How do I know if my coax connectors are good?

In installed a couple of compression style PL-259 connectors to the ends of RG8/U FOAM coax. It's about 16 feet in length. Here is an example of one of the connectors:

How do I know if the coax is "good"? The multi-meter test for shorts doesn't seem like it's a complete test. It lets you know for sure if there's a problem when there's a short. But it doesn't necessarily tell you everything is good when there's an open.

I used my NanoVNA to perform a sweep on the cable with the 50 ohm load at the end. The VSWR and RL tends to go up and down from 1MHz to 500MHz. Is this normal?

Other than those sweeps, is there any kind of "red light; green light" test to make sure a piece of coax and the connectors are good?

• Try your 50-ohm load at the end of the shortest known-good cable you have (or no cable, if possible). That's a pretty bad result, but I think the problem is either the load or the lack of calibration of the VNA, rather than the cable. – hobbs - KC2G Aug 4 '20 at 20:02
• I calibrated both on the VNA itself and then in the VNASaver calibration wizard. – Paul Aug 4 '20 at 20:35
• @Paul how did you calibrate the VNA? Because this really looks like you've built a resonator more than a cable, and I'd agree with hobbs, the most likely explanation is that you've either miscalibrated your measurement equipment, or forgot to actually connect your cable under test. – Marcus Müller Aug 4 '20 at 20:56
• What type of connector does your VNA have? Are you using any adapters, such as N to "UHF" (PL-259, SO-239) ? A good photo showing us how you are connecting the coax to your VNA might tell us a lot! Also, you are correct that a DC ohmmeter is worthless for measuring the RF impedance. – Mike Waters Aug 4 '20 at 21:22
• Hi Mike, It's a NanoVNA H4 from GigaParts gigaparts.com/… It has SMA female connections. It comes with an Open, Short, and Load for calibration. I'm using an SO-259 to SMA adapter on both ends of the coax. One to connect it to the VNA, and the other to connect it to the 50 ohm load i calibrated with. Other coax connections look a little better. I can post an example if needed. – Paul Aug 5 '20 at 1:31

So, what your measurement indicates is two things:

1. yep, the thing is somehow connected. The deep dips in the return loss are points when the characteristic impedance of the connector+cable+connector+termination look like the source impedance of the VNA
2. yep, it's a terrible non-constant complex impedance seen from the perspective of the VNA that's everything but 50 Ω. A "good" cable starts at a return loss of -20 dB, but typically, general-purpose pre-configured coax cabling has more like -30 dB to -50 dB of loss. Your connector reflects power, and pretty strongly so! That's no surprise: In the 1930s, it was designed as a cheap connector for frequencies below 100 MHz, mostly for low power systems. It doesn't have a constant (or even defined) impedance and thus no general guarantees about its performance can be made. The tolerances of the connectors, and the lack of mechanical design specification, in fact force the connector to be bad: there's no way to produce a connector that has constant characteristic impedance over a significant range of frequencies with arbitrary counterparts.

So, sorry, I can't tell you based on these measurement if that cable is "as good as it can be". I can tell you, however, that you really can't rely on PL-259 for UHF (even if it's called "UHF connector"; UHF meant > 30 MHz back when it was invented). You might find one PL-259 jack that works beautifully up to 400 MHz with a given PL-259 connector, and another one that looks like your measurements or even worse.

Really, if you can: replace PL-259 with BNC or Type-N or SMA.

• GREAT answer, but it would help if we knew more about his VNA. His measurements near 1 MHz might rule out the connector. – Mike Waters Aug 4 '20 at 21:27

Per Hobbs-KC2G's comment, you should now try and vary your load resistance, as your load might not equal the impedance of your coax. The procedure below will tell you if that is the case or not.

One way of doing this to to solder a small non-inductive 50 ohm resistor directly to the far end of the coax. Cut the leads just short enough to solder them to the shield and center conductor, or their inductance will affect the reading at higher frequencies. If you have any on hand, I would use some older 1/4 or 1/2 watt resistors with a solid cylinder of carbon inside. Measure them. If necessary, parallel another resistor with it to get exactly 50 ohms.

If you still get these results, then change the resistance value using this same method until the VSWR is nearly flat.

If all you have on hand are smaller resistors that have an internal, spiral cut carbon layer, try (for example) paralleling four 200 ohm resistors using this same method. Start at 1 MHz and go up. You should see a nearly flat VSWR that gradually rises with frequency, due to the resistor's inductance, even with a PL-259 connector at the VNA.

This is assuming that the coax is exactly 50 ohms! It might not be. The goal here is to adjust these resistors so it matches the impedance of the coax. (Hopefully, your VNA is close to that. ;-)

Per the OP's comment, another way is to open the far end of the coax:

I did a Lambda/8 test with the coax. It’s about quarter wave long at 12MHz and appears as a short on the Smith chart when the other end is open. Sliding down to 6MHz and the reactance is around -61j ohms. On good coax the same test shows -50j.

• The coax is Saxton 8283 RG 8/U FOAM. I haven't had much luck finding info on it, but I can only assume it's 50 ohm coax. But I guess if it was 75 ohm that would explain the weird behavior. I tried putting fresh connectors on there. Even without solder, just to eliminate the chance of maybe heat doing something. No matter what I tried, nothing changed. Is there a way to I can test this coax while eliminating connectors from the equation? – Paul Aug 5 '20 at 13:37
• @Paul I thought that I explained that in my answer. No connector at the far end, just resistors. Just use the existing connector to connect to your VNA. It won't matter at HF. – Mike Waters Aug 5 '20 at 14:47
• Hey Mike- I did a Lambda/8 test with the coax. It’s about quarter wave long at 12MHz and appears as a short on the smith chart when the other end is open. Sliding down to 6MHz and the reactance is around -61j ohms. On good coax the same test shows -50j. Something must be wrong with the coax? It’s old and had been up in an attic for years. – Paul Aug 6 '20 at 5:39
• According to the nanovna group, the magnitude of the Lambda/8 reactance value is a close approximation of the cable's wave impedance. – Paul Aug 6 '20 at 15:17
• @mike $Z_{sc}=jZ_0{tan}(\beta x)$ where tan(45)=1 so $Z_{sc\lambda/4}=jZ_0$. It works best if the coax loss is small. Better is to find the frequency where $Z_{OC}=-Z_{SC}$, instead of just using a quarter wave. This averages out some other effects like capacitance of the open end slightly better. – tomnexus Aug 6 '20 at 16:14