Group Actions
Engagement
  • Author Author: shabaz
  • Date Created: Date Created
  • Views 4170 views
  • Likes 9 likes
  • Comments 19 comments
Related
Recommended

Antenna Measurement with the R&S FPC1500 Vector Network Analyzer

shabaz

Introduction

One really useful use-case for a VNA is to characterise antennas. When radio transmitters operate, the signal travels with virtually no loss over a cable (often a coax cable) and then is radiated by an antenna. But some of the energy may be reflected back towards the transmitter, if there is an imperfect match. The VNA operates in a manner where it can simulate a transmitter, and observe how much of the signal is reflected back. This provides an indication of how well the antenna is performing (the assumption is that anything that isn’t getting reflected back is being mostly transmitted; this isn’t always the case, for instance a 50 ohm resistor will absorb it all, but we assume that a reasonable antenna is attached).

 

The VNA goes further into the characterisation by also monitoring the phase difference in the reflection versus the input signal. This provides insight into how to modify the antenna side circuitry to better match the transmitter.

 

I’d recently purchased a low-cost antenna from ebay, and I was curious to see if it really did meet the specifications. This short blog post documents my findings for that antenna, but the steps described here are relevant to any antennas.

image

 

Connecting it up

The antenna came with an SMA connector. As you can see in the photo above, I used an adaptor to deliberately go from there to an N connector plug, and then attached that to an N socket adaption back down to SMA. There is reason to the madness! For more information, see FPC1500 Spectrum Analyzer - Review but in brief, I wanted the cables that connected the antenna to the VNA to be calibrated out of the measurement. To do that, I needed to make the VNA aware of what constituted the cables, and what constituted the antenna. To define this, I need to pick a reference plane and calibrate the VNA up to that plane. The calibration uses three reference impedances and the tool has three connectors for this. The tool looks like a T-shaped piece with a connector on each end and it is known as an OSL mechanical calibrator – OSL refers to Open, Short and Load. It can also be called an OSL T-piece. The FPC1500FPC1500 supports an electronic calibrator that eliminates a lot of effort and risk of inaccuracy, but I only have the OSL mechanical calibrator, so that is what I used.

image

 

It just happens that my OSL tool has N connectors (this is very common) so the cable needs to have an N connector too. The photo above shows the approximate location of the calibration plane.

 

Once the calibration is done, the OSL calibrator tool is removed, and the calibration plane can be attached to the antenna. Since the antenna has an SMA connector, an N-to-SMA adaptor is used. The one in the photo above cannot be removed for the test because it would mess with the calibration plane and the calibration that was just done.

 

Incidentally, a good quality cable is needed. I used a home-assembled cable that consisted of RG-402 semi-rigid coax (so that things don’t move during the measurements) with SMA connectors soldered to each end and ferrite cores strung on the outside in an attempt to isolate the coax shield from RF, so that it doesn’t influence the antenna much. The antenna ought to be mounted as it would be mounted for a real scenario (i.e. if it would be fitted to an enclosure, then it should be fitted to an enclosure for the measurement too).

 

Working with VNA Calibration Tools

As mentioned, the easiest way is to use an electronic calibration toolelectronic calibration tool.

Image source: R&S websiteimage

 

It has a single N plug on one end, and a USB socket on the other. When the USB connector is plugged into the FPC1500, the calibration procedure can be used to electronically switch different reference impedances onto the N connector. Tables of calibration information are also built-in to the tool, so that the FPC1500 is aware of the precise characteristics of the tool’s impedances.

 

Without the electronic calibrator, an OSL tool can be used. I had to send off mine to be measured, and the returned file specified the parameters that I needed to load into the FPC1500.

 

The excellent software from Rohde and Schwarz, called InstrumentView, is used for that purpose. It allows me to enter in the information in a user-friendly manner, and it will convert it to a binary file for uploading (e.g. via USB) to the FPC1500.

 

Here you can see the parameters I entered for my OSL tool:

image

 

After that I clicked on Save to locally generate and save the binary file, and then went into the Instrument->File Transfer menu to transfer the file to the FPC1500.

 

Loading the parameters is just the preparatory part of the calibration procedure. Next, the calibration procedure is executed from the FPC1500 menu. It will prompt on the display to attach the OSL tool at the desired reference plane location in three steps, corresponding to the three impedances that the OSL tool offers.

 

Lots of care is needed during this time, not to disturb any cables, and to double-check that no cables work themselves loose during the procedure.

 

Once the procedure is done, the OSL tool is disconnected and the antenna can be attached to the VNA!

 

Results

The screenshot below shows the voltage standing wave ratio (SWR or VSWR) of the antenna, across the 0-3GHz range. As can be seen, the SWR falls below 2.0:1 at the marker M1 position which is 1.37GHz. This antenna would therefore function from 1.37GHz to 3GHz and beyond. This is consistent with the ebay advert, which states it works from 1.35GHz to 9.5GHz. I could only test to 3GHz of course. A SWR of 2.0:1 represents a loss of about 10% of power, i.e. 10% was reflected back.

 

For a lot of the spectrum the ratio was below 1.5:1. At 1.5:1 SWR, the loss is 4%.

image

 

Another way of seeing the same information is to look at ‘return loss’. It shows the same information, just as a different conversion that shows more detail due to the y-axis being logarithmic. The SWR chart has everything cramped in the useful portion. In the return loss chart below, the marker is set to -9.5dB which is a return loss value of 9.5dB, that is equivalent to 2.0:1 SWR.

image

 

So far the charts have shown scalar information. Now we can go into the vector domain! The swirly diagram here is a Smith Chart. It is discussed in a lot more detail in the FPC1500 review linked to earlier. To cut a long story short, imagine the 0 to 3GHz line in the previous charts, laid out in a different way, which in the case of the screenshot below, happens to begin near the right side of the circular chart, spirals downward and inward, and finally ends up at somewhere else (near the center of the chart in this example).

 

Anything near the center of the chart corresponds to the antenna looking like a 50 ohm resistance at that frequency, and the loss would therefore be very low. Anthing far away from the center corresponds to an impedance that is not close to 50 ohms resistance. The green marker M1 corresponds to the 1.37GHz position on the swirly yellow line, and it is close to the center. The D2 marker is at a delta of 1.63GHz to M1, i.e. it is at 3GHz absolute frequency, and it too is close to the center.

image

 

The Smith Chart view is helpful because it graphically describes what needs to be done to the circuit in order to make the match closer to the center, for any desired portion of the spectrum. This too is discussed in more detail in the FPC1500 review.

 

Summary

It was easy to use the FPC1500 to verify an antenna. The procedure is just a few button-clicks once the calibration procedure is done. The electronic calibration tool would reduce the calibration steps and make that part of the procedure more reliable, but it is perfectly feasible to do it using manual tools too.

 

The VNA provides valuable information on how to tune an antenna using components. The scalar charts provide a good summary of the overall match across the desired frequency range.

Parents

  • I realize this is an old thread but I enjoyed the article.  I am planning an FPC-1500 purchase soon and also have a third party OSL with unknown coefficients.  I am in the U.S.  Can anyone tell me where to I might send it for evaluation so I can get the values needed to enter into InstrumentView?   Thank you!

  • in reply to dc2daylight

    Hi Mark,

     

    Others may have US-located recommendations, but it's a fairly specialist thing and I only found Kirkby Microwave in the UK to do it - costs about £50 GBP, but overseas shipping/taxes may perhaps double that.

    The luxury option however would be the automated OSL tool that's available for the FPC1500 (pricey for hobbyist use though, think about $900 ballpark, but low-cost compared to the traditional manual OSL kits).

    Depending on your requirements, if you're working at low frequencies then you might find the result is not too different if you don't have any coefficients and just do the OSL procedure. As an example, I don't often use the T-shape OSL tool because the N connectors are inconvenient for my needs. I use a home-made SMA OSL tool instead, I don't use it much above 50 MHz or so although it might still give reasonable results beyond that I reckon. The photo of it is below, it says 'Texas Instruments' on it because I used their PCB gerber files, modified slightly (it was a larger TI antenna board, I deleted the antennas and edited part of the file to become the device-under-test ('DUT') portion of it). In the photo I was testing a capacitor, not an antenna (for regular antenna testing with an SMA connector, then a different OSL tool could be manually built, for example there are pictures on the Internet of people hacking SMA connectors to produce this). The gerber files for this board (requires a couple of 0402 sized thick film 100 ohm resistors for the 'Load' portion) are in this zip file.

    image

    image

  • in reply to shabaz

    Very interesting!  Thanks for the quick response.  I may build something like that for the lab.  I had planned to make a couple OSL kits I saw somewhere on the internet with some Nf-SMA and UHFf to SMA adapters.   Just need to cut off and modify the SMA ends also using some SMT resistors for the load soldered to the connector body.  Keeps all the connections real short.

     

    Well,  I followed your link to Kirkby and already got a response from David about possibly sending the OSL in.  Your right about the cost.  The actual cost of the service is very reasonable but it's all the added shipping and taxes and tariffs that add up.  Oh, and not to mention our very unfavorable currency exchange rate at the moment!   Anyway, as luck would have it I received an email file today from the original OSL manufacturer.  I had all but given up on finding them.  They list what they are calling "industry standard" coefficients for the open.  C0-C3.  Also effective lengths and loss.  The values are all there but they don't label them with any units of measurement.  I see no information (L0-L3) about the short.  I'm not sure what is actually needed for the Instrument View software to calibrate correctly.  Any ideas?

  • in reply to dc2daylight

    Hi Mark,

     

    There's a screenshot in the blog post, also pasted below, which shows what values I needed to enter for my cal tool. From what little I know, the OSL should come with a delay value (in picoseconds) or can be a distance measurement, and in the screenshot, the FPC1500 InstrumentView software accepts a distance value (if you have a delay value then that can be converted using the normal wavelength conversion formula lambda=c/f). That needs to be entered for both the open and short values (see yellow arrows). Apparently, it is right that the short length should be very slightly larger than the open length, for a good OSL set.

    Perhaps if you have access to extremely excellent depth gauges, and maybe metalwork shop, this could be self-measured, but I'm guessing it is not as easy as this - It's way out of my knowledge area : (

    The L values are very likely to be all zero, they are unmeasurable mostly. The C values will vary, but apparently, C0 is closely related or linear in some way to the Open Length, so if you know one you can compute the other - I just don't know the formula to do that though! (Edit: thinking about it, I can imagine it's just directly linear, since the only dimension of the capacitor that is changing is the length).  The C values below are in fF and fF/GHz so if your numbers look to be in a similar ballpark then it's likely they are in fF/GHz too. I do not have a Through piece, so that's set to zero.

     

    image

Comment
  • in reply to dc2daylight

    Hi Mark,

     

    There's a screenshot in the blog post, also pasted below, which shows what values I needed to enter for my cal tool. From what little I know, the OSL should come with a delay value (in picoseconds) or can be a distance measurement, and in the screenshot, the FPC1500 InstrumentView software accepts a distance value (if you have a delay value then that can be converted using the normal wavelength conversion formula lambda=c/f). That needs to be entered for both the open and short values (see yellow arrows). Apparently, it is right that the short length should be very slightly larger than the open length, for a good OSL set.

    Perhaps if you have access to extremely excellent depth gauges, and maybe metalwork shop, this could be self-measured, but I'm guessing it is not as easy as this - It's way out of my knowledge area : (

    The L values are very likely to be all zero, they are unmeasurable mostly. The C values will vary, but apparently, C0 is closely related or linear in some way to the Open Length, so if you know one you can compute the other - I just don't know the formula to do that though! (Edit: thinking about it, I can imagine it's just directly linear, since the only dimension of the capacitor that is changing is the length).  The C values below are in fF and fF/GHz so if your numbers look to be in a similar ballpark then it's likely they are in fF/GHz too. I do not have a Through piece, so that's set to zero.

     

    image

Children
  • in reply to shabaz

    That's very helpful.  I think my OSL numbers must be in similar units.   My C0 is much higher than yours but the length is greater so that may explain the differences.  I was a bit confused when seeing that some of the C's are a negative value. The O/S on mine have exactly the same length.  Both the length and the delay is stated.  When I did the math it works out correctly. (Well, I didn't really do the math but used an online converter)  image  Here's some of the data.

     

    Impedance: 50 Ohms

    Frequency: 11GHz (usable to 18 GHz)

    Effective Length: 26.5mm (O/S)

    Load Power Rating: 2W (33dBm)

    Coefficients for open:

    Calibration

    Coefficients

    Co         C1       C2       C3          Delay      Loss

    167.8    -62 .0   4.6      -0 .032    88.3ns     2.2

  • in reply to dc2daylight

    Hi Mark,

     

    That makes sense, it seems reasonable (the 88.3ns is a typo, it's 88.3ps of course).

    Just as an aside, as I understand, for those who do not have this detail for their OSL tool or were not sure if the values were ok, the C0 to C3 values can all be set to zero, and it will not make a huge difference, so if one wanted to double-check that all is well, they could be deliberately set to zero, do any measurement and then set them to the expected values, and re-do any measurement, and compare the results, they should look close, for at least a couple of GHz I'm guessing.

    Also, there's a zero-$ calibration method I'm patenting (not really : ) if you look down the hole of the Short, if you can see metal at the bottom (as it should be) then it's possible to get a soft thin wooden splint (e.g. shave a wood barbeque skewer) until it can be put into the hole (ideally not leaving any wood fibres or touching/scraping anything) until it gently touches the metal at the bottom, and then mark off with a pencil, to the metal cylinder inside the N male connector.. like a car dipstick. And then remove the splint and measure with calipers. This cannot be done with the Open however, because that will use some insulating support and will not have a flat base at the expected depth. But we know the open and short should be extremely similar, so they can both be set the same. And then C0..C3 can all be zero, as can L0..L3. I think with the wood splint, the accuracy can easily be to within 0.1mm if repeated a few times for certainty, and just eyeballing the pencil mark at the top, since it's not possible to precisely draw the mark with a normal pencil. I think it could be good enough for 1-2GHz, but I'm speculating though! I have nothing to base it on. Also, probably not a good idea for super-expensive OSL tools to put wood splints inside them.. but seems reasonable otherwise, and could also be dusted out with a camera-lens type blower.

  • in reply to shabaz

    Yeah, that's odd the paperwork must have a typo.  I checked again and it does say 88.3 nS.  It's unlikely that I will be working much above 1.3 Ghz and that won't be anytime soon.  Heck, I have a NanoVNA that holds it's own up to there!

     

    So, I'm about to pull the trigger on a fully loaded FPC-1500 special bundle available.  How long have you used one and do you have any comments (negative or positive).  I watched a number of online reviews and they have all been very favorable.  Is there any other equipment I should do a comparison on?  I haven't found anything at the price point that has all the features.  A good friend has a KS N9320B with a TG and ithe FPC data sheet appears to compare quite nicely (if I'm not missing something) at a fraction of the cost. 

  • in reply to dc2daylight

    Hi Mark,

     

    I did a video review of it several years ago, viewable here: https://www.youtube.com/watch?v=932HrQYbEEI

    and written review: FPC1500 Spectrum Analyzer - Review

    Since then, it's my main RF instrument. I've also used a N9322C (which I'm guessing is going to be very similar to the N9320B), and the review compares briefly with that. Both are great instruments. The Keysight instrument is a traditional spectrum analyzer, whereas I think (not sure) that internally the FPC1500 is using a mix of digital and analog techniques to do things like have down to 1 Hz resolution bandwidth.

    I really like that the FPC1500 is totally silent (no fan), very good PC software and large amount of features. The corners that are cut to reduce the price are fair things I feel, like the display doesn't have great viewing angles.. although on the plus side the display is larger. The Keysight instrument is physically much larger, and has the fan (although not as noisy as older HP/Agilent spectrum analyzers). Both instruments have very good software. The VNA capability on the FPC1500 is very useful, although if you already have the NanoVNA then you're mostly already covered on that point. For such a price range both the FPC1500 and the Keysight are excellent products to be honest.

  • in reply to shabaz

    I have exhaustedly compared the specs and reviews on a number of entry level SA's.  I have used my friend 's N9320B w/TG  (no VNA capabilities we use an external RLB).  He bought that about 6 years ago for more than 2.5 X the price of the current FPC-1500 fully loaded promotional offer.  At such a cost difference I figured I had to be missing something!  Makes me thing are they about to discontinue this model?   Not that will matter much to me since it can do most everything I need it for for many years to come.  Yeah, still have to get over such good things can actually come in a small/lightweight package.  As I look at the HP 8657 taking up tons of real estate on my bench!  It's shocking that to step up to any higher performing SA the cost will increase 3-4 fold and most of those units don't have the VNA capability.  The NanoVNA is a great little device and is nice for portable use but I'm sure nothing comparable to the FPC-1500.