2
\$\begingroup\$

Are there tools for checking (or that could be adapted to check) if BNC or SMA coaxial cables are in spec for things like proper shielding, impedance (across their bandwidth), etc.? I've found lots of very cheap (<$50) tools for continuity testing, and a couple very expensive (>$20k) analyzers that I think are way overkill, out of budget, and mostly geared towards consumer cables (cat5, etc.). The closest I've seen is this: https://www.tessco.com/product/cable-antenna-analyzer-1-mhz-4-5-ghz-597938 . Would that tell me what I need to know, and would it work with a connector adapter (i.e. to use with BNC or SMA)? It's still a bit pricey ($3.6k) for the intended purpose, but it is within reason. Is there another keyword/type of tool I should look for?

Context: I work in experimental physics where we often do stuff that would probably be better handled by a proper electrical engineer (kind of have to be a jack-of=all-trades). I know enough to get by, but I'm sure I do stuff that would make an RF engineer cringe all the time. I work with signal frequencies anywhere from DC - 5 GHz depending on application, and obviously run into issues more often with the higher frequencies. One thing I've noted is sometimes we run into hard to diagnose problems that ended up being the fault of a BNC or SMA cable (i.e. coaxial cable with BNC or SMA connectors), or a BNC T adapter etc. Sometimes this seems to be because of a cable not really being up to spec with shielding, impedance or bandwidth or something, so they are problems that are a little harder to spot than a simple continuity check. We end up marking cables sometimes ("don't use this one for X", etc.). It'd be nice to be able to just test everything we have in the lab, and dispose of or label the stuff that fails, so we don't waste time trying down trivial issues like this.

\$\endgroup\$
9
  • 1
    \$\begingroup\$ You can can test your cables, but you need a Vector Network Analyzer (VNA) to do it right, along with the expertise to use the tool. Alternately, you can use a cable vendor who will guarantee their cable specs, though those cables will cost more then a generic off the shelf cable. \$\endgroup\$
    – SteveSh
    Commented Nov 3, 2021 at 16:54
  • \$\begingroup\$ Do you know for sure why your cables aren't working in your application? Do you know for sure that they are failing their impedance spec, or shielding requirement? Or are you just venturing an educated guess? \$\endgroup\$
    – SteveSh
    Commented Nov 3, 2021 at 16:56
  • \$\begingroup\$ 'BNC' and 'SMA' only describe the connectors. There are many, many, different types of co-ax cables which can be used with either of those connectors, all with different specs. These cables are typically labeled with their type (RG-58, etc, unless they're really tiny cables) and that's what will tell you what sort of performance you should expect from the cable assembly, not looking at the connectors. \$\endgroup\$
    – brhans
    Commented Nov 3, 2021 at 16:56
  • \$\begingroup\$ And tell us a bit about what your driving requirements are for your cables. Do they need to carry low level analog signals (like from a detector)? What is their loss requirement? Do multiple cables have to be length/delay/phase matched? \$\endgroup\$
    – SteveSh
    Commented Nov 3, 2021 at 16:58
  • \$\begingroup\$ A NanoVNA might be a useful alternative to a full on professional vector network analyser. It won't be calibrated like a pro machine, but it will certainly be able to show you what's wrong with your cables. It has SMA connectors, but you can use adapters. A NanoVNA only costs a couple of hundred bucks. \$\endgroup\$
    – JRE
    Commented Nov 3, 2021 at 17:07

2 Answers 2

1
\$\begingroup\$

if BNC or SMA coaxial cables are in spec

a couple very expensive (>$20k) analyzers

But that's the point here: the specs you're referring to are actually not trivial to measure: you're looking for losses and velocity factors over a large range of frequencies.

In the end, what you need to measure is the complex impedance of your cable using calibrated connectors on a measurement device against the internal calibrated standard of that device. That's the job of a (vector) network analyzer. 20 k€ isn't an "expensive" VNA, but you don't seem to need to go very high in frequency.

Sometimes this seems to be because of a cable not really being up to spec with shielding, impedance or bandwidth or something, so they are problems that are a little harder to spot than a simple continuity check. We end up marking cables sometimes ("don't use this one for X", etc.). It'd be nice to be able to just test everything we have in the lab, and dispose of or label the stuff that fails, so we don't waste time trying down trivial issues like this.

I think the winning move here is not to buy better measurement equipment – because that would still not tell you what is relevant – but understanding exactly what the cable's job is and hence, what requirements it needs to fulfill; for example, to transport a highly accurate reference clock, you don't need a cable that has a very constant attenuation from 10 to 10000 MHz - you actually don't care about the amplitude, as long as you're not loosing too much signal power and thus get into SNR trouble (and that typically takes a while). For other applications, a smooth frequency response is incredibly important.

So, since you're a physicist working with physicists, this is actually easier to explain than to a group of motivated engineering students: every cable carries a specific signal class, and these call for different properties. It's often sufficient to just use the right cable, and they are typically marked with the kind of cable they are. For example, an RG-58 cable with SMA connectors might be nice, sturdy, cheap, abundant and reliable for 10 MHz reference clocks, but the type itself means that it can't really be used for a 1 GHz wide measurement signal centered around 3 GHz, or to carry the 6 GHz excitation for an acousto-optical modulator in a different room.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ also, buying a VNA just to measure a few cables: probably just throwing out all cables, and getting exactly the cables you need would be cheaper. You'd still need to know what you need! Also, chances are, in a modern physics department, you just need to ask your colleagues in a couple other labs until one has such a thing lying around. \$\endgroup\$ Commented Nov 3, 2021 at 17:32
  • \$\begingroup\$ All good points. I didn't mean to imply that $20k is expensive for what it does, just that it is expensive relative to what I need. As you mentioned, at those prices it would be cheaper to replace every cable with a more expensive one from a trusted supplier. But that still doesn't really resolve my question. Obviously there is a need for better education on my part, and you brought up a few things that make obvious sense (but I hadn't really though of before), like that a cable that works for carrying single freq. won't neceessarilly work for multiple frequencies. \$\endgroup\$
    – argentum2f
    Commented Nov 3, 2021 at 18:00
  • \$\begingroup\$ it's less that it doesn't "work badly" for multiple frequency, it's that it might have a different phase and attenuation across your operational frequency range, which can have negative effect on what you're doing - depending on what you're doing. I think if you have an experiment / setup that you know is "picky" on cables, it would be worth asking a question here describing what you're doing (in some detail, even!) and asking about what properties are important for a cable in that application. Might be easier than you think (but would still leave what can be discussed in comments)! \$\endgroup\$ Commented Nov 3, 2021 at 18:08
  • \$\begingroup\$ Yeah, I understand. I think about frequency and phase response all the time in other things, I guess it just hasn't occurred to think about cables that way, beyond the "is the rate for 1GHz?" type of questions. \$\endgroup\$
    – argentum2f
    Commented Nov 3, 2021 at 18:34
1
\$\begingroup\$

Cable testing is best done with a vector network analyzer (VNA) that can do S11 (SWR and/or impedance at the feed-point) and S21 (gain between the opposite ends of the cable under test). If you work at a university, go to the electronics department and they'll probably have a VNA buried in a lab.

Patch cable testing is rather tedious since the setup must free of problems. You need proper adapters to go between the VNA and cable under test and the adapters must be of good quality. It may take many hours of learning how to do this properly.

You'll find that most cables with BNC connectors are junk because the ground connection (spring fingers) isn't designed properly on the connector. Even BNC patch cables supplied by HP (Keysight, Agilent, or whoever they are now) are junk. We've found that Pomona BNC patch cables (current and within the past 20 years) use excellent BNC connectors and have the least amount of issues at higher frequencies because the ground spring fingers are designed properly. Pay the exorbitant costs to get a proper BNC cable. At lower frequencies (below 10MHz), you can get by with BNC cables in a lab environment.

SMA connectors are superior to BNC connectors, especially above 100MHz. For an SMA to work properly, you need to tighten the nut with a wrench. Finger tightening will produce poor results with SMA and 3.5mm connectors. If you are in the GHZ range, use 3.5mm connectors. They look similar to SMA and will mate between series, but have better high frequency specifications. You also need a center-pin connector gauge if you don't want to damage your 3.5mm connectors with an out of specification connector. The cheapo SMA and 3.5mm connectors can have an out of spec center pin.

Another thing to worry about is cable. Testing of various cables in the 1GHZ to 10GHz range shows a lot a variability between cable manufacturers and types.

\$\endgroup\$
2
  • \$\begingroup\$ IIRC the really high end agilent scope (multiple GHz bandwidth) that a RF guy had at work came with BNC connectors, but also came with special BNC to SMA adapters that had a special locking mechanism, not just the sprun bayonet of a regular BNC. \$\endgroup\$ Commented Nov 4, 2021 at 1:53
  • \$\begingroup\$ @PeterGreen The 54609-61609 BNC patch cable that comes with the 8GHz bandwidth scope from Keysight has fancy BNC connectors on the patch cable that are total garbage due to the poorly formed spring fingers on the ground connection (very little frictional force required to mate). When we tested it, it was no better than $4 patch cables you get on e-bay. HP's expensive connectors for multi-GHz use we tested are great. \$\endgroup\$
    – qrk
    Commented Nov 4, 2021 at 2:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.