcely large enough to avoid diffraction losses anyway. if your dish is greater than about 15 wavelengths, then it's definitely worth using better feed systems such as picket potter horns, corrugated horns or at least concentric choke rings. The problem is that getting good performance in terms of edge taper and side globes usually means increasing the diameter of the feed and therefore increasing blocking losses. The other aspect is that a large feed with high performance is heavy and has a lot of wind resistance. 

On dishes less than about 10 wavelengths diameter, I prefer a loop or patch feed with a beam-forming ring to control sidelobes and optimise the edge taper to around -12dB. That isn't going to work well on a dish with an F/D much more than about 0.6 or less than 0.3, but deeper dishes are probably best used in a Cassegrain or Gregorian configuration. The point about waveguide style feeds having a very high attenuation beyond their cut-off frequency is very useful if you have local high-powered sources of cell phones or broadcast TV for instance that are below the cut-off frequency. 

Shallow offset parabolic dishes reduce the blocking problem, and the mechanical arrangements are less difficult. On my 2.4 m solid offset, I can use a full sized dual mode horn even at 1.3 GHz without significant blockage problems. However, the dish is scarcely large enough to be an efficient sub reflector in a Cassegrain, never mind as the main reflector!

Most of the high-quality information that's needed for antenna design is contained within very expensive, very heavy books about microwave engineering. Some of those cost several hundred dollars even secondhand, so you can understand why the Tech Bros behind the generic AI slop generators are stuck with using free sources and content from highly questionable Internet forums and websites.

I suspect it might be useful to show the prompts that were used to generate results so that we can all get better at prompt engineering to elicit sensible and useful answers from these tools. Rather than just saying "ChatGPT told me this", it might be more useful to share the prompts that were used to elicit the response, which LLMs were used, and where they differed.

Now if the tech bros were brave enough to use Sci-Hub, that might make for much better results, although perhaps also lawsuits from the gatekeepers of knowledge who operate paywalled science publishing sites.

Neil

[Stephen Arbogast]

I  consulted  my  book  by John  Kraus  on Antennas   but  no  info on feeds...   Maybe this will help  to get started.....   https://en.wikipedia.org/wiki/Parabolic_antenna

[Stephen Arbogast]

https://durenberger.com/wp-content/uploads/2019/11/Kraus_antennas.pdf

[Neil Smith]

Another good starting point is the Antenna Designer's Notebook website. Lots of nomograms and references there. Lots of different horn, patch and cup dipole designs, loss calculators for tolerance errors, various reflector geometries and small DOS/Linux programs:  https://antennadesigner.org/axial_and_radial_corrugations.html

Neil

[Don Latham]

Hi Neil: 

Thanks for the reminder about this site. Has good information.  I have several books on the subject of microwave horns and the like, starting with the grandpappy MIT Rad lab volume. I have a pet peeve (besides paywall papers that never have quite enough information to actually build a copy of the geewhiz design.)  And that is (please pardon the shout ) ALL AND I MEAN ALL OF THE BOOKS NEVER SHOW HOW YOU EXCITE THE HORNS. NEVER!. The horns usually are shown from left to right, and where the excitation part ought to be is ALWAYS JUST A TUBE OR RECTANGULAR SECTION. There whew. I feel a lot better. Are we supposed to use afflatus to see how to get any energy into or out of them?

I'll save the trouble:

afflatus /ə-flā′təs/

noun

1. A strong creative impulse, especially as a result of divine inspiration.
2. A breath or blast of wind.
3. A divine impartation of knowledge; supernatural impulse; inspiration.

---

From: "Neil Smith" <neil...@gmail.com>
To: "sara" <sara...@googlegroups.com>
Sent: Tuesday, July 8, 2025 6:31:42 PM

Subject: Re: [SARA] Re: Loop vs Cantenna detector

Another good starting point is the Antenna Designer's Notebook website. Lots of nomograms and references there. Lots of different horn, patch and cup dipole designs, loss calculators for tolerance errors, various reflector geometries and small DOS/Linux programs:  https://antennadesigner.org/axial_and_radial_corrugations.html

Neil

--

--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/CAGbfs6VvAeaeZbqcjNpzssD8q1o5SL5Nq0FbiAb759EeuyzoCA%40mail.gmail.com.

[Stephen Arbogast]

R2   and  R2.....     what?  R1    and  R1  what?

To view this discussion visit https://groups.google.com/d/msgid/sara-list/001001dbeff1%24c9c52960%245d4f7c20%24%40ganssle.com.

[Stephen Arbogast]

I would say   don't worry  about loop  vs  dipole  feed   just  get something   working....   to learn about   our Universe  then come  back later  to improve the  Engineering  ...   remember Carl Sagan......   https://www.youtube.com/watch?v=dtCwxFTMMDg

[Neil Smith]

Heh heh, yes there is a general assumption that the signal is already in the waveguide. to get from a coaxial connection to waveguide, there are several approaches that will work. However, if you look at how satellite LNB systems work, they just have a probe inserted into the guide that forms the rear of the antenna and that probe directly connects to the circuitry without any need for a coaxial connection. That means no losses. in general though, using a probe to connect directly to a coaxial socket mounted in front of a wave guide short or just the bottom of a can, you have to balance the pin length and distance from the back short to provide a match between the waveguide impedance and the coaxial impedance. In a rectangular guide, there's an additional variable to consider which is the distance from the side wall. In general, a fatter pin tends to give a wider bandwidth to the transition. You can use a tuning screw in the opposite side from the probe or in the backshort to adjust the match and tuning of the transition.

The distance from the back short, and pdiameter and length of the pin are in some ways like adjusting an L-match at lower frequencies. there's an infinite number of solutions, some have a lower Q factor and generally, that leads to lower losses.

You can also use a loop or an L shaped wire from a socket on the back face of the guide, which is then bent and connected to the internal wall. While there are some advantages to this in that it gives DC grounding, the loss always seems to be larger in any practical implementation when compared with a simple probe. If you want dual polarisation in a round guide, you can have two probes spaced apart by a halfwave of the waveguide wavelength. The reason this works is that the distance to the back short is usually a little less than a quarter wave, and the impedance and match repeat every half wave along the guide axis. 

For circular polarisation, you can use a stepped septum design . Withl a probe on either side of the septum, you can achieve left-hand or right hand polarisation, with some isolation between the two ports.

The W1GHZ antenna book site has some excellent analysis of pin position and backshort distance for various pin sizes.

There are many other ways to manage the transition from a printed circuit board environment into a wave guide, using finlines and another structures, but those tend to be used at higher microwave frequencies.

if the antenna and feed are only being used on receive, there is another complicating factor. Getting the best possible noise performance from practical amplifier front ends usually means that the impedance of the input is definitely not resistive at the point where the noise figure is optimal. That means that you don't necessarily get the best possible results from a pin that is right at resonance when measured using a VNA or transmitter.

Ultimately what you are trying to do is place an antenna inside the wave guide so that it will couple with the desired mode that you are trying to generate. That's usually TE10 in a rectangular guide or TE11 in a round guide. The reason that the pin tends to be shorter and closer to the back wall than you would expect is that the guide impedance is usually a couple of hundred ohms and that needs transforming to the coaxial impedance. Making the pin too short looks like a capacitor and the distance to the back wall being too short looks like an inductor, so the result is slightly similar to using an L match.

I don't have access to commercial modelling software, so I use a mixture of rules of thumb from the W1GHZ site and designs that have been demonstrated to work, but also I use openEMS to model the performance of transitions. I usually make a range of back short distances and try many different pin lengths when I'm designing a completely new antenna probe design. Tuning screws can save a whole world of pain, but you need to consider if you can afford the small extra losses from using screws made from material that is less than an ideal conductor, or has exposed threads inside the wave guide. I sometimes use sapphire tuning rods or solid silver rods to help avoid losses. Silver plated brass screws are fairly close to the ideal unless you are chasing the last tenth of a decibel noise performance.

So long as there is enough of a wave guide section in front of the probe to allow the creation of a fully formed wavefront, you can treat the transition section as almost independent of the remainder of the antenna tube. In most cases that means at least a half wave at the waveguide wavelength in front of the pin and before any elements such as steps to generate other modes within the guide. Dual mode horns often use a step or a taper to force the creation of other modes, which if they are generated with the correct phase and amplitude to form a hybrid mode, can reduce side lobes in the antenna pattern. Turrin and Pickett-Potter horns are examples of that. Corrugated horns and choke rings are other ways to reduce edge currents at the orifice, but they add a lot of mechanical complexity compared with the dual mode horns.

When I was testing a new antenna recently, I machined nine solid aluminium transitions with different wave guide backshort distances so I could determine the best possible balance between ease of tuning and transition bandwidth. CNC machines mean cut-and-try is still alive, at least in my shop!

Neil

[kb3puw]

I tried building an H1 Loop as per the instructions and it seems to be working quite well on first analysis (just SWR).  I do not think I could have done very well if I did not have the VNA to do the final tuning.

[b alex pettit jr]

Looks Good.

Pass the Input SMA of the LNA into the base of the cake pan, anchor with the nut normally attaching it to the LNA case

and you are good to go !

[kb3puw]

Will do.  

Thanks for your help.

I will post again once I get the rest of the antenna system constructed and tested.