Eliminating the cable influence in antenna measurement

[FvM]

network analyzer open short match

Hello,

I want to measure the input impedance of an electrical small monopole on a 433 MHz transmitter PCB board. Cause the system has no clear symmetry plane, it's difficult to place the cable without affecting or possibly completely changing the impedance.

I'll have an opportunity to make the measurements in an university lab, but the staff probably don't has sufficient experience regarding this particular measurement problem.

Can you suggest a suitable technique to reduce or at least validade the cable influence with that measurement? I thought of some cable windings through a ferrite core as common mode choke or try to find a neutral position by varying the cable position and watching the impedance reading.

Thank you for your help in advance!

 

[vfone]

if you want to measure vector network

Calibrate for one-port your Network Analyzer using a thin coax cable that you are using for antenna measurement.

For calibration attach to the coax cable a small SMT GSC type connector.
**broken link removed**

During calibration you have Open, Short and for Load attach to the GSC connector a 50 ohms 0402 resistor (preferable 3 resistors 150 ohms in parallel to minimize the parasitic inductance).

Solder the GSC connector on the board, near the terminals of the antenna that you are looking to measure.

 

[nccu1990]

metal interaction on bazooka antenna

If you want to measure the antenna impedance without the connector, you can measure like below step.

(1)Do the one port calibration OSM ( open ,short ,match->50Ohm) at the Vector Network Analyzer.
(2) Connect the SMA cable (which you want to solder at antenna input) at the VNA test cable
(3) Use the VNA port extension function, offset the SMA cable length
(4) Disconnect the SMA cable at the VNA test cable
(5) Clean the ground plane at the place which you want to solder. Ground is important at the High frequency measurement.
(6) Solder SMA cable at the antenna input port. Be careful solder the signal and ground point.

 

[FvM]

how to block flowing currents on antenna cables

Thank you vfone and nccu1990 for the detailed suggestions!

I see, that my questioned wasn't as clear as I thought. The impedance calibration referred to a reference plane at the cable end respectively antenna input isn't an issue. I'm not familiar to it, so it's good anyway, that you explained it briefly.

The point is, that the antenna radiation itself is affected by the connected cable, that acts as part of the antenna itself. The effect is rather strong, cause DUT is a monopole with a very small ground plane. The ground of antenna input port is a ground for the connected transmitter, but not when seen from the outside.

 

[flatulent]

antenna 433mhz near field dut

Your problem is that the outer conductor of the coax will act as part of the antenna. You have two options.

1. Put plenty of ferrite toroids around the coax as close as possible to the antenna. Use the smallest inner diameter that will fit over the coax.

2. Put a bazooka over the coax. This is a quarter wave cylinder that has one end as close as possible to the antenna and the other end is snorted to the outer conductor of the coax. The antenna side looks like an infinite impedance and keeps currents from flowing.

 

[FvM]

The "bazooka" λ/4 bandstop is an interesting idea. Although the outer cylinder would still act as a passive coupled antenna, it should effectively block the current flow through cable shield.

The data of MnZn ferrite beads (hollow cylinder shape) promise a sufficient high impedance, so they are my first choice.

 

[biff44]

I have been thru all this before. You do not want your cable to be part of the antenna system, since it will make an electrically small monopole look much better than it is.

One idea is to put a battery powered oscillator right at the antenna terminals. Make sure the battery is small and mounted right at the PCB. Use inductors on the battery lines so it does not look like a ground plane extension. Then do the pattern measurements on the antenna is it is moved, and use something like a spectrum analyzer to measure the absolute amplitude. Then, to calibrate your measurement, move the oscillator/battery to a calibrated dipole antenna and repeat the measurement. The difference between the two will give you the antenna gain in dBd. You then calculate dBi=dBd + 2.15

If you Must use a network analyzer and a long cable, you want to load it with ferrite clamps every 6" or so. You can try the idea about shorting the outside cable currents with a quarterwave sleeve, but you are right about it affecting the field pattern.

 

[flatulent]

One way to test for the effectiveness of your efforts is to touch the cable and see if the impedance changes.

 

[FvM]

Thank you for the additional suggestions. The antenna radiation pattern can be easily measured with the transmitter itself.

The impedance measurement is intended to allow a systematic design of matching network. Another lab made impedance measurements and suggested a matching network. But the measurements are inplausible cause they don't correspondend to the proposed and other empirical determined matching networks. The most likley explanation is, that the measurement was performed through a cable without the discussed isolation methods.

 

[RF-OM]

I think your monopole is single-ended antenna and you need to isolate the feeding line and minimize any disturbance to the field around. If the second requirement is important ferrite beads may be not the best choice, they are not small and affect the field a lot. Bazuka is good for narrowband practical antennas, but in your case it may affect your measurements because it is big and narrowband. Both of these ways put some disturbance into feeding line and it may affect your measurements also. There may be one more way. It is to use small differential line connected through a small size balun (usually ceramic chip one). This feeder may be done on flex film and in this case you may move it around and find the best location. Flex film if not so expensive. On the opposite end you may install the same balun and get single-ended line again. This feeder can be easily measured alone and then its S-parameters may be de-embedded from the measurements. You even may calibrate your VNA with this line and get calibration plane at the end of the balun. What it is important is to fasten the balun and the end of the line to your board with some plastic elements. In this case you will not disturb your setup. Usually plastic parts are not a problem and widely available. Feeder may be done with non-flexible substrate too. The main advantage that you have full isolation from the antenna board and differential line provides excellent isolation from the fields and any common mode signals. If you need help with physical design of small differential line please let me know.

 

[FvM]

Thank you for the interesting suggestion. I also wondered, if a transformer could be a solution. It's good to hear from a RF specialist, that it's also an applicable technique.

The point is, that I don't intend very accurate measurements. The DUT won't operate in free field in it's final application anyway. But I want to avoid a matching network design based on systematically incorrect measurements (that apparently happened with the measurements from another lab). I feel, that ferrite cores as common mode chokes can be a sufficient solution and will try it first. I particularly thought of using thin hollow cylinders matching the cable diameter.

Regarding the suggested differential feeder: Why do you think, that a differential line is superior to a thin coaxial cable?

 

[RF-OM]

You are right, ferrite bead can do the job in eliminating high frequency currents through outer conductor of coaxial cable. If they will not interact with antenna field it is okay for your tests. But you need to check it. About differential line I do not think that it is better than coaxial cable. I know it. Coaxial cable just seems to be very good, but it is far from it. Please check, for example, Henry Ott's book and you will find the table with practical numbers for different types of transmission lines parameters. If you have no this book I may send to you this table as scanned page. Coaxial cables are far from to be perfect.

Transformer is a good solution when you need galvanic isolation, but it will not help when you need to minimize any interventions into electromagnetic field around your DUT. In order to minimize it you need to have very thin probe with as smaller metal parts as it possible. In this case small size differential line can be a good choice. It may be done with high impedance, which can be reduced at the ends with transformers, especially for narrow band applications. But even without impedance reducing you may use it because you know this line (probe) parameters and always can de-embed them.

When we are talking about EM field interactions even plastic parts may contribute to final result because they are occupy some space and change the effective dielectric constant of antenna environment. Depends on your application you need to count such effects.

I do not know your particular situation, but very often I use short (about 1 to 2 cm) thin (47 mils or about 1.18 mm) semiridgit coaxial cable as end-lunch. It is good for measurements, but definitely may affect antenna EM field. It is easy to calibrate VNA with this end-lunch probe and you may not be concern about currents flowing through outer conductor.

 

[biff44]

This is the problem with this forum. People seldom ask the question they are looking for. They ask some other unrelated thing, and slowly leak out more information as others respond. We are not amused.

IF all you wanted was to figure out how to match your antenna to a specific transmitter, then set up the transmitter at a fixed distance from a receiving antenna and vary the matching network between the transmitter and antenna. Make up an LC matching network, use a simulation program to postulate various antenna impedances and reactances, change the L's and C's as the transmitter steps across the frequency range. Plot the magnitude vs. frequency at the receiver. You can then deduce what the antenna's impedance was.

 

[FvM]

I don't think that I kept back information. I said that I want to measure an antenna impedance and I also mentioned in my second post that's a transmitter application. You can most likely imagine that an electrical small monopole antenna is involved with a battery powered transmitter rather than a base station.

Your absolutely right that the impedance can be measured indirectly by modifying a matching network with a generator connected. The method can be used, if no VNA is in reach or no meaningful results are expectable from the specific conditions. This would be the case probably with a very small transmitter.

On the other hand, if you regard this method as preferable, you can use it also if the generator doesn't already exist at the device under test, as you suggested previously.

I think however, that it's rather long-winded and limited accurate to get the results from a scalar measurement in the suggested way. At lot of error terms has to be considered in the calculation. Also the systematic modification can be done only by changing very small SMD parts, the hardware durability sets a limit to the number of permutations. But I'll have it as a last resort.

 

[FvM]

As an update: I made a cable with tubular ferrite cores as common mode filter or ground breaker. The total length is 30 cm with 15 cm of ferrite cores. Touching the cable behind the ferrite zone has no noticeable effect on the antenna impedance measurement, while removing the ferrites changes the impedance in a dramatic way and makes the cable also touch-sensitive. I expect that the first cms of cable still modify the antenna impedance, but not to a high extent.

 

[RF-OM]

Very good! I am glad you solved your problem. By the way, there is one ramification for this method: to use less numbers of beads, may be even one, but put more than one turn of the cable around the bead. Of course you need to use semiflexible coax instead of semiridgit, but sometimes it may be better than use long train of beads. people often do such cable as permanent usage probe and keep it as tool. In this case usually teflon tape is used over the ferrite core to fix it in the same position and keep protected.

 

[FvM]

A ferrite core with mutiple turns (at least two) of a flexible coax is also one of my standard means of ground breaking in measurement and instrument design. In case of the present antenna measurement, I had a feeling, that a thin, distributed choke would be superior, but I didn't validade this assumption.

The impedance curves from a ferrite manufacturer suggested a higher impedance for a single turn compared to a double turn choke from 200 to 400 MHz upwards. It promises several 100 ohms at 433 MHz for each ferrite bead, which would effectively leave only a small radiating stub added to the DUT. Another reason for using a semirigid cable was to have it as a support for the antenna as well.

 

[RF-OM]

Absolutely agree. I told you about multiturn version only to share some experience. I knew that it is not for this case, but for future. It may be helpful.

 

[srhoar]

My job is to do antenna measurements.

First:
Cal out the connectors and cables with a network analyzer.(NA)
second:
hook up a known antenna to the output of the NA.
third:
hook up the antenna to the input port of the NA.
fourth:
rotate the antenna and measure the dB loss per degree of rotation.
fifth:
plot

Hope this helps,

Red

 

[RF-OM]

I think the initial question was not about how to measure the antenna directivity but how to connect VNA without disturbing the Antenna Under Test near field by cables and connectors. They may be bigger than AUT itself. For discussed case it was not very easy.