I have a NanoVNA V2 Plus (50kHz - 3GHz) and decided to test my LoRa antennas:

  1. an unknown antenna delivered with a LoRaWAN gateway (RAK831)
  2. a 868Mhz antenna from Linx Technologies (ANT-868-OC-LG-RPS)

As the range was poor between my gateway and my devices using the first antenna (1), I thought that maybe this was an antenna for US LoRa frequencies (915MHz) so I bought the second antenna (2) that works in the 868Mhz band. The datasheet of (2) contains the following figures:

vswr return loss

However, when I measured the antenna response with the NanoVNA I got very different results:

antenna 2 response

In my results, the VSWR is ~2.7 at 868Mhz and never go under 2.5. It also seems from the smith chart that the antenna has too much inductance. As a comparison, the response from the unknown antenna (1) looks much better:

antenna 1 response

I configured the sweep from 840Mhz to 940Mhz with 0.25Mhz steps. I calibrated the VNA with the included calibration kit. The antenna is connected directly to the NanoVNA SMA connector. Since the antennas use SMA-RP, I used an adapter and compensated for it in the calibration by adding 57ps in the port length extension configuration. This is the first time I use a VNA and mostly followed the Andreas Spiess NanoVNA V2 tutorial.

Could such a large discrepancy between the datasheet and my result come from the lack of precision of my cheap VNA? The fact that the SWR is much better for the unknown antenna (1) makes me think that the Linx antenna (2) is crap. However, it does not match the analysis from the datasheet.

I am misusing the VNA? What could explain this behavior?

Edit: I made some new measurement, this time with a 30cm SS405 cable between the antenna and the NanoVNA.

unknown antenna (1) without touching the coax (antenna end): antenna 1 unknown antenna (1) touching the coax (antenna end): antenna 1 touch Linx antenna (1) without touching the coax (antenna end): antenna 2 Linx antenna (1) touching the coax (antenna end): antenna 2 touch

  • $\begingroup$ Connection problems can also cause such large reflections. Very annoying to trouble-shoot, but with the fast scanning of NanoVNA2 you can try to wiggle the connections to see if anything changes. $\endgroup$ – jpa Jun 8 at 7:05

The NanoVNA isn't exactly precision equipment, and the Linx datasheet specifies a maximum VSWR of 2.3, so it's possible the antenna is within spec (barely).

You could double-check your calibration by attaching a 50 ohm load and running a sweep: you should see a 1:1 VSWR at every frequency. Of course, if you use the same 50 ohm load you used in the calibration, this doesn't rule out a bad calibration reference as the issue.

Note that the figures in the datasheet were measured with the antenna straight up above a ground plane:

enter image description here

Were your measurements taken under similar conditions? If not, the results may not be the same.

The antenna is claimed to be ground-plane independent, but it may still rely on some kind of choking at the feedpoint to isolate the feedline. One thing you can do is put your hand over the feedline or significantly change the geometry of the feedline (for example, twist it into a loop) while running a sweep: if you see this significantly affecting the measurements then it's an indication there are common-mode currents on the feedline, and you are measuring the impedance of not just the antenna but also the feedline. In actual use, things may be different especially if the antenna is mounted on a metal box.

Finally, keep in mind VSWR doesn't need to be 1:1. A higher VSWR doesn't mean worse performance per se: it just means the radio will see an impedance other than 50 ohms. Often, radios specify an acceptable range for VSWR although I could not find one for the RAK831. I would hope the VSWR would at least be below the maximum specified in the antenna's datasheet, but without a calibrated reference it's hard to say if there are problems with your antenna or your measurements.

  • 1
    $\begingroup$ I added some new measurements in the question. Putting my hand on the end of the feedline (touching the metal directly) significantly change the response, so there is probably not enough grounding. How can I improve my setup? $\endgroup$ – DurandA Jun 7 at 15:30
  • 2
    $\begingroup$ @DurandA Try drilling a hole in some copper board for the connector, like in the photo from the datasheet. Or maybe your gateway is in a metal enclosure, and you can just use that. $\endgroup$ – Phil Frost - W8II Jun 7 at 16:28

Most probably the discrepancy is caused by lack of ground: surface or radials. The coax and/or the VNA are no good ground and the impedance of this counterpoise adds up to the total antenna impedance.

  • $\begingroup$ How can it be improved? Should I try to better ground the NanoVNA? I do not use any coax cable, the antenna is directly connected to S11. Should I use a coax cable to increase distance between the antenna and the VNA? $\endgroup$ – DurandA Jun 7 at 13:15
  • 4
    $\begingroup$ The first sentence of the Linx antenna's datasheet is "The OC-LG series antennas deliver excellent performance in a ground plane independent dipole configuration." So, if the problem is the ground plane, the antenna isn't doing a very good job of delivering on its claimed performance. $\endgroup$ – Phil Frost - W8II Jun 7 at 13:20
  • 1
    $\begingroup$ An antenna can be independent of a ground plane, but not from other bits of random conducting objects larger than lambda/10 in the near field. However a seperating ground plane can isolate an antenna from those effects to some degree, likely enough to change the measurements. $\endgroup$ – hotpaw2 Jun 7 at 16:11
  • 1
    $\begingroup$ @Hotpaw: the current in the radiating element can only be obtained from a source and two connections. A dipole does not require ground for operation. The topic starter observed a problem with one of the two connections: the counterpoise or ground. And you are right w.r.t. mutual influence from objects, more or less resonating, in the transmitted EM-field. But what do I say in aworld where pundit PF will explain better in every case. And PF is right, but strives to completeness beyond limits.... $\endgroup$ – F. Sessink Jun 7 at 22:04
  • $\begingroup$ And may be I use words that after transalation have a negative sound. It is not a compliment, or is it? In Dutch, before translation: yes. $\endgroup$ – F. Sessink Jun 7 at 22:13

In addition to Phil's excellent answer, these specifications are in an "ideal" free space. Having your NanoVNA in close proximity to the antenna (even having the antenna lying on a wooden desk) will influence your measurements. Try just moving it around, closer or farther from other objects, you'll see your results changing.

These measurements are not that simple to do "right" (I found out the hard way).

Additionally, I wonder... If these antennae function as a dipole, should they not (ideally) have some kind of current choke at the feed point? If so I would like to see if/how they do this.

  • 1
    $\begingroup$ Thanks for your answer. From the various setup I tried, I still think that there is an issue with the Linx antenna as the other (probably cheaper) antenna always has a much better response. The distributor agreed to send me a new antenna so I will check if my particular unit was defective. Also it came slightly bent but this might just be cosmetic. $\endgroup$ – DurandA Jun 7 at 16:09
  • 2
    $\begingroup$ @DurandA You can't necessarily conclude performance is better because VSWR is better. For example, one way to make an antenna relatively insensitive to its surroundings and to improve its bandwidth is to make it lossy. If you make it nothing but loss you get a dummy load, which has an excellent VSWR over a very wide range of conditions. $\endgroup$ – Phil Frost - W8II Jun 7 at 16:32
  • $\begingroup$ @PhilFrost-W8II Thanks for raising this. The antenna measurement page from Wikipedia also explains this. Couldn't an antenna with just VSWR, inpedance and gain specified (e.g. this Pearson antenna) without figures about the efficiency be a dummy load while still being compliant with the spec? Let me know if you think that I should create a new question for this. $\endgroup$ – DurandA Jun 7 at 17:54
  • 1
    $\begingroup$ @DurandA To have a gain of more than -inf dBi the antenna has to radiate a non-zero amount, and so can't be an ideal dummy load. $\endgroup$ – Phil Frost - W8II Jun 7 at 18:14
  • $\begingroup$ Given these specifications, could it theoretically be the case that this 2dB gain antenna is very directive and have a very poor efficiency? I always interpreted the gain as a measure of directivity for passive antennas. $\endgroup$ – DurandA Jun 7 at 18:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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