Alternative to RG58 for home-built RF VNA (network analyzer) benchtop test cable

[wb0gaz]

Posting this here as I think this may be close enough to metrology/calibration that this would be the starting point; if not, please let me know?

Summary Question: I am looking for alternative to RG58 (same general diameter, but not that type specifically) for use in building some home-made RF patch cables usable to 6 GHz on a benchtop VNA.

(background/details)

I need build some short (typically 0.5-1.0M long) coaxial patch cables for benchtop use with a vector network analyzer for RF component/system testing up to 6 GHz. I need the cables to be N-male at one end, SMA-female at other end.

In this application, the cables will be subject to repositioning during use (i.e., between calibration and measurement phases, but not during actual measurements), so mechanical phase stability is a concern. Thermal phase stability is not a concern (room temperature only).

I prefer to use RG58 size (but not specifically RG58) cables as this size seems reasonable mechanically. RG58 itself is probably not a good choice (at least, poor RF shielding if nothing else.)

I don't wish to source commercially offered VNA test cables due to cost and I'd like to learn from the home-build experience.

Cable type options I'm aware of, with observations:

RG58 - OK cost, flexible (good), single braid shielding at best (bad)
RG142 - OK cost, double shielded (good), bad flexibility (solid center conductor, PTFE dielectric)
RG400 - OK cost, double shielded (good), bad flexibility (stranded center conductor, PTFE dielectric)
RG223 - OK cost, double shielded (good), bad flexibility (solid center conductor, solid PE dielectric)
RG223/U-02 - Very high cost (bad), double shielded (good), probably good flexibility (stranded center conductor, solid PE dielectric)
LMR195 - OK cost, double shielded (good), bad flexibility (solid center conductor, foam PE dielectric probably susceptible to phase shift during movement?)
LMR195 Ultraflex - OK cost, double shielded (good), probably good flexibility (stranded center conductor, foam PE dielectric probably susceptible to phase shift during movement?)

I could shift to LMR240 Ultraflex (can switch connector types) but remain concerned about foam PE dielectric vs. phase stability

Thanks for any advice, and happy to provide any needed problem clarifying information.

[shabaz]

Hi,

Regarding the comment:

"cables will be subject to repositioning during use (i.e., between calibration and measurement phases, but not during actual measurements)"

Why is this, if you don't mind me asking?

I'm worried it might invite an additional chance of error by doing that, and normally, it should be possible to set the cables appropriately, make the calibration, and then move the calibration pieces out of the way to connect to the DUT. If there's not enough space, the entire VNA and cable could be moved together so that none of that needs to change between calibration and measurement.

Can I ask why you're not considering semi-rigid cable? It's possibly easier to self-assemble, and it will retain shape between calibration and measurements. Anyway, it's an option to rule in or out. There are great flexible cables, too; I just don't have experience with self-assembling other than just a few types of coax, nor have I used the ones you list specifically with a VNA to have formed an opinion for that use case, unfortunately. I'm sure others will know better in that area.

Personally, if I can, I like using RG-402 semi-rigid coax with a VNA; it is for sure thinner than RG-58, but not so thin as to be awkward to work with relative to scenarios that may have used RG-58.

[wb0gaz]

Hello Shabaz,

From your reply, you immediately and correctly identified missing information in my original post, so thank you for opportunity to provide!

I anticipate moving the test cables (probably just small amount) because for a 2-port connection, it is not possible (I think) to remove the calibration components and insert the device under test without some movement in the test cables. The movement is not large, but I think it would prevent a fully rigid test cable configuration.

In a 1-port connection, I agree with your idea that the test port end point (where it would attach to DUT) is fixed. In that case, RG402 (I have some UT250 which I think is similar except for solid copper jacket instead of braided material) would be good material.

In some 1-port test cases, I just attach DUT to the existing test port of the VNA (usually through adapter), so after removing calibration part, there is no cable movement because there is no cable. This is not always practical, but eliminates in many cases any concern about test port cable integrity, because there is none.

Anyway, thank you again for helping me further understand the use cases!

[tooki]

Cable type options I'm aware of, with observations:

RG58 - OK cost, flexible (good), single braid shielding at best (bad)
RG142 - OK cost, double shielded (good), bad flexibility (solid center conductor, PTFE dielectric)
RG400 - OK cost, double shielded (good), bad flexibility (stranded center conductor, PTFE dielectric)
RG223 - OK cost, double shielded (good), bad flexibility (solid center conductor, solid PE dielectric)
RG223/U-02 - Very high cost (bad), double shielded (good), probably good flexibility (stranded center conductor, solid PE dielectric)
LMR195 - OK cost, double shielded (good), bad flexibility (solid center conductor, foam PE dielectric probably susceptible to phase shift during movement?)
LMR195 Ultraflex - OK cost, double shielded (good), probably good flexibility (stranded center conductor, foam PE dielectric probably susceptible to phase shift during movement?)

Are your “observations” from hands-on experience or just assumptions? I ask because RG223 isn’t actually that stiff. At least not the good stuff I have at work from Huber+Suhner. Being solid it does have some memory, but many coaxial cables with stranded conductors can be stiffer. The dielectric makes a difference, but the insulation makes a HUGE difference. In addition to the H+S RG223 at work, I also have H+S RG174 and RG174 from Tasker. The difference in flexibility between the two brands of RG174 is enormous. The Tasker PVC is quite rigid and makes the cable very stiff. The H+S PVC is supple and soft, and thus so is the cable.

I also have H+S K_02252_D, which is RG316D (double-shielded RG316). Like other RG316, it’s somewhat stiff. And if you strip off a good length of the FEP jacket, you notice that it’s the FEP jacket, not the dielectric or shields, which makes it so stiff — the stripped part is super flexible.

So while I certainly wouldn’t use a solid-conductor coax for things that are moved frequently (and definitely not in tight bend radii!), many are probably not as stiff as you think.

[wb0gaz]

Hello (again!) Tooki,

Thank you for the comments!

I have no hands-on experience with RG223; I was inferring the potential characteristics from hands-on comparison of LMR400 (solid center conductor) vs LMR400 Ultraflex (stranded center conductor) where there is quite a bit of difference. Of course that's completely different material, but without physical sample/experience with RG223, I had classified it as not likely to be practical. For that reason, your hands-on experience provides quite valuable input, and consideration of the different contributors to flexibility is invaluable, as I had not considered that (I had tended to assume that center conductor and dielectric were primary contributors, and I now believe my assumption was in error or at least quite incomplete!)

Fortunately I found a section of HS RG223 is not too expensive, so I have placed order online to try. In particular, that will satisfy my curiosity as to it's susceptibility to phase/delay shifts during movement, which I can roughly observe by calibrating a length, leaving the calibration short in place, then watching phase or smith chart while sweeping frequency on VNA.

So, thanks again for your input!

[Overspeed]

Hello

Be aware of forces on connectors generated by stiff RF cable , more your RF cable is stiff better your cable / connector shall be .

Regards
OS

[tooki]

I have no hands-on experience with RG223; I was inferring the potential characteristics from hands-on comparison of LMR400 (solid center conductor) vs LMR400 Ultraflex (stranded center conductor) where there is quite a bit of difference. Of course that's completely different material, but without physical sample/experience with RG223, I had classified it as not likely to be practical. For that reason, your hands-on experience provides quite valuable input, and consideration of the different contributors to flexibility is invaluable, as I had not considered that (I had tended to assume that center conductor and dielectric were primary contributors, and I now believe my assumption was in error or at least quite incomplete!)

Fortunately I found a section of HS RG223 is not too expensive, so I have placed order online to try. In particular, that will satisfy my curiosity as to it's susceptibility to phase/delay shifts during movement, which I can roughly observe by calibrating a length, leaving the calibration short in place, then watching phase or smith chart while sweeping frequency on VNA.

Getting samples is definitely the way to go, where possible!

I have never used either kind of LMR400, but in addition to being a much larger cable family with a much, much thicker conductor than RG223, the regular and ultraflex versions of LMR400 appear to differ nor only in the solid vs. stranded conductor, but also in the jacket material. The tough PE typically used for jackets is quite stiff (compared to the much softer, waxier PE used for dielectrics), so I’m not at all surprised that the regular LMR400 is much stiffer than the TPE (thermoplastic elastomer) jacketed ultraflex version.

When I found the H+S RG223 in storage at work — bearing in mind I wasn’t familiar with that type at the time — was actually surprised to discover it has a solid conductor, because it didn’t feel particularly rigid at all. More like… RG58 with a bit more memory.

[wb0gaz]

OK good on all - will re-surface when I can provide an initial result with RG223 sample.

As for connector vs. cable (torque), I experienced this with HP APC-7 test port cable (there are two kinds, I was trying the thicker one with quite stiff cable, it has 2291 in the part number); movement in the cable would often slightly reverse tightness on APC-7 connector (which has fairly low torque specification), impairing measurement results.

Thanks again for the most helpful and informative replies!

[David Hess]

I like RG400 (double shield, stranded, Teflon, 0.2" OD), however doubled shielded RG316 (double shield, stranded, Teflon, 0.114" OD) exists.

[Smoky]

Arrow has 5 CalTest CT4443-100 cables on sale that are N to SMA for $13 each regular price ~$65:

https://www.arrow.com/en/products/ct4443-100/cal-test-electronics

They use cable type RG-142B/U (Mil-C 17/60-RG142 is printed on the cable) rated to 6GHz.

I bought two last month when they were $10 each. Maybe the price will come back down again

Maybe a little stiff for your needs but they come in a small roll about 5" in diameter.

Datasheet attached:

*Oops, you're looking for the SMA end to be female. Disregard these cables then.

[wb0gaz]

Thanks, Smoky for the pointer to the Caltest cables on Arrow - correct - SMA female is needed, but I appreciate the pointer and will keep that in mind.

As for my original conjecture regarding flexibility - RG223 sample (Suhner) did arrive, and it is very nicely flexible, so my original claim about bad flexibility for RG223 was indeed wrong. The sample cable does not appear to be labeled _U02.

Thank you again Tooki!

What I have learned is that flexibility of coaxial cable has also dependency on outer jacket construction/material (along with center conductor, dielectric, and shield).

In this same time I received sample RG400 type (thanks, David H for your comment) - it is less flexible, however, once connectors arrive I will be able make test cables to compare phase stability where the only variations will be the cable material and repeatability of my (home workbench) construction effort.

I am now waiting arrival of connectors.

[Smoky]

Here's another cable on sale on Arrow that testing-folks might want to look at:

https://www.arrow.com/en/products/095-902-531m100/amphenol-rf

It's $3.16 with a regular price of ~$40 and there are 18 in stock.

It's a Type N to SMA male cable using RG316. It's 1 meter long (~39") and tested to 6GHz.

You could cut off the SMA male connector and put on a female and it would still be over 36" long after you're done

All of the RG316 used by Amphenol is brand name Harbour or Thermax too.

The cost of just the N connector used on this cable is like $15.

I bought 4 today just to have them