N2PK Vector Network AnalyzerAn inexpensive Vector Network Analyzer of a Lab quality
A simple narrow-band transverter extends N2PK VNA frequency range to 450MHzThe Vector Network Analyzer developed by Paul Kiciak, N2PK, covers frequency range of 0.05...60MHz and has proven accuracy over these frequencies. It is desirable, however, in many cases, to perform measurements at higher frequencies, especially on 2m and 70cm (VHF/UHF) bands which are of great interest for many amateurs. The transverter presented here can do frequency translation in a narrow-band fashion and its working range is only limited by the frequency range of the mixers and bandpass filters used.
The idea of extending the VNA frequency range by using up- and downconverter with same LO came from Bill Carver, W7AAZ. Paul, N2PK, then presented a working concept and accompanying documentation. Unfortunately, there was no board layout available at the time which would allow for replication. I started this project with a goal in mind to develop a compact transverter which could be fairly easy repeated by other N2PK VNA owners with little or no access to the equipment.
Finished VNA transverter enclosed in a Hammond 1455C801 extruded aluminum box, pictured close to actual size. Right angle PCB mount SMA connectors interface the N2PK VNA to a DUT.
Original N2PK's schematics was adopted as the basis for the transverter because it was well characterized and prototyped. It was then modified a few times in attempts to achieve a compromise between performance, simplicity and repeatability. But in the end it was found that these design objectives in fact intersect, and the final version of the schematics did not deviate much from the original one presented by Paul. There was a bit of change introduced to the RF components base to slightly improve some parameters and satisfy SMT type layout design objective.
Specification is given for the UHF version because it was harder to build and test, but the VHF one is no different in size or construction and has even better parameters.
|Frequency range:||40...50MHz low side, 430...450MHz high side|
|Dynamic range (Transmission mode):||90dB|
|Dynamic range (Reflection mode):||70dB|
|Upconverter output power drift:||+/-0.1dB over 1 hr|
|Downconverter output power drift:||+/-0.1dB over 1 hr|
|400MHz LO frequency stability:||+/-5Hz after 1hr of warmup at room temperature|
|DUT port in/out impedance:||50 Ohm|
|DUT Out port Return Loss (To DUT):||25dB|
|DUT In port Return Loss (From DUT):||21dB|
|Input power to DUT:||-20dBm @50 Ohm|
|Power consumption:||9V @130mA|
|Size:||3-1/4 x 2-1/8 x 7/8'' ( 85x55x23mm)|
|UHF Transverter Bill of Materials||VHF Transverter Bill of Materials|
Theory of operation
In short what the transverter does, it takes the VNA's RF DDS signal as input and upconverts it to a higher frequency using a mixer and local oscillator. The upconverted signal then passes through a bandpass filter to get rid of the unwanted mixer image and drives the Device Under Test (DUT). The thru or reflected signal from the DUT is then downconverted back to the VNA working frequency range. The transverter is only interfaced to the VNA and does not require a connection to the computer.
|Quadrature signal necessary for the VNA to work is preserved over the frequency translation. The image to the left shows the transverter output at 40MHz transverter input. Alternating 0/90 degree test signal was upconverted to 440MHz, then routed back to the downconverter via a DUT thru. The resultant 40MHz low-side output signal has its 90 degree shift preserved.|
The transverter is able to extend the VNA frequency range to as high as 1GHz, but does not cover all frequencies continuously. First, theoretical maximum frequency span is limited to approximately 50MHz which N2PK VNA is able to cover. Second, the narrow-band bandpass filter sets limits on the working range. Say, a 70cm bandpass filter tuned to 425...455MHz would ultimately define the transverter frequency range. In order to use the transverter over some other frequency range, an appropriate bandpass filter and Local Oscillator need to be built.
My interests (and I believe many other's) would not normally go beyond 2m and 70cm amateur bands, that is why for sake of portability I designed the PCB to have an on-board filter. The PCB layout is universal and by installing an appropriate LO oscillator and BPF the board implements either 2m or 70cm transverter. For simplicity and performance the transverter PCB was designed to accommodate only a single on-board bandpass filter, but the builder may opt to use an external BPF to have more flexibility. I tried the idea of switchable dual-band on-board filter, but gave up because at these frequencies any complication of the PCB layout and use of additional components increase stray signals and cross-talk because of capacitive coupling. This would also increase drift of transverter parameters over time and temperature. Therefore, a decision was made to produce a single-band layout with possibility of use of an external LO/BPF combination. For example, to cover both his 2m and 70cm needs, the builder could use 2 PCBs and build 2 separate transverters with on-board filters (very compact solution), or use one PCB with unpopulated LO/BPF parts and use an external LO signal and an external BPF (more flexible but bulky solution), allowing for tuning to any frequency up to the mixer limit).
Minicircuits SBL-2LH part was chosen as the frequency mixer for the transverter because of its better port-to-port isolation specs. A good isolation was required to prevent leakage of the input transverter signal to its output via mixer LO ports, bypassing the up- and downconversion paths. Such leakage could very possibly reduce the dynamic range of the VNA in transverter mode, because the input RF signal in this case would only partially pass through the DUT, and the transverter output would have a sum of the useful and leaked signal.
Because the transverter is of a narrow-band design, separate Los are needed for VHF and UHF. An easiest option is a use of an integrated sine wave oscillator with low phase noise. However finding such an oscillator may be not a trivial task. A fairly high drive level is needed by the mixers (+5...10dBm), and in order for the mixers to reduce in the output the images and their combinations a sinewave type of oscillator would be needed.
The commercially made integrated oscillator manufactured by Connor-Winfield was chosen from very limited number of options. It delivers +12dBm into 50 Ohm load. A drawback is that these kind of oscillators are custom made and have to be ordered with 2 months delivery time from the manufacturer who requires a minimum order. The advantage is the oscillator is compact, hermetically sealed and shielded by design. The LO oscillator specification is this:
|Supply voltage:||+5V +/-5%|
|Supply current:||50mA max|
|Oscillator frequency:||400MHz for UHF, 180MHz for VHF|
|Frequency stability by specification:||+/-50ppm|
|Frequency stability measured after 1hr of warmup at room temperature:||+/-5Hz over 1hr of observation|
|Total harmonic distortion:||-35dBm|
|Temperature range:||0 to +70C|
The second and third LO harmonics are well outside of the Transverter working range and suppressed by the BPF right after they exit the mixers. The oscillator phase noise chart provided by the manufacturer can be found here.
The 70cm bandpass filter is an SMT version of Paul N2PK's UHF filter described in his transverter documentation. The PCB presented here has a universal layout allowing to implement either 2m or 70cm bandpass filter on-board. The BPF bandpass and the LO frequency ultimately define the transverter's working frequency range. If an external filter is preferred, the associated components on the PCB should not be populated, and the BPF in/out coaxial connectors should be soldered in and used instead. Because the SMT trimmer capacitors have much lower Q then, say, a piston type trimmers, overall insertion loss of the on-board BPF is higher then of its standalone counterpart, such as the BPF described in Paul N2PK's transverter documentation. The 144MHz on-board filter was measured to have 6dB insertion loss, and the 440MHz one had 14dB insertion loss. In this design higher SMT filter insertion loss was compensated by increased gain of the amplifier stage that followed the BPF.
|The 70cm on-board BPF response. A 400MHz LO was used in 70cm transverter, and the closest 430MHz image located at 370MHz is well suppressed by the BPF. The graticule center is at 440MHz, vertical scale is 10dB/ , horizontal scale 20MHz/ . The filter's flat top covers 430...450MHz.|
Two MMIC amplifier stages were used in the design, one in up- and the other in downconverter path. To simplify the BOM both stages use same low noise 26dB SGA4586 amplifiers. This gain figure differs from the original Paul N2PK's schematics in upconversion path because more gain was heeded to compensate for the added on-board BPF insertion loss.
Return loss bridge
The return loss bridge senses the reflected power when the transverter is in Reflection mode. Quality of the reflection bridge is important for accurate VNA measurements. Several designs of the return loss bridge were evaluated for the best possible performance. The candidates included a Minicircuit power splitter ADP2-1W, a Minicircuit directional coupler ADC10-4, and a resistive bridge with MA/COM ETC1-1-13 1:1 balun. Table below summarizes their own in-circuit return loss and other parameters measured with my HP spectrum analyzer and ADI logarithmic detector at 440MHz for each one of them. When taking RL measurements on a port all other ports were terminated with 50 Ohm loads.
|MCL splitter ADP2-1W||MCL coupler ADC10-4||MA/COM balun ETC1-1-13|
|DUT port RL||12 dB||7 dB||23 dB|
|RF port RL||13 dB||7 dB||18 dB|
|Open/Short||1.6 dB||10 dB||0.3 dB|
|Directivity||18 dB||21 dB||24 dB|
The data collected ruled out the MCL ADC10-4 directional coupler from the design list as a definite loser on return loss and Open/Short tests. Out of 2 candidates left the MA/COM balun-based resistive bridge showed better parameters, especially on DUT port return loss and Open/Short test. It was then chosen as the return loss bridge for the transverter project.
Downconverter output low-pass filter
|The output low-pass filter has 3dB cut-off frequency of 52MHz and leaves behind all unwanted mixer images and the LO residue. Attenuation at 60MHz is 27dB, increasing up to 80dB at 80MHz. The screenshot to the left has center frequency of the graticule set to 60MHz, 10db/ vertical and 5MHz/ horizontal scales.|
|Transverter PCB top. Seen are the shielded bandpass filter, mixers, and the Local Oscillator. The filter sections are additionally separated with walls inside the shield to reduce mutual coupling between sections.|
|Transverter PCB bottom. The BPF trimmer capacitors are at mid top. The reflection bridge at the bottom and was built on the MA/COM RF balun - based resistive reflection bridge.|
Testing of this little transverter required engagement of a serious equipment. The HP 8565A spectrum analyzer shown below was the leader on the bench. Without one many problems such as parasitic oscillation of the MMIC amplifiers in 2...4GHz range would have gone unnoticed and caused phantom problems.
Because of working at VHF/UHF, beside using the appropriate test gear, having quality RF supplies and tools was also a must. At the beginning the items such as 50 Ohm loads, some of BNC nickel-plated RF connectors, leaky jumper cables and others that performed poorly at the test frequencies were identified and removed from use. Whenever possible, semirigid, hardline and double-shield coax jumpers were used for interconnections. Silver-bearing solder was used throughout the project. Some of the accessories, such as high value attenuators and VNA calibration standards where made at home using SMA connectors and SMT chips. Where it was not possible to make, the needed accessories where purchased from Minicircuits. The equipment used in the project included:
|HP 8565A Spectrum analyzer 0.01...22GHz|
|HP 8568B Spectrum analyzer 100Hz...1.5GHz|
|HP 5334A Frequency counter|
|HP 5327B Counter/Timer/DVM|
|HP 3456A Digital Voltmeter|
|HP 8656B Signal Generator 0.01...990MHz|
|HP 8444A Tracking Generator 0.5...1500MHz|
|Tektronix 2467B 400MHz Oscilloscope|
Most of the equipment was cross-checked and agreed to each other. Based on that it was then assumed the equipment is holding valid calibration.
To validate proper functionality of the transverter a series of tests were conducted. For most of the tests the goal was to set them in such way that the results would be known in advance to see if the transverter performs to expectations.
|Transmission mode. UHF Transverter noise floor. Dynamic range is better than 100dB throughout UHF band|
|Transmission mode. UHF Transverter noise floor with new CDS sampling mode supported in software. Dynamic range extends to unprecedent 115dB.|
|Reflection mode. A length of Alpha Wire RG58 A/U coax open at far end. The test shows the proper behavior of the complex reflection coefficient detection, which follows circular path centered at 1:0. Radius of the circle equals magnitude of the reflection coefficient.|
|This test shows angle of the reflection coefficient in Reflection mode. A length of Alpha Wire RG58 A/U coax w/short at far end was tested. Test shows good linearity in phase detection.|
|Transmission mode sweep of a 20cm long shunt stub w/short at far end made from a rigid coaxial jumper. Assuming velocity of propagation 0.69 and adding 4.7cm of connectors and termination lengths its electrical length adds to 33.68cm, which would make half wavelength at 445MHz. Because the far end had a short and a half lambda transmission line repeats its far end impedance, we should expect the transmission function to dip at this frequency. The VNA indeed showed a dip at 442MHz.|
|Reflection mode. Scan of my UHF J-Pole 5/8th wave antenna. It was tuned a couple of years ago to the middle of the UHF band with an SWR meter and at the time showed SWR less than 2 over the entire band. Now I can confirm that with the Transverter. Red line is Return Loss, blue line is SWR.|
|Side to side transmission test of a UHF cavity filter, VNA Transverter vs HP 8565A spectrum analyzer. I tried to scale the pictures to bring them close to size of each other. Both scans were swept over 434...454MHz range and vertical scale span is 40dB top to bottom for both.|
Next test was conducted on extreme loads of 2.5 Ohm, 501 Ohm and 1 kOhm. Table below shows their DC resistance in Ohms measured with HP 3456A digital voltmeter, and the correspondent real part of the load impedance values measured with the transverter at UHF. The test was done at 7 samples/s sampling rate. The transverter values shown averaged over the 430...450MHz working range.
|HP 3456A DVM||2.476||501.6||1000.0|
|VNA + Transverter||2.464||499.2||977.2|
For SWR=20 the test showed accuracy better than 1% for low impedance loads and better than 3% for high impedance loads. At this high frequencies and SWRs this is simply an amazing result for a piece of homebrewed equipment.
Calibration standards and test loads
For the transverter we need a set of Calibration Standards to calibrate the VNA/Transverter tandem. Normally they are the Open, Short and Load standards. The existing VNA calibration standards can be used for the transverter calibration, if they were built properly with RF approach in mind.
|Calibration standards. Left to right are the Open, Short and Load standards built using SMA female connectors. The Load standard is based on two 100 Ohm 0.1% resistors of 0603 SMT type.|
The Open standard connector has its center pin shaved down to the base. Same is the Short standard, after soldering the short which is a disk made of copper foil. The Load standard has excess of the center pin removed as well after soldering the resistors. At frequencies that high any extra millimeter of length would introduce angle error in measurements of the Reflection Coefficient.
|Test loads used to evaluate the transverter. SMT 0805 and 0603 type resistors were used. You do not have to build these ones, but if you need one of some specific value, follow the construction concept.|
The Test Loads were used to validate and spec the transverter performance. They are based on same type of SMA connectors as the Calibration Standards to eliminate errors introduced by differences between different connector part numbers.
After the prototype transverter was built, I realized that this is possible to use its downconvert path as a downconverter to monitor the amateur UHF band with an HF general coverage receiver or scanner. The converter connects between the antenna and the receiver (the RL bridge to the antenna, transverter VNA output to the receiver) , which is switched to FM mode and tuned to 40...50MHz. The signals from 440...450MHz amateur UHF band will then be available for listening on the HF radio. The transverter was tested in this mode and its functionality in this mode was confirmed. Because the downconvert amplifier is of a wide-band type, for best result a UHF bandpass filter could be connected between the antenna and the RL bridge. Your external 70cm BPF transverter filter would work perfectly for that purpose.
Another potential application of the downconverter would be waveform monitoring of the output signal of a UHF transmitter/repeater. In this application the transverter bridge input could be connected to the coupled port of a directional coupler inserted inline with the TX antenna. The downconverter would then translate the signal down to 40...50MHz range where it can be analyzed with low frequency range oscilloscopes or spectrum analyzers. To prevent damage to the transverter the power developed at the coupled port of the directional coupler should not exceed -20dBm into 50 Ohm with the highest power applied to the antenna.
Software to use with the Transverter
There are several freeware packages that fully support the Transverter. You specify the Transverter LO frequency (400MHz for the UHF Transverter, 180MHz for the VHF one), and they plot charts with the frequency axis labels corrected, as was shown in the screenshots above. They also set the VNA DDS frequency automatically based on if the LO is below or above the frequency of interest. The DOS-based program package was written by Paul N2PK. It creates data files that later can be plotted with my IVManLite program. This package only supports the Parallel port interface. Dave Roberts, G8KBB, developed a nice Windows-based program called myVNA, which supports both parallel port and USB interface. Tomas Baier, DG8SAQ, has a software called VNWA that he wrote for VNA of his own design, but that software also supports N2PK VNA hardware. Currently there is no problems with the Transverter software.
- - DOS package by Paul Kiciak, N2PK
- - myVNA with USB support by Dave Roberts, G8KBB
- - VNWA by Tomas Baier, DG8SAQ
PCB and parts kits availability
Please check this link for availability of the Transverter kits. The PCB kits include the universal VHF/UHF Transverter PCB (two PCBs will be needed to construct VHF and UHF transverters), and the hard-to-get parts:
- The PCB, Minicircuits mixers and power splitter, 400MHz Connor-Winfield Local Oscillator, MA/COM RF balun, and Coilcraft air coils for the BPF.
Rest of parts will need to be purchased directly from DigiKey by the builder. Cost of this DigiKey package is approximately $30 not including the box and SMA connectors. The BOM and PCB kit does not include the RF connectors. The PCB can take either SMA or SMB type and decision is left on the builder what type and kind of connector to use. Depending on the enclosure, you may decide to use either straight or right angle connector, or even the edge type one (some filing work may be needed to fit this one in). Optionally, if a spacious box RF tight box is used, you could install panel mounted bulkhead RF connectors of any type (including BNC or N ones) and solder the pigtail jumpers directly to the PCB.
Please follow this link for the Transverter construction information.
A versatile transverter unit was built and characterized, extending the N2PK VNA frequency range. The work lasted more than a year and was aimed to fill the gap in needs of the N2PK VNA owners to perform measurements on VHF and UHF amateur bands.
I would like to thank the people, who helped me to make this page available, for their contribution. Here is the names list in alphabetical order:
- Bill Carver, W7AAZ, for his initial idea of quadrature signal frequency translation;
- Harold Johnson, W4ZCB, for reviewing the manuscript and confirming the technicalities;
- Paul Kiciak, N2PK, for reviewing the article and providing valuable comments;
- Yuri Romanov, VE3XB, for arranging access to an HP 8568B Spectrum analyzer to measure the 400MHz LO phase noise.