2

I am using the Siglent SSA3032A-R spectrum analyser which has a Vector Network Analyser (VNA).

Before measuring the impedance of an antenna under test so that I can match this antenna, I have to calibrate the VNA. I perform a 1-port calibration, doing an Open, Short and Load calibration.

The method for calibrating is as follows:

  • Open, connect the 50Ohm cable, soldered to the PCB to the VNA and disconnect the antenna from the 50Ohm cable.

Open calibration

  • Short, using a 0Ohm resistor, connect the antenna feed line to ground.

Short calibration

  • Load, using a 50Ohm resistor, solder the 50Ohm load between the antenna feed line and ground.

Load calibration

When I perform the calibrations, the open, short and load is plotted on the smith chart in the expected location (open plotted on the right, short plotted on the left, load potted in the centre).

I then remove the 50Ohm load and connect the antenna to the antenna feed line using a 0Ohm resistor (as shown below). Connecting this to the VNA results in a negative resistance reading off of the smith chart plot. A negative reactive component would indicate a capacitive load but a negative resistive component should not be possible.

DUT testing

Here is the smith chart plot seen when the antenna under test is connected to the VNA after calibration.

Smith chart plot seen on VNA after connecting antenna under test

Can anyone advise what I might need to do differently?

I have been following the 'How to do VNA calibration on PCB' https://www.baseapp.com/iot/antenna-tuning-for-beginners/

Image of test setup below. Test setup

MRB
  • 382
  • 3
  • 13
  • The open and short must be at the same place, and define your 'reference plane'. It's simplest if this is where you connect the device under test. You've used the terms '50 ohm cable' and 'antenna feed line'. Are these synonyms, or are they the same thing? Post a picture and a schematic of your measurement setup. The VNA is only as accurate as your cals, your open doesn't sound very 'open', and I've found a 50 ohm resistor doesn't often make a good load, two 100s in parallel tend to be better, depnding on frequency. – Neil_UK Feb 23 '22 at 14:15
  • I perform the open and short at the same place. I say antenna feed line, referring to the cable inside the coaxial cable. I said 50Ohm cable to confirm I was using a 50Ohm cable and the cable was not the source of the issue. We are actually using two 100Ohm resistors in parallel as these were readily available. – MRB Feb 23 '22 at 14:33
  • Could you add a plot of the smith chart to see the location of the antenna on it? – Big6 Feb 23 '22 at 14:50
  • @Big6 this has been added. The negative resistive component I cannot make sense of. I read that a negative resistive component suggests incorrect calibration, however I have tried calibration countless times over several days, testing the PCB is soldered correctly (i.e. there is a short or open when performing those calibraitons) – MRB Feb 23 '22 at 14:57
  • I wouldn't get too caught up in the negative resistance part, for sure it is an artifact of calibration, especially in the absence of quality calibration standards. Now, if you were to test your calibration against some resistors, say 10, 22, 33, 68, 82, etc, what do you see on the Smith chart? – Big6 Feb 23 '22 at 15:13
  • Calibrating OSL, results are on the "same" line. Probably default "amplitude" of VNA. Use a quarter line (shorted or open) to verify. Or a 1/8 Lambda. Or a shorted "sliding" line. – Antonio51 Feb 23 '22 at 15:42
  • @Big6 If I test the calibration against just resistors, they measure correctly on the VNA. 100Ohms measures as 100.45+j2.88Ohms and 50Ohms measures as 49.75+j0.762Ohms. So if the VNA/ calibration can measure resistors reasonably well, why would the resistive component of the antenna not measure correctly? – MRB Feb 23 '22 at 15:49
  • 2
    Your open standard actually has some capacitance, and your short standard has some inductance. Since you don't know what these are, you (or the instrument's software) assumed they are 0. The error caused by this assumption can make it appear as if your DUT has negative resistance. Your options are, don't worry about it, do the measurement without correction, or obtain a fancier VNA with well characterized cal standards. – The Photon Feb 23 '22 at 15:51
  • @ThePhoton okay so if I negate the capacitance and inductance in my open and short calibrations for now, and I read a negative resistive component when measuring the impedance of the antenna (e.g. -5-j78Ohms) how can I obtain the antenna's impedance from this? Do I simply ignore the -ve resistive component and read it as if it were +ve? – MRB Feb 23 '22 at 15:55
  • @MRB, one option is design a preliminary matching circuit with the data you have (maybe project your measurement to the nearest point on the Smith chart with R=0), and re-measure with the matching circuit, then tune up the matching circuit based on those measurements. – The Photon Feb 23 '22 at 16:56
  • Also, now that I think about it, it's likely the problem with your standards is that they have some loss (radiation or imperfect resistance values). This would make the instrument think your feedline has more loss than it does, and cause it to correct your DUT measurement by increasing \$|\Gamma|\$, leading to the kind of result you saw. But again the only real solution for that is to get fancier standards and an instrument that can include their nonidealities in the correction process. – The Photon Feb 23 '22 at 16:58
  • How long is the grounding connection between the coaxial outer, and the ground connection for the DUT and the cal standards? Let's have a photo of your PCB, the one I asked you for 3 hours ago. At 900 MHz, every little detail counts. – Neil_UK Feb 23 '22 at 17:42
  • Have you tried with a shorted line lambda/4 ? – Antonio51 Feb 23 '22 at 18:05
  • > ... I read that a negative resistive component suggests incorrect calibration ... < You can have something like this when you have an "active" DUT (some RF "amplifiers"). – Antonio51 Feb 23 '22 at 18:08
  • @Neil_UK the coaxial cable connected to the PCB and DUT is 14cm long. – MRB Feb 24 '22 at 12:29
  • What is this "board" ? And how is connected the antenna ? – Antonio51 Feb 24 '22 at 12:38
  • The PCB is a simple test PCB that follows the manufacturer's recommended layout. The PCB only has the antenna and pads for a pi network so that the antenna can be disconnected from the coaxial cable during calibration and matching components added when the impedance of the antenna was found- only I cannot seem to obtain a correct impedance measurement for the antenna. – MRB Feb 24 '22 at 12:41
  • Ok. So you made calibration at the "socket" of the antenna? Or am I wrong? – Antonio51 Feb 24 '22 at 12:57
  • Is your antenna receiving something? Another nearest active system at the antenna frequency? You should use some absorbers around the antenna. – Antonio51 Feb 24 '22 at 13:03
  • I performed the calibration with the antenna disconnected so to remove any impedance from the cable or PCB so when the calibration is complete and the antenna is connected and impedance measured only the impedance of the antenna itself is measured. – MRB Feb 24 '22 at 13:15
  • @Antonio51 no the antenna is not receiving anything in this test. I am simply trying to measure its impedance so that I may match this antenna. I will then connect the antenna to an RF source and use a spectrum analyser to measure its gain and radiation pattern to compare against other antennas. – MRB Feb 24 '22 at 13:43
  • we don't want a photo of your VNA with a tiny PCB, we want a photo of the PCB itself showing in detail how the coax terminates (makes ground connection) to the board, and how the two boards connect. Having a good repeatable transmission lines, 50 ohms if possible, is vital. Having short connections is vital. Having clean connections at your cal pieces and DUT is vital. – Neil_UK Feb 24 '22 at 13:49
  • @Neil_UK I cannot get a good quality photo of the PCB as it is quite small. We made sure to obtain a 50Ohm coaxial cable. the coaxial cable is 14cm, we could make this smaller but it is doubtful this is the sole/ main cause. As for the ground connection, the ground sheath from the coaxial cable has been soldered directly to the PCB, ensuring it does not make contact with the core cable of the coaxial cable. Before each calibration, the PCB is tested to ensure there are no shorts/ opens and that the solder joints are secure. – MRB Feb 24 '22 at 14:10
  • @MRB the length of the cable, 14 cm or not, is totally irrelevant. I am disturbed that you think it's worth mentioning that length. What I want to know is the detail of how the cable connects to the PCB, especially the length of the ground connection from the end of the cable to the ground connection on the board, DUT, and cal pieces. That is the detail that needs addressing, and that is why I want a high resolution photograph of the PCB. – Neil_UK Feb 24 '22 at 19:30
  • @MRB Are you sure that the antenna does not receive something (GSM, or other in the neighborhood)? When you read a negative resistance on VNA, it is because the antenna is really a "light" "generator". – Antonio51 Feb 24 '22 at 20:43

0 Answers0