Modularized Spectrum Analyzer
into a Vector Network Analyzer.
The MSA/TG/VNA
This
Page was Started Oct. 17,
2007.
Updated 1-18-08. Revise paragraph "The SLIM Phase Modification"
This page is a guide for expanding the MSA, to become a 0 to 1000 MHz RF Vector Network Analyzer, the MSA/TG/VNA. I will restrict this page content to the hardware aspects of a VNA, and not give much discussion on the mathmatic processes of vector analysis. The first portion of this page contains general information for the topology of a generic VNA, and is relevant when constructing either the Original MSA/TG/VNA or the SLIM MSA/TG/VNA. The second portion is specific to the SLIM version.
What is a Vector Network Analyzer?
When an RF signal is applied to a network, such as a filter, amplifier, or transmission line, that signal is altered in magnitude and phase. If the magnitude and phase of the altered signal can be compared to the magnitude and phase of the originating RF signal, the characteristics of that network can be evaluated. The term "magnitude" is just another name for voltage or power. "Phase" is a derivation of time delay versus frequency.
Here is a block diagram of a highly simplified Vector Network Analyzer (VNA):
The network, or component, being tested is called the Device Under Test (DUT). The RF signal that is input to the DUT is the Reference signal. (It is not an absolute necessity that the Reference Signal Generator reside within the VNA, but it is a common practice.) The DUT will alter the Reference signal's two components, Magnitude and Phase. The DUT will change the magnitude component, due to it's resistive natures. It will alter the phase component due to it's reactive natures. These two altered components of the Reference signal are measured by the magnitude and phase comparators within the VNA. The key word, here is "compare".
The DUT Altered Magnitude is compared to the Magnitude of the stable Reference signal by the Magnitude Comparator. The output of the Magnitude Comparator is some value that represents the difference between the voltage, or power, of it's two input signals. This value of differential magnitude is called the Magnitude Vector.
The DUT Altered Phase is compared to the stable Phase of the Reference signal by the Phase Comparator. The output of the Phase Comparator is some value that represents the difference between the phase of it's two input signals. This value of differential phase is called the Phase Vector.
These two Vector values can be a voltage level or a digitized word. Either way, the Processor converts the Vector values to some format that the human operator can use, see and evaluate. Old VNA's only had a visual output, the CRT (Cathode Ray Tube). The old CRT has been replaced with color LCD displays. The new VNA's are totally computer controlled and will output Vector formats in various ways. Since the Vectors are digitized and processed by computers, the data can be converted to any format recognizable to software evaluation programs.
For a VNA to have any value, it must be able to take Magnitude and Phase measurements over a range of frequencies. This is called the Operating Range, or Frequency Range of the VNA. In the previous diagram, the Reference Signal Generator would be a Variable Frequency Signal Generator. A VNA with this scheme would have an operating range equal to the frequency range limitation of the comparators. Comparator circuits do not have an unlimited bandwidth. I hesitate to show actual comparator frequency values, because they vary widely among commercial and home-brew VNA's. Bandwidths from a few MHz to several hundred MHz can be obtained. As bandwidths increase, so does the cost. In the case of the MSA/VNA, the comparator limitation is 8 MHz to 12 MHz. Therefore, Example 1 VNA would have an operating range of 8 MHz to 12 MHz.
All VNA's have a limit to their operating range, and some commercial VNAs will operate higher than 30 GHz. In all cases, the VNA's magnitude and phase comparators will operate at a lower frequency. So, for a VNA to have a high frequency operating range, some type of frequency conversion sheme must be used to convert the Reference and Altered signals to the frequency range of the VNA's comparators.
In the following example, the Reference Signal Generator is a much higher frequency. The Reference signal and the DUT's Altered Reference signal is converted to a low frequency for proper measurement by the VNA's magnitude and phase comparators.
VNA block diagram with frequency conversion :
The conversion oscillator could be a simple fixed frequency oscillator. However, the VNA range would be limited to a narrow spectrum of frequencies. For wide-band operating ranges, both the Reference Signal Generator and the Conversion Oscillator are variable. All VNA's use some type variable frequency oscillator (VFO) as the conversion oscillator. This frequency conversion scheme dictates that the operating range of the VNA will be limited to the bandwidth of both the Reference Signal Generator and the Variable Conversion Oscillator. In this scheme, the Reference Generator and Conversion Oscillator will always be separated in frequency by the value of the frequency of the Comparator circuit. I show a value of 10.7 MHz, since that is the value recommended for the MSA.
In most cases, variable frequency oscillators are limited to an octive of frequency range. In the diagram I show that the Reference Signal Generator will tune from 1000 MHz to 2000 MHz, and the Conversion Oscillator from 1010.7 MHz to 2010.7 Mhz. This gives the VNA an operating range of 1000 MHz to 2000 MHz. But many other ranges are used. Again, as Commercial VNA operating ranges increase, so do the prices!
It must be pointed out that the two conversion mixers and low pass filters will cause a variation in the magnitude and phase of the signals being converted. This is not a problem, since these variations can be "calibrated out" of the final measurement results. (I will elaborate on the calibration technique later). The variations only become problematic if the variations are "short term", meaning, variations occur during the time measurements are taking place.
I must make another, very important point here. The Reference Phase Signal and the DUT Phase signal MUST be the exact same frequency. They will rarely be the same phase, but the frequencies must always be exactly, the same. This requirement is met in the above example by using the same Conversion Oscillator for both the Reference and DUT signal lines. The magnitudes and phases of the two lines will differ, but the two final converted frequencies will be identical.
Example 2, is a Single Conversion scheme for a VNA with an operational range of 1000 MHz to 2000 MHz. But, multiple conversion schemes are usually the norm. Multiple conversion schemes are necessary to create the required Operation range for a VNA. Shown below is a Dual Conversion scheme for a VNA.
The Reference Signal Generator is variable from 0 MHz to 1000 MHz. (Obviously, this is greater than an octive of bandwidth and I will explain how it does this, later). The Reference and DUT Altered Reference signals are upconverted by the variable Conversion Oscillator 1, to a fixed, 1013.3 MHz. Then it is downconverted by the fixed Conversion Oscillator 2. This dual conversion scheme creates a VNA with an operating range of 0 to 1000 MHz.
Up and down converting will create undesired frequencies that are unwanted, and detremental to VNA operation. Therefore, narrow bandwidth filters are included to attenuate the unwanted mixing products. Just as the mixers do, the filters will cause a variation in the magnitude and phase of the signals being converted. These variations are either fixed, or long term (due to temperature changes) and are of no consequence. They can be calibrated out during short term measurements.
In example 3, the Reference Signal Generator operated from 0 MHz to 1000 MHz. The following scheme shows how this is done.
Instead of a stand-alone Reference Signal Generator, a new Conversion Oscillator 3 is added to the circuit and is identical to Conversion Oscillator 1. It is used to create two Reference Signals at one time. Conversion Oscillator 3 and Conversion Oscillator 1 are mixed together to create the Magnitude and the Phase Reference signal, at 10.7 MHz. Conversion Oscillator 3 and Conversion Oscillator 2 are mixed together to generate the Reference signal for application to the DUT. In this case, down conversion produces frequencies from 0 Hz to 1000 MHz. However, for a higher frequency VNA, the up conversion product can be used as a Reference signal for an operating range of 2048 MHz to 3048 MHz.
As the VNA is tuned across it's operating frequency, both Conversion Oscillator 1 and Conversion Oscillator 3 are changed at the same rate. The two Oscillators will always track each other with a 10.7 MHz difference.
The combination of Mixer 3 and Conversion Oscillator 3 constitute the Reference Signal Generator, also referred to as a "Tracking Generator". It's output signal can be named "Reference", in VNA applications, or Tracking Generator Output, in Spectrum Analyzer applications.
This is the scheme used when converting the MSA into a VNA. Not because it is the best scheme, but, because it takes advantage of the Original MSA's signal topology. No substantial modifications to the MSA are required to incorporate this scheme. The MSA already contains the Conversion Oscillator 1, Conversion Oscillator 2, the top two mixers, and the 1013.3 MHz band pass filter. The additional items required to convert the Basic MSA into a VNA are Mixer 3, Mixer 4, Conversion Oscillator 3, Phase Comparator, and a form of Magnitude Comparison.
The Basic MSA does not have a Magnitude Comparator. However, a VNA does not actually require one. In all of the previous diagrams, the Magnitude Comparator is shown as a single block that compares the Magnitude of the Reference Signal to the Magnitude of the DUT Altered Signal. One might assume that it be a requirement that this comparison be taken dynamically. Years ago, when there was no such thing as "digitizing", "memory", or even computers, this was definitely a requirement. However, these days, it is not necessary. The following diagram is a very practical method of Magnitude Comparison.
The previous Magnitude Comparator has been replaced with a block called Magnitude Detector. It is simply an RF power to voltage converter. It's output voltage is converted to a digital word by an analog to digital converter and stored in the memory (digitized). No dynamic comparison is involved, just an absolute power measurement. Older VNA's change frequency versus time, with a smooth, linear response. New VNA's do not. They sweep in a "step" motion. That is, the VNA will move to a particular frequency, stop, make a measurement, store that measurement in memory, and then move to the next frequency. These movements are quite small, and somewhat unnoticeable on a display. If it takes 400 steps to sweep a particular frequency spectrum, then there will be 400 magnitude measurements and 400 memory locations.
If there is a DUT in place, the VNA will measure the DUT Altered Reference Power 400 times and put the results in memory. If the DUT is then replaced with a short circuit, the VNA will measure Unaltered Reference Power 400 times and put the results in memory. We call this Unaltered power, the Reference Magnitude. There are now 800 power measurements in the memory. We can recall the 400 Altered measurements and the 400 Reference Magnitude measurements and compare them to each other with a calculator (ie, the home computer). The difference in magnitude of the Memorized Altered Signal Power and the Reference Magnitude Power becomes the Magnitude Vector. There will be a Magnitude Vector for each step of the VNA sweep. By the way, the memory is not inside the MSA/VNA, it is inside the home computer.
Likewise, there is a Phase measurement digitization for each step of the VNA sweep. When the DUT is in place, there will be 400 Phase Measurements of Altered Phase. When the DUT is replaced with a short circuit, the VNA will measure Unaltered Phase 400 times and put the results in memory. The computer can read the memory at any time, compare the Altered and Unaltered Phase measurements to each other, and calculate the correct Phase Vector for each step of the sweep.
Modification of Basic MSA into a Vector Network Analyzer :
The Basic MSA operates from 0 MHz to 1000 MHz and has a frequency conversion scheme identical to the previous Examples 4, and 5. Some names have changed, but the functions are the same.
As you can see, a large portion of a VNA is already constructed. To keep the discussion simple, I have not shown some amplification and filtering in the Final I.F. path. We need to add only a few devices to the Basic MSA to complete a VNA, as shown in the blue and red sections, below.
In the MSA, the final signal from the last mixer is called the Final I.F. When the MSA/VNA is used as a Spectrum Analyzer, it is just the final intermediate frequency. When used as a VNA, it contains all the frequency components that were previously named, DUT Altered Magnitude and Phase, and DUT Unaltered Magnitude and Phase. In the MSA, the Magnitude Detector is called the Log Detector. The Log Detector will measure absolute signal magnitude. Since the MSA is controlled by a home computer with memory, Magnitude Vector measurements are made using the digitizing method, previously described. The Log Detector also has an R.F. limited output, to supply the DUT Phase Signal to the Phase Comparator.
To create a Reference Signal, to inject into the DUT, a Tracking Generator is added to the Basic MSA. The Tracking Generator (blue section) consists of the Tracking Generator Oscillator (LO 3) and Mixer 3. The Basic MSA now becomes an MSA/TG.
Of course, the MSA/TG does not contain any form of Phase Comparison network. The Phase Modification to the MSA (red section) consists of Mixer 4, a low pass filter, a Phase Detector, and an Analog to Digital Converter. The terms, Phase Comparator and Phase Detector are the same thing, and all my documentation uses the term Phase Detector, or Phase Detector Module, or PDM.
Expanding the MSA into the MSA/TG/VNA
There are three steps for constructing the MSA/TG/VNA.
1. Build the Basic MSA. Go to the Main Web Page for info on the Basic MSA.
2. Add the Tracking Generator (TG) to the Basic MSA. Go to Tracking Generator page for more details. This creates the MSA/TG.
3. Add Phase Modification to the MSA/TG. This creates the MSA/TG/VNA.
If you are converting the Original MSA/TG to an MSA/TG/VNA, go to Original MSA/VNA.
If you are expanding the Basic SLIM MSA into a SLIM MSA/TG/VNA, but have not added the Tracking Generator yet, go to this page for Adding the Tracking Generator to the SLIM MSA. Then, return here to complete the SLIM Phase Modification.
This rest of this web page will detail the expansion of the SLIM MSA/TG into the SLIM MSA/TG/VNA.
The SLIM Phase Modification
The Phase Modification is the addition of the Mixer 4 Module, the Phase Detector Module, and their connections to the SLIM MSA/TG. Some modifications to the SLIM MSA/TG may be necessary and are covered in the next paragraphs. The Mixer 4 Module has the part number, SLIM-MXR-2. It is identical in parts and construction as Mixer 2 in the Basic MSA. The Phase Detector Module has the part number, SLIM-PDM. Click on either part number to go to it's page, for construction details.
The following is the SLIM MSA/TG/VNA Block Diagram, including RF and coaxial connections. Signal and Power connections are shown in the Wiring Diagram, later.
As an option, both DDS modules have spare DDS outputs, not shown in the diagram. One or both can be brought out to the front panel for other experiments.
Modules in SLIM MSA/TG and Mods. required to expand into SLIM MSA/TG/VNA
SLIMs Added for Phase Modification
Additional Coaxial Interconnections for the Phase Modification to the SLIM MSA/TG
Modifications to the SLIM MSA/TG when adding the Phase Mod:
1. The added Phase Detector Module (PDM) requires signal commands from the SLIM Control Board. The SLIM-CB-NV, Control Board, has been built with these connections in place; therefore, no mods are necessary to the Control Board.
2. The added Mixer 4 requires a sample of the MSA's PLO 1 tuning signal, at +7 dBm. It is possible that the SLIM-PLO-1, PLO 1 Module, was built with only one buffer section. If so, the second buffer must be added.
3. Mixer 4 also requires a sample of the MSA/TG's PLO 3 fixed signal, at -4 dBm. PLO 3 was part of the Tracking Generator addition. If it was consructed without the intention of converting to the VNA, it is possible that the SLIM-PLO-3, PLO 3 Module, was built with only one buffer section. If so, the second buffer must be added.
4. The added Phase Detector Module requires the RF Limited signal from the MSA's Log Detector. It is possible that the Log Detector, SLIM-LD-8306, was built without the Limiter circuit. If so, it must be completed.
5. The Phase Detector Module's output must be digitized by the MSA's Analog to Digital Conversion Module. It is possible that the AtoD Module, SLIM-ADC-16 (or SLIM-ADC-12), was built with only one A/D circuit. If so, the second one must be added.
6. The PDM is designed with an internal diplexer/filter on the J1, Reference Input. This diplexer is not needed when used in this VNA expansion. The reason is, PDM's J1 follows Mixer 4. The SLIM Mixer 4 has an internal diplexer/filter. Both are not needed. Therefore, delete L1, L2, R10, C17, and C18. Bypass the L2 solder pads with a wire short.
7. One other modification to the AtoD Module, and it is optional: Remove U1, the 5 volt regulator I.C. Connect a short jumper or ferrite bead from the input pad to the output pad (was U1, pins 3 and 1). This module can receive it's 5 volt operating power directly from the added Phase Detector Module. This modification increases the VNA Phase Measurement accuracy.
Wiring Diagram for the SLIM MSA/TG/VNA
The following diagram consists of all the Power and Control signals for the SLIM MSA/TG/VNA. For RF and coaxial connections, refer to the MSA/TG/VNA Block Diagram, shown previously.
The signal names, internal to the blocks, are general names given to the individual SLIM Modules. The external signal names are assigned to the MSA/TG/VNA, as the SLIMS are integrated.
Layout for the SLIM MSA/TG/VNA
The following layout is proposed for the SLIM MSA/TG/VNA.
This layout is not a requirement, just a suggestion. This is also how the SLIM pwb's are ordered from ExpressPCB. There are 4 panels, each one is 3.8 x 2.5 inches. 3 Panels are required for construction of the Basic SLIM MSA, and the 4th panel is required for the Tracking Generator addition. No other pwb's need to be ordered when adding the Phase Modification to the SLIM MSA/TG. Soon, Cash Olsen will offer this layout as a single panel, containing all 4 sections. It can be constructed without slicing the individual pwb's from the panel. I will update this page as soon as it is ready.
Some Notes on the SLIM MSA/TG/VNA
1. I would like to point out that, the Phase Detector Module responds to the difference of phase between its two input signals. Not amplitude differences. If one input signal has phase noise and the other does not, the PDM will output a noisy product. If both input signals have identical, and correlated phase noise, the PDM output will be clean, since the differential phase noise is "zero". Look back at Example 1, Block Diagram. The Reference Signal Generator supplies an identical signal to the DUT and the Phase Comparator circuit (the phase detector). This signal will contain phase noise. This is because the Reference Signal Generator is not perfect, no oscillator is. If the DUT causes a phase change (and it will), the phase noise from the Reference Signal Generator will be shifted (in the Time Domain) when passing through the DUT. It will not be shifted when entering the Phase Comparator. This causes the phase noise to become uncorrelated, and the phase comparator will differentiate the uncorrelated noise as a noisy output. However, the output signal can be highly integrated (smoothed) with a low pass filter, to reduce measurement noise error. This condition exists in all VNA's and certainly the MSA/VNA. If we could build oscillators without phase noise, this problem would not exist.
There is another form of noise that will enter one port of the PDM, but not the other. And, that is the noise generated by the I.F. Amplifier, and the Log Detector. The Log Detector will limit this noise, by the action of the Limiter Output, but will be noticeable when the Input Signal to the MSA chain is low. This noise is the predominant factor during low level VNA measurements. Still, this VNA has an exceptional dynamic range for a home-brew device, about 80 dB. See the screen plots near the bottom of this page.
2. The Phase Detector Module, PDM, contains a Video Selection Switch that has the provision for 3 positions of filter capacitance (the integrator). As the capacitance becomes larger (in Farads), the noise contribution becomes smaller. However, more capacitance means a longer integration time, which means it takes longer to acquire an accurate Phase Measurement. The integration capacitors, shown in the PDM schematic, are my recommendations, and can be changed at the preference of the builder. As a matter of fact, the 3 position switch could be replaced with a multiposition, rotary switch with multiple capacitors. This would allow more selections of integration times.
3. Some comments about the Master Oscillator. The phase noise contribution discussed in Notes 2. and 3. is considered high frequency noise, and can be filtered out using low pass filtering. There is another, very low frequency Phase Noise component, which cannot be filtered out. This low frequency Phase Noise is determined by the combination of the 3 conversion oscillators in the MSA/VNA. Each one is a Phase Locked Loop system and is "solid as a rock". The "rock" is the Master Oscillator. In time, and over temperature, the Master Oscillator will drift in frequency. It is a slow drift, but does create the low frequency Phase Noises in the conversion oscillators. This low frequency Phase Noise can be renamed Frequency Drift. That is, the 3 conversion oscillators will drift in frequency relative to the frequency drift of the Master Oscillator.
It would sound as though a very accurate, stable, and expensive Master Oscillator is required for a VNA. Stability, is the attribute we are looking for. And, even very inexpensive oscillators can be built with good stability. I will use the following block diagram example 8, with cases, to explain how the Master Oscillator effects VNA measurements.
Consider case 1, the Master Oscillator is exactly 64 MHz. The VNA is commanded to 1000 MHz. LO 1 is at 2013.3 MHz. LO 2 is at 1024.0 MHz. LO 3 is at 2024.0 MHz.
The Reference frequency to the DUT is created by Mixer 3 and is:
LO3 - LO2 = 2024 - 1024 = 1000 MHz.
The Reference frequency goes through the DUT and is converted by the top mixer to:
LO1 - 1000 MHz = 2013.3 - 1000 = 1013.3 MHz.
It is reconverted by the second mixer to be an input to the Phase Comparator:
LO2 - 1013.3 = 1024 - 1013.3 = 10.7 MHz
The other input to the Phase Comparator is the difference mixing of Mixer 4:
LO3 - LO1 = 2024 - 2013.3 = 10.7 MHz.
The two input frequencies to the PDM are exactly the same frequency.
Now let's look at case 2, where the Master Oscillator has drifted by 1000 Hz. Its frequency is now 64.001000 MHz. Even though the VNA is still commanded to 1000 MHz, the conversion oscillators frequencies will change, relative to the Master Clock's frequency change. In paranthesis().
LO1 will change to 2013.331458 MHz (a 31,458 Hz change)
LO2 will change to 1024.016000 MHz (a 16,000 Hz change)
LO3 will change to 2024.031625 MHz (a 31,625 Hz change)
The Reference frequency to the DUT will now be:
LO3 - LO2 = 2024.031625 - 1024.016000 = 1000.015625 MHz (a 15,625 Hz change)
The Reference frequency goes through the DUT and is converted by the top mixer to:
LO1 - 1000.015625 MHz = 2013.331458 - 1000.015625 = 1013.315833 MHz (a 15,833 Hz change)
It is reconverted by the second mixer to be an input to the Phase Comparator:
LO2 - 1013.315833 = 1024.016000 - 1013.315833 = 10.700167 MHz (a 167 Hz change)
The other input to the Phase Comparator is the difference mixing of Mixer 4:
LO3 - LO1 = 2024.031625 - 2013.331458 = 10.700167 MHz (a 167 Hz change)
The two input frequencies to the PDM are exactly the same frequency (10.700167 MHz), even though both are 167 Hz higher than when the Master Oscillator was stable. The PDM is not frequency sensitive enough for this minor frequency change to create a phase measurement error.
The actual frequencies that arrive at the inputs of the PDM are given by the following two formulas:
Signal Frequency, S = LO2-[LO1-(LO3-LO2)]
Phase Frequency, P = LO3-LO1
If any, or all of the LO's change frequency, the differential frequency of the two inputs to the PDM will always be zero. Therefore, the Master Oscillator stability will not have any effect of the quality of the PDM.
However, there are two areas of concern, where the Master Oscillator stability can effect Phase Measurements. One area is within the MSA itself. In previous example blocks, a LPF is shown as the only element in the Final I.F. path. In actuality, a Resolution Band Width Filter (RBWF) is in this path. If the RBWF is a narrow band crystal filter, the phase shift can change quite dramatically, with a frequency change. In Case 2, the final I.F. frequency changed 167 Hz when the Master Oscillator changed 1 KHz. For a crystal filter with a 2 KHz bandwidth, the phase change associated with a 167 Hz change could be as much as 100 degrees. Normally, filters with wider bandwidths have less phase change versus frequency change. Therefore, a wider bandwidth RBWF is preferred when the MSA/VNA is in the VNA Mode.
The second area of concern is the effect that the Master Oscillator stability has on the DUT, itself. All DUTs will have a phase change vs. frequency. After all, that is the purpose of a VNA, to measure this quantity. If you think you are sweeping a DUT around 1000 MHz and the actual sweep frequency is 15 KHz in error (due to Master Oscillator drift in Case 2) the phase shift through the DUT will be in error. Probably not very much, but it is still an error.
Let us recalculate this error by changing the VNA sweep spectrum from the 1000 MHz, shown in example 8, to a lower spectrum, say 50 MHz. If the Master Oscillator has drifted the same 1000 KHz, the actual Reference frequency is not 50.0 MHz, it is:
LO3 - LO2 = 1074.016781 - 1024.016000 = 50.000781 MHz (a 781 Hz change). Again, this Reference Frequency error may not create a large phase error, but, it is still an error that must be minimized.
I have stated before, in other pages, that the absolute frequency of the Master Oscillator is of no concern, when used in the MSA. The software is "told" what the actual Master Oscillator frequency is. The same reasoning holds true when used in the MSA/VNA. The frequencies for the three Local Oscillators are calculated and adjusted for any absolute Master Oscillator frequency error. But short-term drift cannot be calibrated out of the measurement error. Therefore, it is a requirement that the Master Oscillator be stable over a short period of time. I have built several Master Oscillator Modules using very inexpensive integrated crystal oscillators. They all have these "problems" in common. They are never the exact frequency they are labeled. Very close, but not exact. They will change frequency as they warm up or change temperature. They will change frequency if their Vcc supply voltage is changed (pushing). They will change frequency if their load values change (pulling).
All of these problems can be minimized with some proper designing. A very stable supply voltage will minimize pushing. This is why I include an internal Voltage Regulator. Loading the oscillator with stable buffers will minimize pulling. Allowing the temperature of the oscillator to stabilize and remain at a fixed temperature is probably the best design goal. Just a "breath" of air flowing around the oscillator will cause a frequency change. I like the idea of sealing the Master Oscillator to prevent any wind currents from entering the module. Accordingly, wrapping the Master Oscillator Module in some type of "thermal blanket" is very effective. I have used styrofoam "boxes" to enclose a module. I have even used the expandable foam, used to fill gaps around water pipes (available at your local hardware store). You will not see any pictures of this method on my web site, because it will prevent you from seeing the actual module.
I have built several of these "cheapie" oscillators with the design techniques, described. Once the oscillators have been allowed to temperature stabilize (about 30 minutes) the phase noise is confined to very low frequency "warbling". It is usually due to "shot noise" within the oscillator's internal amplifier. This "warbling" is simply the frequency moving a few Hz around a stable point (low frequency jitter). So, the bottom line is, you don't need a super-duper expensive Master Oscillator. But, hey, if you have a super-duper TCXO (temperature compensated crystal oscillator), use it.
Updated 1-18-08. Revise paragraph "The SLIM Phase Modification"
This page is a guide for expanding the MSA, to become a 0 to 1000 MHz RF Vector Network Analyzer, the MSA/TG/VNA. I will restrict this page content to the hardware aspects of a VNA, and not give much discussion on the mathmatic processes of vector analysis. The first portion of this page contains general information for the topology of a generic VNA, and is relevant when constructing either the Original MSA/TG/VNA or the SLIM MSA/TG/VNA. The second portion is specific to the SLIM version.
What is a Vector Network Analyzer?
When an RF signal is applied to a network, such as a filter, amplifier, or transmission line, that signal is altered in magnitude and phase. If the magnitude and phase of the altered signal can be compared to the magnitude and phase of the originating RF signal, the characteristics of that network can be evaluated. The term "magnitude" is just another name for voltage or power. "Phase" is a derivation of time delay versus frequency.
Here is a block diagram of a highly simplified Vector Network Analyzer (VNA):
The network, or component, being tested is called the Device Under Test (DUT). The RF signal that is input to the DUT is the Reference signal. (It is not an absolute necessity that the Reference Signal Generator reside within the VNA, but it is a common practice.) The DUT will alter the Reference signal's two components, Magnitude and Phase. The DUT will change the magnitude component, due to it's resistive natures. It will alter the phase component due to it's reactive natures. These two altered components of the Reference signal are measured by the magnitude and phase comparators within the VNA. The key word, here is "compare".
The DUT Altered Magnitude is compared to the Magnitude of the stable Reference signal by the Magnitude Comparator. The output of the Magnitude Comparator is some value that represents the difference between the voltage, or power, of it's two input signals. This value of differential magnitude is called the Magnitude Vector.
The DUT Altered Phase is compared to the stable Phase of the Reference signal by the Phase Comparator. The output of the Phase Comparator is some value that represents the difference between the phase of it's two input signals. This value of differential phase is called the Phase Vector.
These two Vector values can be a voltage level or a digitized word. Either way, the Processor converts the Vector values to some format that the human operator can use, see and evaluate. Old VNA's only had a visual output, the CRT (Cathode Ray Tube). The old CRT has been replaced with color LCD displays. The new VNA's are totally computer controlled and will output Vector formats in various ways. Since the Vectors are digitized and processed by computers, the data can be converted to any format recognizable to software evaluation programs.
For a VNA to have any value, it must be able to take Magnitude and Phase measurements over a range of frequencies. This is called the Operating Range, or Frequency Range of the VNA. In the previous diagram, the Reference Signal Generator would be a Variable Frequency Signal Generator. A VNA with this scheme would have an operating range equal to the frequency range limitation of the comparators. Comparator circuits do not have an unlimited bandwidth. I hesitate to show actual comparator frequency values, because they vary widely among commercial and home-brew VNA's. Bandwidths from a few MHz to several hundred MHz can be obtained. As bandwidths increase, so does the cost. In the case of the MSA/VNA, the comparator limitation is 8 MHz to 12 MHz. Therefore, Example 1 VNA would have an operating range of 8 MHz to 12 MHz.
All VNA's have a limit to their operating range, and some commercial VNAs will operate higher than 30 GHz. In all cases, the VNA's magnitude and phase comparators will operate at a lower frequency. So, for a VNA to have a high frequency operating range, some type of frequency conversion sheme must be used to convert the Reference and Altered signals to the frequency range of the VNA's comparators.
In the following example, the Reference Signal Generator is a much higher frequency. The Reference signal and the DUT's Altered Reference signal is converted to a low frequency for proper measurement by the VNA's magnitude and phase comparators.
VNA block diagram with frequency conversion :
The conversion oscillator could be a simple fixed frequency oscillator. However, the VNA range would be limited to a narrow spectrum of frequencies. For wide-band operating ranges, both the Reference Signal Generator and the Conversion Oscillator are variable. All VNA's use some type variable frequency oscillator (VFO) as the conversion oscillator. This frequency conversion scheme dictates that the operating range of the VNA will be limited to the bandwidth of both the Reference Signal Generator and the Variable Conversion Oscillator. In this scheme, the Reference Generator and Conversion Oscillator will always be separated in frequency by the value of the frequency of the Comparator circuit. I show a value of 10.7 MHz, since that is the value recommended for the MSA.
In most cases, variable frequency oscillators are limited to an octive of frequency range. In the diagram I show that the Reference Signal Generator will tune from 1000 MHz to 2000 MHz, and the Conversion Oscillator from 1010.7 MHz to 2010.7 Mhz. This gives the VNA an operating range of 1000 MHz to 2000 MHz. But many other ranges are used. Again, as Commercial VNA operating ranges increase, so do the prices!
It must be pointed out that the two conversion mixers and low pass filters will cause a variation in the magnitude and phase of the signals being converted. This is not a problem, since these variations can be "calibrated out" of the final measurement results. (I will elaborate on the calibration technique later). The variations only become problematic if the variations are "short term", meaning, variations occur during the time measurements are taking place.
I must make another, very important point here. The Reference Phase Signal and the DUT Phase signal MUST be the exact same frequency. They will rarely be the same phase, but the frequencies must always be exactly, the same. This requirement is met in the above example by using the same Conversion Oscillator for both the Reference and DUT signal lines. The magnitudes and phases of the two lines will differ, but the two final converted frequencies will be identical.
Example 2, is a Single Conversion scheme for a VNA with an operational range of 1000 MHz to 2000 MHz. But, multiple conversion schemes are usually the norm. Multiple conversion schemes are necessary to create the required Operation range for a VNA. Shown below is a Dual Conversion scheme for a VNA.
The Reference Signal Generator is variable from 0 MHz to 1000 MHz. (Obviously, this is greater than an octive of bandwidth and I will explain how it does this, later). The Reference and DUT Altered Reference signals are upconverted by the variable Conversion Oscillator 1, to a fixed, 1013.3 MHz. Then it is downconverted by the fixed Conversion Oscillator 2. This dual conversion scheme creates a VNA with an operating range of 0 to 1000 MHz.
Up and down converting will create undesired frequencies that are unwanted, and detremental to VNA operation. Therefore, narrow bandwidth filters are included to attenuate the unwanted mixing products. Just as the mixers do, the filters will cause a variation in the magnitude and phase of the signals being converted. These variations are either fixed, or long term (due to temperature changes) and are of no consequence. They can be calibrated out during short term measurements.
In example 3, the Reference Signal Generator operated from 0 MHz to 1000 MHz. The following scheme shows how this is done.
Instead of a stand-alone Reference Signal Generator, a new Conversion Oscillator 3 is added to the circuit and is identical to Conversion Oscillator 1. It is used to create two Reference Signals at one time. Conversion Oscillator 3 and Conversion Oscillator 1 are mixed together to create the Magnitude and the Phase Reference signal, at 10.7 MHz. Conversion Oscillator 3 and Conversion Oscillator 2 are mixed together to generate the Reference signal for application to the DUT. In this case, down conversion produces frequencies from 0 Hz to 1000 MHz. However, for a higher frequency VNA, the up conversion product can be used as a Reference signal for an operating range of 2048 MHz to 3048 MHz.
As the VNA is tuned across it's operating frequency, both Conversion Oscillator 1 and Conversion Oscillator 3 are changed at the same rate. The two Oscillators will always track each other with a 10.7 MHz difference.
The combination of Mixer 3 and Conversion Oscillator 3 constitute the Reference Signal Generator, also referred to as a "Tracking Generator". It's output signal can be named "Reference", in VNA applications, or Tracking Generator Output, in Spectrum Analyzer applications.
This is the scheme used when converting the MSA into a VNA. Not because it is the best scheme, but, because it takes advantage of the Original MSA's signal topology. No substantial modifications to the MSA are required to incorporate this scheme. The MSA already contains the Conversion Oscillator 1, Conversion Oscillator 2, the top two mixers, and the 1013.3 MHz band pass filter. The additional items required to convert the Basic MSA into a VNA are Mixer 3, Mixer 4, Conversion Oscillator 3, Phase Comparator, and a form of Magnitude Comparison.
The Basic MSA does not have a Magnitude Comparator. However, a VNA does not actually require one. In all of the previous diagrams, the Magnitude Comparator is shown as a single block that compares the Magnitude of the Reference Signal to the Magnitude of the DUT Altered Signal. One might assume that it be a requirement that this comparison be taken dynamically. Years ago, when there was no such thing as "digitizing", "memory", or even computers, this was definitely a requirement. However, these days, it is not necessary. The following diagram is a very practical method of Magnitude Comparison.
The previous Magnitude Comparator has been replaced with a block called Magnitude Detector. It is simply an RF power to voltage converter. It's output voltage is converted to a digital word by an analog to digital converter and stored in the memory (digitized). No dynamic comparison is involved, just an absolute power measurement. Older VNA's change frequency versus time, with a smooth, linear response. New VNA's do not. They sweep in a "step" motion. That is, the VNA will move to a particular frequency, stop, make a measurement, store that measurement in memory, and then move to the next frequency. These movements are quite small, and somewhat unnoticeable on a display. If it takes 400 steps to sweep a particular frequency spectrum, then there will be 400 magnitude measurements and 400 memory locations.
If there is a DUT in place, the VNA will measure the DUT Altered Reference Power 400 times and put the results in memory. If the DUT is then replaced with a short circuit, the VNA will measure Unaltered Reference Power 400 times and put the results in memory. We call this Unaltered power, the Reference Magnitude. There are now 800 power measurements in the memory. We can recall the 400 Altered measurements and the 400 Reference Magnitude measurements and compare them to each other with a calculator (ie, the home computer). The difference in magnitude of the Memorized Altered Signal Power and the Reference Magnitude Power becomes the Magnitude Vector. There will be a Magnitude Vector for each step of the VNA sweep. By the way, the memory is not inside the MSA/VNA, it is inside the home computer.
Likewise, there is a Phase measurement digitization for each step of the VNA sweep. When the DUT is in place, there will be 400 Phase Measurements of Altered Phase. When the DUT is replaced with a short circuit, the VNA will measure Unaltered Phase 400 times and put the results in memory. The computer can read the memory at any time, compare the Altered and Unaltered Phase measurements to each other, and calculate the correct Phase Vector for each step of the sweep.
Modification of Basic MSA into a Vector Network Analyzer :
The Basic MSA operates from 0 MHz to 1000 MHz and has a frequency conversion scheme identical to the previous Examples 4, and 5. Some names have changed, but the functions are the same.
As you can see, a large portion of a VNA is already constructed. To keep the discussion simple, I have not shown some amplification and filtering in the Final I.F. path. We need to add only a few devices to the Basic MSA to complete a VNA, as shown in the blue and red sections, below.
In the MSA, the final signal from the last mixer is called the Final I.F. When the MSA/VNA is used as a Spectrum Analyzer, it is just the final intermediate frequency. When used as a VNA, it contains all the frequency components that were previously named, DUT Altered Magnitude and Phase, and DUT Unaltered Magnitude and Phase. In the MSA, the Magnitude Detector is called the Log Detector. The Log Detector will measure absolute signal magnitude. Since the MSA is controlled by a home computer with memory, Magnitude Vector measurements are made using the digitizing method, previously described. The Log Detector also has an R.F. limited output, to supply the DUT Phase Signal to the Phase Comparator.
To create a Reference Signal, to inject into the DUT, a Tracking Generator is added to the Basic MSA. The Tracking Generator (blue section) consists of the Tracking Generator Oscillator (LO 3) and Mixer 3. The Basic MSA now becomes an MSA/TG.
Of course, the MSA/TG does not contain any form of Phase Comparison network. The Phase Modification to the MSA (red section) consists of Mixer 4, a low pass filter, a Phase Detector, and an Analog to Digital Converter. The terms, Phase Comparator and Phase Detector are the same thing, and all my documentation uses the term Phase Detector, or Phase Detector Module, or PDM.
Expanding the MSA into the MSA/TG/VNA
There are three steps for constructing the MSA/TG/VNA.
1. Build the Basic MSA. Go to the Main Web Page for info on the Basic MSA.
2. Add the Tracking Generator (TG) to the Basic MSA. Go to Tracking Generator page for more details. This creates the MSA/TG.
3. Add Phase Modification to the MSA/TG. This creates the MSA/TG/VNA.
If you are converting the Original MSA/TG to an MSA/TG/VNA, go to Original MSA/VNA.
If you are expanding the Basic SLIM MSA into a SLIM MSA/TG/VNA, but have not added the Tracking Generator yet, go to this page for Adding the Tracking Generator to the SLIM MSA. Then, return here to complete the SLIM Phase Modification.
This rest of this web page will detail the expansion of the SLIM MSA/TG into the SLIM MSA/TG/VNA.
The SLIM Phase Modification
The Phase Modification is the addition of the Mixer 4 Module, the Phase Detector Module, and their connections to the SLIM MSA/TG. Some modifications to the SLIM MSA/TG may be necessary and are covered in the next paragraphs. The Mixer 4 Module has the part number, SLIM-MXR-2. It is identical in parts and construction as Mixer 2 in the Basic MSA. The Phase Detector Module has the part number, SLIM-PDM. Click on either part number to go to it's page, for construction details.
The following is the SLIM MSA/TG/VNA Block Diagram, including RF and coaxial connections. Signal and Power connections are shown in the Wiring Diagram, later.
As an option, both DDS modules have spare DDS outputs, not shown in the diagram. One or both can be brought out to the front panel for other experiments.
Modules in SLIM MSA/TG and Mods. required to expand into SLIM MSA/TG/VNA
| Block
Diagram Items |
Part
Number of SLIM, as Block Diagram Item |
Modification
to SLIM Required? |
| Mixer 1 | SLIM-MXR-1 | No |
| DDS 1 |
SLIM-DDS-107
rev A |
DDS Spare Out
is Optional |
| PLO 1 |
SLIM-PLO-1 | Both outputs
used |
| Mixer 2 | SLIM-MXR-2 | No |
| PLO 2 |
SLIM-PLO-2 | Both outputs
used |
| I.F. Amplifier | SLIM-IFA-33 | No |
| Final Xtal Filter | SLIM-MCF-106 | No |
| Log Detector | SLIM-LD-8306 | Limiter
Output used |
| Master Oscillator | SLIM-MO-64 | All 3 buffers
used |
| Control
Board |
SLIM-CB-NV | Both +20v DC
Converters used |
| AtoD
Converter, or |
SLIM-ADC-16 | Both A/Ds
used, Regulator optional |
| AtoD Converter | SLIM-ADC-12 | Optional 12
Bit, prefer 16 Bit |
| DDS 3 |
SLIM-DDS-107
rev A |
DDS Spare Out is Optional |
| PLO 3 |
SLIM-PLO-3 | Both outputs active |
| Mixer 3 |
SLIM-MXR-1 | Mxr
1 and 3 are identical |
SLIMs Added for Phase Modification
| Block
Diagram Items |
The
SLIM, to Replace the Block Diagram Component |
Modification
to SLIM Required? |
| Mixer 4 | SLIM-MXR-2 | Mxr 2 and 4
are identical |
| PDM |
SLIM-PDM | Delete
internal Diplexer |
Additional Coaxial Interconnections for the Phase Modification to the SLIM MSA/TG
| Signal
Name |
From |
To |
Recommended Cable Type |
Approximate Length, Inches |
| LO1 | SLIM-PLO-1, J2 | SLIM-MXR-4,
J1 |
RG-085 |
3.5 |
| LO3 | SLIM-PLO-3, J2 | SLIM-MXR-4, J3 | RG-085 | 1.6 |
| Phase
Reference |
SLIM-MXR-4, J2 | SLIM-PDM, J1 | RG-085 or RG-188 | 1.6 |
| Limited
IF |
SLIM-LD-8306, J3 | SLIM-PDM, J2 | RG-085
or RG-188 |
1.8 |
| Phase
Volts |
SLIM-PDM, J3 | SLIM-ADC-16, J2 | RG-085 or RG-188 | 1.8 |
Modifications to the SLIM MSA/TG when adding the Phase Mod:
1. The added Phase Detector Module (PDM) requires signal commands from the SLIM Control Board. The SLIM-CB-NV, Control Board, has been built with these connections in place; therefore, no mods are necessary to the Control Board.
2. The added Mixer 4 requires a sample of the MSA's PLO 1 tuning signal, at +7 dBm. It is possible that the SLIM-PLO-1, PLO 1 Module, was built with only one buffer section. If so, the second buffer must be added.
3. Mixer 4 also requires a sample of the MSA/TG's PLO 3 fixed signal, at -4 dBm. PLO 3 was part of the Tracking Generator addition. If it was consructed without the intention of converting to the VNA, it is possible that the SLIM-PLO-3, PLO 3 Module, was built with only one buffer section. If so, the second buffer must be added.
4. The added Phase Detector Module requires the RF Limited signal from the MSA's Log Detector. It is possible that the Log Detector, SLIM-LD-8306, was built without the Limiter circuit. If so, it must be completed.
5. The Phase Detector Module's output must be digitized by the MSA's Analog to Digital Conversion Module. It is possible that the AtoD Module, SLIM-ADC-16 (or SLIM-ADC-12), was built with only one A/D circuit. If so, the second one must be added.
6. The PDM is designed with an internal diplexer/filter on the J1, Reference Input. This diplexer is not needed when used in this VNA expansion. The reason is, PDM's J1 follows Mixer 4. The SLIM Mixer 4 has an internal diplexer/filter. Both are not needed. Therefore, delete L1, L2, R10, C17, and C18. Bypass the L2 solder pads with a wire short.
7. One other modification to the AtoD Module, and it is optional: Remove U1, the 5 volt regulator I.C. Connect a short jumper or ferrite bead from the input pad to the output pad (was U1, pins 3 and 1). This module can receive it's 5 volt operating power directly from the added Phase Detector Module. This modification increases the VNA Phase Measurement accuracy.
Wiring Diagram for the SLIM MSA/TG/VNA
The following diagram consists of all the Power and Control signals for the SLIM MSA/TG/VNA. For RF and coaxial connections, refer to the MSA/TG/VNA Block Diagram, shown previously.
The signal names, internal to the blocks, are general names given to the individual SLIM Modules. The external signal names are assigned to the MSA/TG/VNA, as the SLIMS are integrated.
Layout for the SLIM MSA/TG/VNA
The following layout is proposed for the SLIM MSA/TG/VNA.
This layout is not a requirement, just a suggestion. This is also how the SLIM pwb's are ordered from ExpressPCB. There are 4 panels, each one is 3.8 x 2.5 inches. 3 Panels are required for construction of the Basic SLIM MSA, and the 4th panel is required for the Tracking Generator addition. No other pwb's need to be ordered when adding the Phase Modification to the SLIM MSA/TG. Soon, Cash Olsen will offer this layout as a single panel, containing all 4 sections. It can be constructed without slicing the individual pwb's from the panel. I will update this page as soon as it is ready.
Some Notes on the SLIM MSA/TG/VNA
1. I would like to point out that, the Phase Detector Module responds to the difference of phase between its two input signals. Not amplitude differences. If one input signal has phase noise and the other does not, the PDM will output a noisy product. If both input signals have identical, and correlated phase noise, the PDM output will be clean, since the differential phase noise is "zero". Look back at Example 1, Block Diagram. The Reference Signal Generator supplies an identical signal to the DUT and the Phase Comparator circuit (the phase detector). This signal will contain phase noise. This is because the Reference Signal Generator is not perfect, no oscillator is. If the DUT causes a phase change (and it will), the phase noise from the Reference Signal Generator will be shifted (in the Time Domain) when passing through the DUT. It will not be shifted when entering the Phase Comparator. This causes the phase noise to become uncorrelated, and the phase comparator will differentiate the uncorrelated noise as a noisy output. However, the output signal can be highly integrated (smoothed) with a low pass filter, to reduce measurement noise error. This condition exists in all VNA's and certainly the MSA/VNA. If we could build oscillators without phase noise, this problem would not exist.
There is another form of noise that will enter one port of the PDM, but not the other. And, that is the noise generated by the I.F. Amplifier, and the Log Detector. The Log Detector will limit this noise, by the action of the Limiter Output, but will be noticeable when the Input Signal to the MSA chain is low. This noise is the predominant factor during low level VNA measurements. Still, this VNA has an exceptional dynamic range for a home-brew device, about 80 dB. See the screen plots near the bottom of this page.
2. The Phase Detector Module, PDM, contains a Video Selection Switch that has the provision for 3 positions of filter capacitance (the integrator). As the capacitance becomes larger (in Farads), the noise contribution becomes smaller. However, more capacitance means a longer integration time, which means it takes longer to acquire an accurate Phase Measurement. The integration capacitors, shown in the PDM schematic, are my recommendations, and can be changed at the preference of the builder. As a matter of fact, the 3 position switch could be replaced with a multiposition, rotary switch with multiple capacitors. This would allow more selections of integration times.
3. Some comments about the Master Oscillator. The phase noise contribution discussed in Notes 2. and 3. is considered high frequency noise, and can be filtered out using low pass filtering. There is another, very low frequency Phase Noise component, which cannot be filtered out. This low frequency Phase Noise is determined by the combination of the 3 conversion oscillators in the MSA/VNA. Each one is a Phase Locked Loop system and is "solid as a rock". The "rock" is the Master Oscillator. In time, and over temperature, the Master Oscillator will drift in frequency. It is a slow drift, but does create the low frequency Phase Noises in the conversion oscillators. This low frequency Phase Noise can be renamed Frequency Drift. That is, the 3 conversion oscillators will drift in frequency relative to the frequency drift of the Master Oscillator.
It would sound as though a very accurate, stable, and expensive Master Oscillator is required for a VNA. Stability, is the attribute we are looking for. And, even very inexpensive oscillators can be built with good stability. I will use the following block diagram example 8, with cases, to explain how the Master Oscillator effects VNA measurements.
Consider case 1, the Master Oscillator is exactly 64 MHz. The VNA is commanded to 1000 MHz. LO 1 is at 2013.3 MHz. LO 2 is at 1024.0 MHz. LO 3 is at 2024.0 MHz.
The Reference frequency to the DUT is created by Mixer 3 and is:
LO3 - LO2 = 2024 - 1024 = 1000 MHz.
The Reference frequency goes through the DUT and is converted by the top mixer to:
LO1 - 1000 MHz = 2013.3 - 1000 = 1013.3 MHz.
It is reconverted by the second mixer to be an input to the Phase Comparator:
LO2 - 1013.3 = 1024 - 1013.3 = 10.7 MHz
The other input to the Phase Comparator is the difference mixing of Mixer 4:
LO3 - LO1 = 2024 - 2013.3 = 10.7 MHz.
The two input frequencies to the PDM are exactly the same frequency.
Now let's look at case 2, where the Master Oscillator has drifted by 1000 Hz. Its frequency is now 64.001000 MHz. Even though the VNA is still commanded to 1000 MHz, the conversion oscillators frequencies will change, relative to the Master Clock's frequency change. In paranthesis().
LO1 will change to 2013.331458 MHz (a 31,458 Hz change)
LO2 will change to 1024.016000 MHz (a 16,000 Hz change)
LO3 will change to 2024.031625 MHz (a 31,625 Hz change)
The Reference frequency to the DUT will now be:
LO3 - LO2 = 2024.031625 - 1024.016000 = 1000.015625 MHz (a 15,625 Hz change)
The Reference frequency goes through the DUT and is converted by the top mixer to:
LO1 - 1000.015625 MHz = 2013.331458 - 1000.015625 = 1013.315833 MHz (a 15,833 Hz change)
It is reconverted by the second mixer to be an input to the Phase Comparator:
LO2 - 1013.315833 = 1024.016000 - 1013.315833 = 10.700167 MHz (a 167 Hz change)
The other input to the Phase Comparator is the difference mixing of Mixer 4:
LO3 - LO1 = 2024.031625 - 2013.331458 = 10.700167 MHz (a 167 Hz change)
The two input frequencies to the PDM are exactly the same frequency (10.700167 MHz), even though both are 167 Hz higher than when the Master Oscillator was stable. The PDM is not frequency sensitive enough for this minor frequency change to create a phase measurement error.
The actual frequencies that arrive at the inputs of the PDM are given by the following two formulas:
Signal Frequency, S = LO2-[LO1-(LO3-LO2)]
Phase Frequency, P = LO3-LO1
If any, or all of the LO's change frequency, the differential frequency of the two inputs to the PDM will always be zero. Therefore, the Master Oscillator stability will not have any effect of the quality of the PDM.
However, there are two areas of concern, where the Master Oscillator stability can effect Phase Measurements. One area is within the MSA itself. In previous example blocks, a LPF is shown as the only element in the Final I.F. path. In actuality, a Resolution Band Width Filter (RBWF) is in this path. If the RBWF is a narrow band crystal filter, the phase shift can change quite dramatically, with a frequency change. In Case 2, the final I.F. frequency changed 167 Hz when the Master Oscillator changed 1 KHz. For a crystal filter with a 2 KHz bandwidth, the phase change associated with a 167 Hz change could be as much as 100 degrees. Normally, filters with wider bandwidths have less phase change versus frequency change. Therefore, a wider bandwidth RBWF is preferred when the MSA/VNA is in the VNA Mode.
The second area of concern is the effect that the Master Oscillator stability has on the DUT, itself. All DUTs will have a phase change vs. frequency. After all, that is the purpose of a VNA, to measure this quantity. If you think you are sweeping a DUT around 1000 MHz and the actual sweep frequency is 15 KHz in error (due to Master Oscillator drift in Case 2) the phase shift through the DUT will be in error. Probably not very much, but it is still an error.
Let us recalculate this error by changing the VNA sweep spectrum from the 1000 MHz, shown in example 8, to a lower spectrum, say 50 MHz. If the Master Oscillator has drifted the same 1000 KHz, the actual Reference frequency is not 50.0 MHz, it is:
LO3 - LO2 = 1074.016781 - 1024.016000 = 50.000781 MHz (a 781 Hz change). Again, this Reference Frequency error may not create a large phase error, but, it is still an error that must be minimized.
I have stated before, in other pages, that the absolute frequency of the Master Oscillator is of no concern, when used in the MSA. The software is "told" what the actual Master Oscillator frequency is. The same reasoning holds true when used in the MSA/VNA. The frequencies for the three Local Oscillators are calculated and adjusted for any absolute Master Oscillator frequency error. But short-term drift cannot be calibrated out of the measurement error. Therefore, it is a requirement that the Master Oscillator be stable over a short period of time. I have built several Master Oscillator Modules using very inexpensive integrated crystal oscillators. They all have these "problems" in common. They are never the exact frequency they are labeled. Very close, but not exact. They will change frequency as they warm up or change temperature. They will change frequency if their Vcc supply voltage is changed (pushing). They will change frequency if their load values change (pulling).
All of these problems can be minimized with some proper designing. A very stable supply voltage will minimize pushing. This is why I include an internal Voltage Regulator. Loading the oscillator with stable buffers will minimize pulling. Allowing the temperature of the oscillator to stabilize and remain at a fixed temperature is probably the best design goal. Just a "breath" of air flowing around the oscillator will cause a frequency change. I like the idea of sealing the Master Oscillator to prevent any wind currents from entering the module. Accordingly, wrapping the Master Oscillator Module in some type of "thermal blanket" is very effective. I have used styrofoam "boxes" to enclose a module. I have even used the expandable foam, used to fill gaps around water pipes (available at your local hardware store). You will not see any pictures of this method on my web site, because it will prevent you from seeing the actual module.
I have built several of these "cheapie" oscillators with the design techniques, described. Once the oscillators have been allowed to temperature stabilize (about 30 minutes) the phase noise is confined to very low frequency "warbling". It is usually due to "shot noise" within the oscillator's internal amplifier. This "warbling" is simply the frequency moving a few Hz around a stable point (low frequency jitter). So, the bottom line is, you don't need a super-duper expensive Master Oscillator. But, hey, if you have a super-duper TCXO (temperature compensated crystal oscillator), use it.
