AI Designed 20-1300MHz Dipole

[Enforcer52]

A friend owns a company that is designing AI software for engineering purposes and asked me to do some test runs on it. After having the system scrounge the internet for information on antenna designs I then asked it to design a small 20-1300MHz coiled dipole antenna that required no ground plane. After crunching the numbers for a while it turned out a design for an overall 24 1/4" dipole.

Got the wire diameter it suggested and then headed to another friends shop that works on racing cars and has all the equipment I needed. Turned the coils on the wire as directed by the AI design, and made both halves of the antenna. Connected both halves to a non-conductive center piece and soldered the LMR coax to both halves. Then placed the antenna into a 27 inch piece of 1/2 inch pvc, capped the top end, and placed a 1/2 inch T on the bottom end with the coax run out thru it.

Next was mounting the antenna on my suv luggage rack, which required two rather large U-clamps because of the width of the luggage rack.

Tested out the antenna on my SDS200 against my 20-1300MHz mag mount antenna. RSSI strength was any where from 10-25 points better on the home made dipole than on the mag mount depending on the range of the system being monitored. This was while parked in my driveway and riding around the neighborhood. I am going to take the antenna on a road trip and see how it works on the road.

My next step will be to build a second antenna, place it in pvc, cap both ends and raised it up on a pole the same height as my Omni X and compare the two antennas. If I get good results will then raise it up over a tree limb about 50 feet in the air and see how it does.

So far pleased with the results and hope the road test and pole tests are just as promising.

 

[prcguy]

Can you post info on the coil(s) and other antenna specifics?

 

[Enforcer52]

I'll check with the engineering company and see what they say since it was their computer who came up with the design.

We are still just checking it out and seeing if it needs tweaks, as soon as that is done and with approval will post the design.

 

[hazrat8990]

I'm interested. Any updates??

 

[merlin]

That poses a-LOT- of interesting questions. I'd like to see an antenna test range's results.

 

[Enforcer52]

I am still doing range tests on the design with the antenna on my vehicle, but most have been only when traveling within a 75 miles of my house. I am planning a trip to the DFW area in a couple of weeks and will be able to compare the antenna with my regular mag mounted (20-1300Mhz) antenna during the trip. Will have a stretch of about a 100 miles with nothing but conventional systems sandwiched between Houston Area and DFW area.

Will let y'all know how it goes.

 

[eorange]

"Ai for engineering" is pretty broad. Does this company include RF work in their portfolio? That makes me wonder what data they used to train their models.

 

[Enforcer52]

"Ai for engineering" is pretty broad. Does this company include RF work in their portfolio? That makes me wonder what data they used to train their models.

Yes they design and build microwave antennas, both commercial and military. Recently branched out into AI and hired my friend's AI company for assistance. Not privy to all their training models, just advised them of what I would like to have for scanners and sdr's. The owner has a vacation house in my sub-division on the lake, and knowing I was into scanning (because of all the antennas on my house), asked for my input on the lower bands which they had never worked on.

 

[eorange]

Wow, nice. Keep us updated!

 

[Ubbe]

After crunching the numbers for a while it turned out a design for an overall 24 1/4" dipole.

Turned the coils on the wire as directed by the AI design, and made both halves of the antenna. Connected both halves to a non-conductive center piece and soldered the LMR coax to both halves. Then placed the antenna into a 27 inch piece of 1/2 inch pvc, capped the top end, and placed a 1/2 inch T on the bottom end with the coax run out thru it.

It's artificial intelligence and not real intelligence. The program needs to be coded by programmers so it will understand what it would need to find on the internet and then average the result. You can't make a 20Mhz-1300MHz antenna. The span are too wide even for a discone that would be closest antenna type to use, and 20MHz would make any antenna huge in size.

It seems it made a dipole with shortening or loading coils that are tuned to one frequency and its multiplar. The coax are connected to the antenna without any proper balun or decoupling and the coax braid are connected to the lower element and that braid goes in the tube along the whole lower element in parallel close to a direct connection to it and out at the bottom?

/Ubbe

 

[dlwtrunked]

A friend owns a company that is designing AI software for engineering purposes and asked me to do some test runs on it. After having the system scrounge the internet for information on antenna designs I then asked it to design a small 20-1300MHz coiled dipole antenna that required no ground plane. After crunching the numbers for a while it turned out a design for an overall 24 1/4" dipole.

Got the wire diameter it suggested and then headed to another friends shop that works on racing cars and has all the equipment I needed. Turned the coils on the wire as directed by the AI design, and made both halves of the antenna. Connected both halves to a non-conductive center piece and soldered the LMR coax to both halves. Then placed the antenna into a 27 inch piece of 1/2 inch pvc, capped the top end, and placed a 1/2 inch T on the bottom end with the coax run out thru it.

Next was mounting the antenna on my suv luggage rack, which required two rather large U-clamps because of the width of the luggage rack.

Tested out the antenna on my SDS200 against my 20-1300MHz mag mount antenna. RSSI strength was any where from 10-25 points better on the home made dipole than on the mag mount depending on the range of the system being monitored. This was while parked in my driveway and riding around the neighborhood. I am going to take the antenna on a road trip and see how it works on the road.

My next step will be to build a second antenna, place it in pvc, cap both ends and raised it up on a pole the same height as my Omni X and compare the two antennas. If I get good results will then raise it up over a tree limb about 50 feet in the air and see how it does.

So far pleased with the results and hope the road test and pole tests are just as promising.

"AI" has become a much abused term like "technology". In what way is this an "AI" design?

 

[nd5y]

You can't make a 20Mhz-1300MHz antenna.

It sounds like they need to hire a real engineer that actually knows something about antenna design and testing.

 

[Enforcer52]

The main company are a bunch of engineers with many years experience in microwave antenna design and manufacture and many computer programs to help them out. They ventured into AI as a way to help improve things. I just got a chance with the AI team to go outside their normal parameters to see what the AI could learn and produce.

The AI company doesn't build antennas, and the main company is not interested in producing "hobby" antennas, so I took on the task to build the design myself and try it out. I know it has a long way to go yet, and the mount on my vehicle is not the ideal way to mount a dipole, but that will be figured out as time goes on.

It's a work in progress, so give me a break!!!!!!!!!!!!!!!!!

 

[JustinWHT]

design a small 20-1300MHz coiled dipole antenna

In Yagi antenna design larger diameter elements increase the bandwidth.

/sarcasm moden
Use a 55 gallon drum as radiator. Should have amazing bandwidth.

 

[Ubbe]

In Yagi antenna design larger diameter elements increase the bandwidth.

/sarcasm moden
Use a 55 gallon drum as radiator. Should have amazing bandwidth.

The Omni-X use two wide aluminium pipes as its UHF dipole to cover a wide frequency range. The problem are that the wider the element the more you are drifting away from 75 ohm. I can't remember what modeling it in Eznec said about the impedance at different frequencies but it became a problem.

/Ubbe