Antenna measurement

The article describes the methods of measuring SWR and antenna impedance as well as the problems associated with it. However, it should be noted at the outset, that a potential applicant for construction will no longer do without basic knowledge of programming, whether PC or microprocessors. Hardware simplicity is redeemed by more complicated mathematical calculations.

1. Basic principle

Only bridge methods are used to measure a complex quantity such as the antenna impedance. These work with sufficient accuracy and, when properly designed, process a large range of frequencies up to GHz. A partial disadvantage is the need to supply a measuring frequency generator with relatively strict requirements for spectral purity and considerable power.. The bridge method itself allows various types of measurements not only on antennas but also on lines and power supplies. A bridge of this type is also used in the well-known MFJ259 device. The connection is trivially simple:

After connecting the generator, setting the required frequency and connecting a complex load (antennas) the bridge supplies three voltage information, which are sufficient to calculate the impedance and PSV (SWR). In the following description, we will follow this marking:

FWR…..half voltage of the measuring frequency generator
REF…..differential voltage in the diagonal of the bridge characterizing the PSV
VZA…..voltage on the measured load of a complex nature

2. Practical design

Due to the nonlinearity of the diodes in the field of rectification of low voltages used as detectors, it is necessary to at least partially compensate for this nonlinearity and amplify the output voltages to a sufficient level for measurement purposes.. The single channel compensation amplifier is shown in the following figure:

Each of the bridge outputs has its own amplifier. In the practical version, we set the same gain of each branch by the resistor R2. Diode D1 should be of the same type as used in the bridge. The most suitable are Shotky diodes with ZERO BIAS properties specially designed for the detection of low voltages, e.g.. 1PS79B62 (Philips) suitable up to GHz, but to 500 MHz their choice is more than rich… Triple types are very suitable, e.g.. HSMP-386L firmy Agilent, but they are made by many manufacturers. Operational amplifiers are of the RAIL to RAIL type with one power supply, these are also blessed.
Resistors in bridge R1, R2 and R3 should be non-inductive, eg 1206 but TR191 without wire leads are also very suitable, which can be used up to several GHz. It does not matter the absolute value, it is important that they have the same value, which is then included in the calculation as Rx.

3. Measuring frequency generator

It is possible to use any generator of the required frequency range with sufficient power at least 20 dBm and low distortion. Conventional LC oscillators are suitable, do 500 MHz it is possible to make a mixing generator without any problems. The following example shows a possible solution for the shortwave region. On the Internet you can find a detailed diagram of the HF analyzer called RAINBOW from which is the following diagram:

The parts used are completely common. Even all broadband, you don't even have to wind and tune impedance transformers and low-pass filters. American company Coilcraft http://www.coilcraft.com/ it has it all in its production program and, in addition, it sends it to such a banana republic completely free of charge – just fill out the form and talk a little in English. At least it has worked for me for about two years now…
Elegant generator solution for HF up to 50 MHz is using a DDS circuit, which is direct frequency synthesis. The involvement with the company's IC is extended Analog Devices typ AD9851 whose reworked example is in the following figure. The circuit is usually controlled by a microprocessor, which is a suitable solution, because, among other things, there are no problems with frequency measurement – the programming word corresponds to the frequency, which can be easily displayed. A similar solution was once described in RŽ 3/98.

It must be said that the company also produces more powerful DDS circuits operating at much higher frequencies. These are less suitable for amateur use (many feet, small cases) and are very expensive. In addition to AD, some such circuits are also manufactured by HP, but for amateurs without much importance.

4. Own calculation

The calculation is performed in absolute units – Volty, Ohmy, Farady a Henry – thus the following formulas are compiled. We expect, that the measuring instrument shows us actual voltages all three quantities and the generator supplies the power frequency Fx 20 dBm (100mW/2.2V/50 Ohm) spectrally pure as “the word of God…”:

Standing wave ratio – PSV , the REF limit values ​​must be checked, for which it is not relevant to continue the calculation, because PSV would be too high:

FWD = 2 * FWR P = REF / FWD
PSV = (1 + P) / (1 – P)Characteristic impedance – Like this , it is necessary to check the limit values ​​of VZA which indicate unconnected load or load in short circuit. In the formula, Rx is the value of resistors R1 ÷ R3, in our case 50 Ohms.

So = (Rx * VZA) / (FWD – VZA)The real part of the impedance – R:

R = (Rx² * Like this²) * PSV / (50 * (PSV² + 1))Imaginary part of impedance – X:

X = SQR (Like this² – R²) where SQR is the square root

Determining the nature of impedance:

1. Increase the generator frequency Fx by a small value
2. Recalculate the characteristic impedance value Z1 = (Rx * VZA) / (FWD – VZA)
3. if Z1 > Like this then the impedance is inductive in nature L
4. if Z1 < Like this then the impedance is capacitive in nature C

If you decide to build such an analyzer and use a microprocessor with voltage converters for calculation, e.g.. PIC… or AWR… I recommend 10-bit types. Then set the reference range of the AD converter to 2.5V and do not forget to be aware when programming, that the read bit value of the measured voltage in the register has from the actual “very far…” and needs to be converted to fair value:

One = (2.5 / 2¹⁰) * B₁₀where B₁₀ is the decimal value of the contents of the measured voltage register. Applies to 10-bit converter, 2.5At reference voltage and full range 10b.

5. Elimination of measurement errors

As can be seen from previous calculations, a significant proportion of the error in the impedance calculation is due to inaccuracies in the determination of PSV. This is understandable because at low PSV values ​​the measured voltage REF is in the region of the most non-linear part of the diode characteristic, which cannot be compensated and so the error is transferred to the calculation of the real and imaginary part of the impedance. In boundary conditions, the error can also exceed 15-20% which devalues ​​the measurement results.
Fortunately, there is an elegant solution for determining PSV using specialized circuits of the company MAXIM-DALLAS type MAX2016. The circuit consists of two logarithmic detectors operating in the frequency range from LF to 2,5 GHz and with a dynamic range up to 80 dB. A detailed description and use of the circuit can be found at Maxim-IC.com where to download datasheet. Its big disadvantage for amateur use is the QFN-28 case with dimensions of 5x5mm, which can only be soldered by surface mounting technology. Therefore, I will only present the principal connection for the measurement of PSF and the related calculations, especially for those, who do not know English.

Directional binding (created e.g.. PCB guidance) it is connected to the inputs of logarithmic detectors. Exit OutD is the output of the internal differential amplifier, whose inputs measure the differential voltage of the outputs of the logarithmic detectors. The calculation itself is in principle simple. Reflection losses are calculated first RL v dB (PSV measurement is little used in professional practice, the expression of reflection losses in dB prevails):

RL = (VoutD – Vcenter) / Slope where

VoutD……Differential voltage converted to an absolute value in [V]
Vcenter….Medium output voltage OutD, typically 1V for R1 = 0, as you can see from the attached chart
Slope……MV / dB conversion ratio, typically 25mV / dB for R1 = 0

The PSV value is simply calculated from the reflection losses expressed in dB:

P = 10 ^{-(RL / 20)}
PSV = (1 + P) / (1 – P)

Logarithmic detectors usable for this purpose are also manufactured by Analolog Devices e.g.. AD8362, or type AD8364 very similar to MAX2016 but in yet “more unsuccessful” 32 outlet housing with dimensions 5x5mm. All of these circuits can be used to measure gain, performance at real load, as a sensitive RSSI meter for routing and adjusting antennas, etc..

Today, every more serious company provides its customers with extensive technical assistance, including the development of applications offered by integrated circuits and the best and the FREE SAMPLES delivery service. (Free samples). Unfortunately, AD is not one of them, represented in post-communist countries.

DALAS-MAXIM also manufactures other types of logarithmic detectors, one of them MAX2015 is shown in the picture on the left in the circuit as a sensitive RSSI detector. It is a single-channel type with a high sensitivity of -65dBm to + 5dBm (citlivost 0.125mV/50 Ohm) in the frequency range 0,1 to 2,5 GHz. It is housed in an 8uMAX SMD package, which is more suitable for amateur use.

There are many more in the company's catalog, for radio amateurs interesting circuits such as e.g.. MAX2620 what is LC oscillator to 1050 MHz, separating, broadband amplifiers, etc.. Too bad only, that most of the most interesting ones are in miniature SMD packages which complicates amateur use. Unfortunately, this is a world trend and maybe about 10-15 years, we will know the discreet part only some earlier births…

But the best part is, that normally sends two pieces of each selected type for FREE, so it really pays to visit their site even though they have changed the structure in recent months and since then I have not been able to get to the FREE SAMPLE ORDER…but maybe it's just a new one, perfectly non-debugged delivery service.

6. Measure or model antennas ?

…That is a question. If we consider HF bands, measurement is not very important. Modern modeling programs such as e.g.. MMANA, NEC-2, EZNEC, NEC WIN + and others will serve us as well as an antenna analyzer. In addition, no effort and cost of construction, while the error we can count on is practically negligible and the adjustments on the antenna “we do” comfortable at home at the table. We do not have to “Niesakovať” monkey climbing on the roofs.

The situation with modeling in the 2m and 70cm bands is similar, if not better. The accuracy of hardware measurement decreases slightly with increasing frequency, while mathematical modeling (regardless of the MININEC algorithm used, NEC2 or NEC4 is always a torque method and the individual types differ only in the comfort of the operator) gives about to 500 MHz astonishingly accurate results.

If we intend to model antennas other than YAGI, vertical or wireframe – e.g. parabolas or Helical – it is more advantageous to use a program working with the NEC algorithm, which models planar or geometric surfaces as solid structures while the older MININEC algorithm (it is used by MMANA) calculates these structures as individual primitives, which significantly extends the computation time. by the way, The unofficial Numerical Electromagnetic Code will serve those interested in antenna modeling (NEC) Archives.

The situation at high frequencies is different. The prices of measuring instruments reach astronomical heights, but even modeling programs are not exactly cheap…Of the better known, I will mention GENESYS, HFSS92, ZEALAND, MAXWELL_SV9 and many more, whose possibilities exceed the needs of even the most demanding radio amateur. They can also model such structures, which are practically impossible to measure and many technical solutions would not be possible without these programs!

So you, dear friend, have to find the answer to the question yourself. Measure or model…?

---

(C)2006 Ivan Urda, OM7UR