EEVblog Electronics Community Forum
Electronics => RF, Microwave, Ham Radio => Topic started by: tchicago on April 20, 2024, 05:46:44 am
-
Could someone please recommend the "type" and/or source of the 120 or 125 Ohm coaxial cable that is not too thick?
I know of RG63 "type", but it is a little too thick for my purpose. I need it to experiment with matching a folded dipole to 50 Ohm coax using the "stepped impedance matching" method. So obviously, I only need small lengths of this cable.
-
Why not use twisted pair? It's also balanced which is nice for the antenna, and has the right impedance (or nearby). Oh, maybe it's a balun and you're using the shield mode for a 1/4 wave transformer? Dunno.
Tim
-
Yeah, 1/4 wave transformer aka "stepped impedance matching". The piece of 125 Ohm coax would fit neatly inside the pipe that forms the folded dipole, if it wasn't that thick.
-
RG-63 is the only standard coax I'm aware of with Z0>100ohms. High Z0 requires a large OD/ID ratio, so a small cable diameter is asking for a lot...
What frequency are you working at? If <1GHz I would suggest a lumped element phase shifter. If much higher, maybe a PCB transmission line would work?
-
Well, fitting coax inside a pipe isn't going to help with shield-mode balun action. You're making a tri-ax assembly, where the second dielectric is a mix of air and whatever the coax jacket is (usually PVC). So the lengths will be different, and losses likely high.
But at this point we require a diagram to say anything for sure.
Tim
-
Is 100ohm close enough?
You can make 100ohm coax by paralelling 2 identical pieces of 50ohm coax together, join the shields at each end and use the two center cores as the two conductors.
-
The question posed by tchicago is interesting, but can not be modelled here!
Does anybody use a FEM pkg with a graphical model UI that can do this?
Xnec2c (linux version here) works well at HF, UI is not so fancy, best used by us old guys who started with Hollerith cards! Xnec can not model coax cable as far as I know. Only thin solid conductors.
I have QuickField ( purchased pkg) here with fast UI, which can do a 2D of coax, but it can not do Maxwell, only low freq electromag, and static capacitor etc.
I never had much success with installing FOSS FEM . I tried CFD and some others but they typically had complicated data entry, too difficult to learn.
Edit. Info on using RG63 etc in series and parallel:
http://on5au.be/content/a10/trans/spcoax.html (http://on5au.be/content/a10/trans/spcoax.html)
-
It is called a "Sinclair Antenna" design. A short description is here: http://www.gorum.ca/sinc_ant.html (http://www.gorum.ca/sinc_ant.html) There are tons of commercial implementations of this design, as this design is very simple mechanically. But requires a rare RG63 cable. Which I figured is itself very hard to get in reasonably quantities.
I don't see where the coax being inside a tube will affect anything: 50-ohm coax enters the antenna at zero-potential point, and exits at the place where the adjusted 292 Ohm impedance is needed. The braid of internal coax is connected to one side of active element, and the other side is connected to the internal conductor.
It would be super nice if someone could model this design. (Someone with access to modeling software and skills to use it)
The stepped impedance matching using one 1/4 wave line step is used as described here: http://www.gorum.ca/sinc_ant.html (http://www.gorum.ca/sinc_ant.html)
Z01 = sqrt(ZS*ZL)= sqrt(50*292)=120.8 <- that's the required impedance of 1/4 wave transformer line.
My target frequency is the 915MHz ISM band. So, it is kind of a borderline where I can use some magnetics, but the choice of material and design of the balun will play a very critical role.
Another option could be to make my own improvised coax with air dielectric by hanging a wire with appropriate diameter inside the tube of antenna, using some plastic holders. But it is quite challenging to do properly.
-
can you remove the insulation from the RG63 and then put copper tape on it instead of braid to make it thinner? then seal with a ultra thin heat shrink or use kapton tape.
the braid and insulation take up alot of room.
-
Hopefully twisted pair etc might be an option, otherwise, since it is not required in a very long length, perhaps a thick wire (so that it remains straight) inside a copper tube is feasible? I've never done this, so I'm speculating unfortunately; apologies if this doesn't help.
I used this Matlab (https://www.mathworks.com/help/rf/ref/txlinecoaxial.html) link.
Example where inner wire is 0.8 mm radius, and the outer shield is 6.0 mm radius on the inside (i.e. a tube with 12 mm inner diameter), and air dielectric, results in approx 120 ohm it seems:
>> txline = txlineCoaxial('OuterRadius',6.0e-3,'InnerRadius',0.8e-3,'EpsilonR',1.0006,'SigmaCond',5.8e7);
>> z0_f = getZ0(txline,915e6)
z0_f =
1.2077e+02 - 4.6379e-02i
-
Hmm, twisted pair might not work anyway, because it can't exactly be unshielded inside a metal pipe. The Zo will be lower than expected, including normal mode.
Tim
-
You can use 75 ohm coax but it won’t be 1/4 wave long, it’ll be less. 1/4 wave would give you 19 ohms. Can’t find the online calculator I used to know for series coax matching.. SimSmith would probably tell you.
-
Hmm, twisted pair might not work anyway, because it can't exactly be unshielded inside a metal pipe. The Zo will be lower than expected, including normal mode.
Tim
Oh very true! Since it's a dipole, I assume OP needs a balun as well. One simple way would be a ferrite sleeve current balun at the end of 50ohm coax, then 120Ohm UTP through a plastic pipe.
-
Assuming RG63 is available eg Belden 9857 which has O.D of 0.405 inch,https://catalog.belden.com/techdata/en/9857_techdata.pdf
it might be possible to purchase copper tube to suit.
I just measured some ordinary plumbing tube here, it is 0.500 OD by 0.0413 Wall , measured by caliper at ID of 0.417 inch.
I have used O2-Free High Conductivity (OFHC) copper for induction heating, in the following size: OD of 0.500, wall 0.032 which gives ID of 0.436 inch
like this (selling cut lengths, but don't say minimum order) https://www.atlasbronze.com/C10100-Tubing-s/1963.htm (https://www.atlasbronze.com/C10100-Tubing-s/1963.htm)
As the thin wall is soft, it dents easily which might jam the coax.
To bend the thin wall tube into semicircle, it may be necessary to pack it with sand first.
Plumbing tubes and pipes have standard dimensions, with 3 grades of wall thickness
Engineering ToolBox: "CTS-CopperTube Sizes" "NPS-Nominal Pipe Size"
At 915MHz the 1/4W transformer is about 81mm or 3.2 inch before trim, to stuff into the tube.
-
Oh very true! Since it's a dipole, I assume OP needs a balun as well. One simple way would be a ferrite sleeve current balun at the end of 50ohm coax, then 120Ohm UTP through a plastic pipe.
No, extra balun is not needed in Sinclair design. The coax going through the pipe from the zero potential point to the max potential point IS the balun in this case.
-
Plumbing tubes and pipes have standard dimensions, with 3 grades of wall thickness
Engineering ToolBox: "CTS-CopperTube Sizes" "NPS-Nominal Pipe Size"
My biggest issue is the inability to bend the thicker pipes. It was kind of OK to bend the thinner "refrigeration" copper pipes, but those thicker ones - I have no idea how to do that without special equipment.
-
Next option might be: run regular coax all the way through, then put an LC network at the tip of the arm, of equivalent Zo and electrical length (Fo). Equivalent (calculated) values should be close enough that VSWR is okay, but this does of course get dangerously close to "eh just put a matching network on it" and then you need a VNA (well, some manner of NA) to be sure.
Tim
-
Bending, packed with sand, you might need a oxy-acet #1 tip or oxy propane torch to re- anneal during the 180 degree bend. Need dull red near 700C.
But why not use the thinnest possible tube?
-
Use the HP Appcad freeware and design your own with standard size tubing and some oddball dielectrics or spacers ?
Steve