History

This is yet another antenna analyzer, but with a twist. The analyzer was originally designed to replace my good old MFJ-259.
(not the B version) Although the MFJ has proven to be a useful tool, it is not a serious measurement instrument. That's fine for most applications, but if you want evaluate system loss or verify antenna models you really need something more accurate. There are a lot of these analyzers available both from well known manufacturers like MFJ, AEA, Timewave, Palstar etc. as well as individual amateurs and radio clubs.

The majority of these instruments are based on a reflection bridge in some shape or form. As I will explain this limits the impedance range of an instrument for a given accuracy. Since I mainly work with open line fed antennas I needed an instrument that could cope with the broad impedance range these antennas exhibit. So I decided to design and build my own analyzer based on a different measurement method.

Although the concept for this analyzer was conceived early 2005, the first prototype, which I will refer to as version 1 was built in 2007. This version had the basic functionality and was designed around an Atmega8 controller. During the second part of 2007 the software was rewritten and a number of additional features were added. To accommodate the extra code I needed to switch to a different controller with a little more memory space.

Version 2 is based on an ATMega168, which is pin compatible with the original Mega8. Therefor only minor changes in the hardware were needed to make it work. Although some routines were made “more elegant” and additional routines were added, most of the code remained unchanged. After a period of debugging and testing I now feel that Version 2 is ready to be released.


Measurement method

The problem with the reflection based method is its accuracy over a wide range of impedances. Let me illustrate this with an example:
A 1kΩ load would result in a reflection coefficient ρ of 0.904 in a 50Ω system. 2kΩ gives a value for  ρ of  0.975.  So a 100% increase in impedance only results in a 8% increase in the value of ρ. Given the fact that ρcan only be measured with a finite accuracy the overall error for the impedance will increase rapidly as ρapproaches unity. This is illustrated in the diagram on the left. For a 1% error in ρ the resulting error in the calculated impedance is already 10 times larger for a value of ρ just under 0.9. The problem gets even worse if the unknown impedance is highly reactive. For pure reactive loads ρ is always 1 irrespective of the reactance. The actual value is now solely determined by the phase of ρ. Accurate phase measurements on HF, however, are difficult. So I decided to adopt a more classical approach andsimply measure the voltage across the load and the current through it as well as the phase difference between the two. Strangely enough most professional units are also based on a reflection measurement and I know of only one instrument that uses the voltage  current method : The  Tomco TE1000


Design objectives

  • Frequency range of 1 - 30 MHz
  • Good accuracy over a broad range of impedances
  • Low cost.
  • Low power.
  • Stand alone solution.
  • Battery operated.
  • PC interface for frequency scans and impedance plots
  • In-circuit programmable

Circuit description

Crucial to this design is the choice of the type of detector. In an application like an antenna analyzer the detector should have a good linearity over a wide dynamic range. In most other designs either diode detectors or  logarithmic amplifiers are used. These often lack linearity and accuracy. In this case a synchronous detector was chosen. In addition to a good linearity over a wide enough range a synchronous detector has some other distinct advantages; It is a narrow band detector, which helps rejecting signals received by the antenna itself, and it combines the function of both amplitude and phase detector. 
The DC voltage at the output is proportional to Vin*COS(Ф), where Ф is the phase difference between the input signal and the reference signal. So, to determine the magnitude and phase of either Voltage or Current two measurements are necessary. One at 0° and one at 90°. This is known as In-phase and Quadrature detection, in short I/Q. From these four values  (Vi,Vq,Ii and Iq) the real and imaginary parts of the impedance can be directly calculated:

History

This is yet another antenna analyzer, but with a twist. The analyzer was originally designed to replace my good old MFJ-259.
(not the B version) Although the MFJ has proven to be a useful tool, it is not a serious measurement instrument. That's fine for most applications, but if you want evaluate system loss or verify antenna models you really need something more accurate. There are a lot of these analyzers available both from well known manufacturers like MFJ, AEA, Timewave, Palstar etc. as well as individual amateurs and radio clubs.

The majority of these instruments are based on a reflection bridge in some shape or form. As I will explain this limits the impedance range of an instrument for a given accuracy. Since I mainly work with open line fed antennas I needed an instrument that could cope with the broad impedance range these antennas exhibit. So I decided to design and build my own analyzer based on a different measurement method.

Although the concept for this analyzer was conceived early 2005, the first prototype, which I will refer to as version 1 was built in 2007. This version had the basic functionality and was designed around an Atmega8 controller. During the second part of 2007 the software was rewritten and a number of additional features were added. To accommodate the extra code I needed to switch to a different controller with a little more memory space.

Version 2 is based on an ATMega168, which is pin compatible with the original Mega8. Therefor only minor changes in the hardware were needed to make it work. Although some routines were made “more elegant” and additional routines were added, most of the code remained unchanged. After a period of debugging and testing I now feel that Version 2 is ready to be released.

Measurement method

The problem with the reflection based method is its accuracy over a wide range of impedances. Let me illustrate this with an example:
A 1kΩ load would result in a reflection coefficient ρ of 0.904 in a 50Ω system. 2kΩ gives a value for  ρ of  0.975.  So a 100% increase in impedance only results in a 8% increase in the value of ρ. Given the fact that ρcan only be measured with a finite accuracy the overall error for the impedance will increase rapidly as ρapproaches unity. This is illustrated in the diagram on the left. For a 1% error in ρ the resulting error in the calculated impedance is already 10 times larger for a value of ρ just under 0.9. The problem gets even worse if the unknown impedance is highly reactive. For pure reactive loads ρ is always 1 irrespective of the reactance. The actual value is now solely determined by the phase of ρ. Accurate phase measurements on HF, however, are difficult. So I decided to adopt a more classical approach andsimply measure the voltage across the load and the current through it as well as the phase difference between the two. Strangely enough most professional units are also based on a reflection measurement and I know of only one instrument that uses the voltage  current method : The  Tomco TE1000

Design objectives

  • Frequency range of 1 - 30 MHz
  • Good accuracy over a broad range of impedances
  • Low cost.
  • Low power.
  • Stand alone solution.
  • Battery operated.
  • PC interface for frequency scans and impedance plots
  • In-circuit programmable

Circuit description


Crucial to this design is the choice of the type of detector. In an application like an antenna analyzer the detector should have a good linearity over a wide dynamic range. In most other designs either diode detectors or  logarithmic amplifiers are used. These often lack linearity and accuracy. In this case a synchronous detector was chosen. In addition to a good linearity over a wide enough range a synchronous detector has some other distinct advantages; It is a narrow band detector, which helps rejecting signals received by the antenna itself, and it combines the function of both amplitude and phase detector. 
The DC voltage at the output is proportional to Vin*COS(Ф), where Ф is the phase difference between the input signal and the reference signal. So, to determine the magnitude and phase of either Voltage or Current two measurements are necessary. One at 0° and one at 90°. This is known as In-phase and Quadrature detection, in short I/Q. From these four values  (Vi,Vq,Ii and Iq) the real and imaginary parts of the impedance can be directly calculated:

 and 

The subscript S denotes the series equivalent of the impedance as in Z = Rs + jXs. Note that the calculation does not involve trigonometric functions like SIN(Ф) and COS(Ф), which are difficult to evaluate with a simple controller. Some designs using a separate phase detector, notably the ones based on a AD8302 have the problem of phase ambiguity. So whilst these analyzers are able to measure the magnitude of a reactance, they cannot tell whether it is inductive or capacitive. The MFJ269 is an example of such an instrument. In the TAPR VNA this is solved by using the I/Q technique with two AD8302 circuits. The synchronous detector does not have this problem. The output of the detector can go negative as well as positive. So the phase angle is defined in all four quadrants. (0° - 360°)

The measurement signal is generated by a classical PLL synthesizer. No single VCO will be able to run from 1 MHz to 30 MHz without switching inductances or capacitors. Therefore the VCO runs at 49 MHz to 78 MHz and is than down converted to 1-30 MHz by mixing it with a 48 MHz X-tal oscillator. This arrangement also makes it possible to generate the necessary I/Q reference signals. The fixed 48MHz signal is split into a +45° and a -45° component which are fed to two separate down converting mixers. The output of the mixers are therefore always 90° out of phase. 

To prevent systematic errors due to gain differences a single detector –amplifier chain is used for all voltage and current measurements. Reference, voltage and current signals are fed to the detector through diode switches. The current measurement is done by measuring the voltage across a sense resistor in series with the unknown load.

Look here for a more details

Test Results

This table shows test results from ARRL , Bob W5BIG and myself. Notice that most commercially available analyzers perform rather poorly at high impedances. This is partly due to the limitations of the measurement method and partly because of a more general problem with this type of measurement : stray capacitance. Every connector or test fixture has a certain amount of capacitance between its terminals. Not taking this parallel capacitance into account will result in large measurement errors. The reason why my design and Bob's AIM perform so well is that we both do calibrate for stray capacitance and correct the measurement results.

Calibration

Since stray capacitance is the most prominent source of error it is measured during the calibration procedure. The measured capacitance is than used to correct each reading. In version 1 calibration was automatically done at start-up. 
In practice this was a bit annoying because every time the load had to be disconnected. In version 2 calibration values are stored in EEPROM. A new calibration is performed when the function key is held down during start-up.
Appart from the zero reference of the ADC / Amplifier nothing else has to be calibrated because all measurements are essentially relative.


Construction

The analyzer is built on two separate PCB's. The HF PCB contains the PLL, VCO, Mixers, Switches, LPF, 48MH oscillator and synchronous detector. The Controller PCB carries the Controller, Differential amplifier, Display and RS232 level converters. The HF board is a double layer board of which the component side is used as a ground plane.
The connections between the two boards carry only DC signals, so wiring is not at all critical.

Look here for downloads and details


Menu

The extra features available in version 2 made it necessary to extend the menu structure.
You can cycle trough the menu items by repeatedly pressing the function key. Each setting can then be changed by rotating the encoder. The instrument will return to its normal operation after 2 seconds of inactivity in menu mode.
The only setting that is not set from the menu is the frequency step size. Changing the step size is done much more quickly by pressing and holding the function key and turning the encoder. 

Construction

The analyzer is built on two separate PCB's. The HF PCB contains the PLL, VCO, Mixers, Switches, LPF, 48MH oscillator and synchronous detector. The Controller PCB carries the Controller, Differential amplifier, Display and RS232 level converters. The HF board is a double layer board of which the component side is used as a ground plane.
The connections between the two boards carry only DC signals, so wiring is not at all critical.

Look here for downloads and details


Menu

The extra features available in version 2 made it necessary to extend the menu structure.
You can cycle trough the menu items by repeatedly pressing the function key. Each setting can then be changed by rotating the encoder. The instrument will return to its normal operation after 2 seconds of inactivity in menu mode.
The only setting that is not set from the menu is the frequency step size. Changing the step size is done much more quickly by pressing and holding the function key and turning the encoder. 

CIR
 The first menu item is CIRcuit; originally display mode. You can choose between SERies equivalent circuit and PARallel  equivalent circuit.
AVG
In version1 the number of averages was fixed. Now you can set it from 1 to 128 in a power of 2 sequence. Increasing the number of averages improves readout stability for high impedances.
BKL
Backlight intensity works basically the same as in V1, only the level is now displayed as a percentage.
BAR
This is a new feature. You can select the Bar display source : SWR, X or R or select OFF to return to the normal alpha numerical display. By pressing and holding the function key you can set the upper value of the Bar to the current measured value. So you can set the sensitivity of the display to any value you like.
Z0
In version 1 the SWR was calculated relative to a fixed characteristic impedance of 50Ω. In version 2 you can select the required value from the following set of standard values:
50Ω, 75Ω, 300Ω, 450Ω.


To support frequency scans and impedance plots I wrote a Windows program that will control the analyzer and make graphs of the measured results. It features auto scaling, has a zoom function and will display multiple scans. The option to save and retrieve measurement data makes it easy to compare results with earlier measurements. Measurement data is stored in text format to make it possible to import data from another source or to use the data in Excel.
It also offers a separate Smith Chart window to display complex reflection coefficient data. Moving the cursor over the impedance plot will show the measurement results for each individual frequency point, both on the linear graph and Smith chart.