This page describes the assembly and construction of my N2PK VNA which is based on the January, 2004 Group Build kit.
The time investment at this point has been about 20 hours. I spent approximately 9 hours soldering all of the passive components. 6 hours were spent soldering the active components. This step included making voltage checks as the parts were added. The remaining 5 hours were consumed in drilling holes in the enclosures, mounting connectors, and painting/labeling the enclosures.
My costs above and beyond the boards and their parts have been a little under $100 (USD). This includes the two, die-cast aluminum enclosures, and the connectors. The connectors are the costly items. The connector family choice, and shopping techniques, will have a large impact on the price of the connectors.
At this point, I do not have a final power supply solution. I've left open the door for using an external battery pack, or an external AC-input power supply. During initial testing and checkout, I've used a good quality wall-wart supply, which is not the best long term solution, since the quality of regulation and reduction of noise/ripple will be low. Once the dust settles, I suspect I'll spend some additional money to obtain a better AC-input supply.
Here is a picture of the VNA I assembled with the companion T1-6T bridge. Please click on the picture for a larger view.
The left side of the enclosure includes the power jack, On-Off switch, and DB-25 connector. A standard printer cable is used to connect the VNA to the printer port of the computer.
All of the other external connections are made on the top lid. There are two sets of three in-line connectors. One set has the RF DDS output in the center, and the outer connectors represent the Detector #1 and Detector #2 RF inputs. The other set of three connectors has the LO DDS output in the middle, and is flanked by the Detector #1 and Detector #2 LO inputs. As shown in the picture, the T1-6T bridge spans the RF DDS output and Detector #1 RF input. A jumper, made from N connector plumbing connects the LO DDS output to the Detector #1 LO input.
The current VNA includes a single detector, based around a 24-bit ADC (analog to digital converter). This is wired as Detector #1 on my enclosure. Paul, N2PK, is working on a second detector design, which will have the property that it can make faster measurements, at some acceptable reduction in accuracy. The current ADC produces approximately 7 conversions per second. Two conversions are needed to make one (reflection) impedance measurement. As a result, the VNA can make a little less than 4 complete measurements per second. For many applications, this is more than fast enough. In other cases, usually involving sweeping a frequency range, a higher conversion rate is desirable. For example, consider sweeping the (USA) 80 meter band, with 1 KHz frequency spacing. 500 measurements would be needed. At 4 measurements per second, that would take 125 seconds, or a little more than 2 minutes. Paul's VNA documents, available from his web site, include more information on the existing and planned detectors.
The spacing between the center connectors and both matching outside connectors is the same. This means that devices such as bridges or jumpers, can be used to select either Detector #1 or #2, depending upon which pair of connectors is used. While you can simply pick a single detector, and make that the only option, that would preclude the VNA from making certain measurements (such as IP3) where two detectors are needed.
In this configuration, the VNA is ready to make reflection impedance measurements through the female N connector on top of the bridge.
Most of the assembly time was spent soldering the parts to the board. There are approximately 150 separate parts that must be attached. Most are two-terminal resistors or capacitors. The active devices are mainly 8-pin SO-8 (1.27 mm pitch) packages. The hardest parts to solder are the two DDS chips. They are in SSOP-28 (0.65 mm pitch) packages..
I followed the order of assembly suggested in the N2PK VNA Build Notes. The VNA Build Notes were originally created by Ian, G3SEK. They are part of the VNA Build Info available on the N2PK web site The basic plan is to first install all of the passive components on the bottom of the board, then install all of the passive components on the top of the board. Power can be applied at that point, and there should be no smoke or overheating, and expected voltage readings should be found at various points. From that point, the active components are added. The order of addition begins with the chips that convert and regulate voltages. These can be checked as added by measuring their output voltages. The process winds through the parts, ending up with the 28-pin DDS chips, which, to me, present the greatest assembly challenge. By that point, however, your SMT soldering skills should be pretty good, and you should be ready for the final soldering. Don't consume caffeine if it makes you nervous. I thought that the assembly notes made a lot of sense, and I would strongly recommend that they be used.
While following the suggested assembly order, I also referred to printouts of the top and bottom of the board. These pictures include part annotations, and are part of the N2PK build information. I also had my kit documentation, and the 50 labeled zip bags. When soldering the passive components, I would try to find several of the same part value in a local area and do them as a single batch. This reduced the opening and closing of the bags. Dropping a part, or confusing two parts, can be a disaster.
Here is a top-side picture of my board, with some slight optical distortion due to the macro lens on my camera.
At least 4 coaxial cables must attach to the main board, coming in from the bottom side. By the way, they are not shown in the previous picture, which was taken before the cables were attached. Following the suggestions in the N2PK documentation, I cut short lengths of 1/8" brass tubing, and soldered them into the through holes placed at every coax cable attachment location. The tubes are nearly flush with the top side of the board, and extend almost 1/4" above the bottom side of the board. The center conductor of the coaxial cable is put through the tube, on its way to the top side, and the braid of the coax can be soldered to the tube itself, making a low-cost, all-soldered, connection solution. This process is described in the N2PK build information.
My small modification to this process was to use a second short section of tubing cut from 3/16" diameter stock. This second tube is threaded on the coax prior to installation. After the braid is placed over the smaller tube, the larger tube is brought down over the braid, where it makes a snug fit over the braid. For testing purposes, especially if you think that you might need to remove the coax, it is not necessary to solder the braid - the outer tube will hold it tight.
Small diameter copper or brass tubing can be obtained from most hobby stores, or, from a well-stocked hardware store. The tubing can be easily cut with a pipe cutter designed for small diameter tubes.
The connector choice can be difficult. Many of the initial VNA implementations were built around SMA connectors. Given their relatively low cost, and high performance, they are an excellent choice for what I would call an indoor VNA. I tend to be more interested in antenna-related measurements, and I've been locating most of my antennas outside lately, so I was looking for a little more durability and strength.
The most appropriate choice for me seemed to be the tried and true N connector.
Although I had a few N connectors of various types in the junk box, I needed a number of new connectors to complete the VNA. After some looking around, I found I could obtain everything that I needed from RF Parts Company. Companies such as Pasternack Enterprises have an incredible range of products, but I found their prices a little high. I suspect that you could shave a dollar or two off of the price of each connector at a ham fest, but, the selection is always an unknown.
RF Parts also sells the RG-316 Teflon coax cable which is used with some of the connectors I selected. RG-316 is the Teflon version of the popular RG-174, which is a small diameter cable often used within equipment enclosures to interconnect subassemblies. I used RG-316 in my VNA.
Here is a picture of the N connectors which I used in my VNA. All of these connectors are available from RF Parts, and I will use their part numbers in my description. As always, you can click on the picture for a larger view.
The top row shows two 90 degree adapters (RFN-1012-1), and a male to male barrel adapter (RFN-1014-1). When mated together, the result is a male to male rigid jumper which is used to transfer the LO DDS output to the desired detector. This rigid jumper sets the center to center spacing of the holes on the top lid of the VNA box for the LO DDS connectors. Since there might be some variation between these connectors, I'll keep my spacing a secret. Mine was a little more than 2.75".
The left two connectors on the lower row are female (RFN-1021-5) and male (N-MGM) connectors which feature small mounting flanges. These are used on the T1-6T bridge enclosure, where space is at a premium.
The last connector, in the lower right corner of the picture, is the female bulkhead connector (RFN-1022-8) which I used on the top of the VNA enclosure. The connector has the desirable property that it is installed from the inside to the outside of the panel, so that it can be unscrewed, and taken out through the inside of the cabinet. This means that the connections do not need to be removed in order to remove the connector. The flange mount connectors, in contradistinction (I never get to use that word), mount from the outside into the cabinet.
The bulkhead connector is a crimp-style connector. A small pin (under the connector in the picture) is soldered to the end of the coaxial cable. The pin is inserted into the end of the connector, and makes a sliding contact with the center pin. The braid is dressed around the outside of the tube holding the center conductor. A metal cylinder, acting as a form of ferrule (I don't use that word much either), is brought down around the braid, trapping it in place. The final step uses a special hex crimping tool to lock the outer cylinder in place. For some reason, which is so unlike me, I broke down and purchased a ratcheting crimper tool. I did not purchase the fancy one with interchangeable dies available from RF Parts. I went with a less expensive version from Jameco. This connector is specified to work with RG-174, RG-188, or, RG-316 cable. I found RG-178 simply too small, and it could not be used.
Needless to say, if these crimped connections prove unreliable over time, the consistent accuracy of my VNA will be compromised. A solder connection is preferable to a sliding connection, but I did not find an attractive soldered alternative. In this regard, SMA connectors are king, since they are small enough that they can be directly soldered to PC boards.
I decided to use die-cast aluminum enclosures. They will stand up to any abuse short of a nuclear weapon, and the walls are thick enough so that they can be tapped to accept bolts, if needed. Die-cast boxes make fine heat sinks, and have good RF shielding properties.
The size of the VNA enclosure was dictated by considering the PC board size and the connector layout. Additional room was also reserved for a future second detector. I ended up with a case which is 7.6" X 4.3" X 2.4". The case is available from Circuit Specialists (their 03-125C), and Jameco (their 11973). This case is apparently not make by Hammond, perhaps the leading name in die-cast enclosures. A Hammond part which is just about the same size is the 1590R1. Hammond products are available from Mouser (where they are very expensive).
The female bulkhead connectors are cut with a pair of flats so that they will fit into a non-round hole. This keeps the connector from rotating if the nut and lock washer become loose, and, you can hold the connector in place with an open-end wrench while tightening the nut. I really wanted to use that feature, but I had no way to easily cut that shape with the tools I have. This is especially a problem when considering that the connectors also had to be in a precise location so that the connectors on the bridge would mate with the VNA. I decided to drill full-sized holes, and then create what amounted to a non-round shape by using several pieces of shim stock on the inside of the top cover. This shim stock was riveted in place with aluminum rivets. The stock cuts across a part of the mounting hole, providing the needed shape. I've heard a report that these female connectors, with the sharp edges where the flats are cut, can scrape the plating off of the mating male connector.
The next two pictures show interior views of the VNA. When they were taken, the power and data cables were in place, but the 4 coaxial cables were not yet installed.
The left picture shows the VNA board, which is mounted on 1.5" standoffs. It is mounted over the 4 connectors which accept cables from the board. The other two connectors will go to the second detector board. There is clearly a lot of room in the box for another board. The shims which take the connector holes out of round can be seen in both pictures.
The right picture shows the handful of power supply components which are located in the lower corner of the box, using point to point wiring. They are described in the next section.
The enclosure for the T1-6T bridge is smaller, and is ideally the same internal size as the bridge PCB, so that the ground plane naturally extends to contact the enclosure and the connectors. The best enclosure I could find was the Hammond 1590A. The combination of the bridge board and the enclosure was not a perfect fit, however. The board was not tall enough to span from one side of the box to another, and the 2.5" spacing between the RF input and detector output was a little tight - the connectors were too close to the edge of the enclosure. For these reasons, a recent layout of the bridge board, which will be used on the March, 2004 Group Build, was modified to fit conveniently into the Hammond 1590A enclosure. Thanks to Todd Nichols and Paul, N2PK, for the layout work.
Here is a picture of the earlier version bridge board in the enclosure. I used brass shim stock to increase the size of the board ground plane so that it would make contact with the enclosure. The outer set of screws on the male connectors are right up against the corner posts in the enclosure. The new layout will bring the bottom two connectors a little closer together to avoid the corner posts, and the board will be taller, to span the entire inside height of the box.
I am curious as to the impact in accuracy due to bridge construction differences. I plan on building another bridge from the next batch of boards, and compare results between the two.
Even the Hammond 1590A is not perfect, since the sides have a slight bevel. This means that the top female DUT (device under test) connector is not perpendicular to the top of the VNA. It is off of perpendicular by a few degrees.
There are some more pictures of this general construction approach on the N2PK VNA page. In particular, the units made by G3SEK and G4PMK.
The N2PK documentation defines the power requirements of the VNA as +5 VDC at 310 mA (max), and +12 VDC at 25 mA (max). In order to achieve and preserve accuracy, a high quality supply should be used. A design is presented by N2PK in his documentation which has both quality and protection.
For outdoor use, I wanted to be able to run the VNA off of batteries. Combined with my laptop computer, this would create a highly portable VNA.
Battery operation forces a consideration of the required voltage and current. If a diode was added to the +12 VDC feed to provide protection against a reverse polarity connection, the required input voltage would rise to nearly 13 volts. This would be around the limit for 10 AA cells in series. I'm really happy with the NiMH AA batteries which have reached a power capacity of 2200 mAH. I use them in my digital camera, portable radio, MFJ-269 analyzer, and several other gizmos around the house. I have a drawer full of them at all times. They are inexpensive, especially when bought at ham fests. If the VNA input voltage could be reduced, using a reasonable number of these batteries would be a great solution.
A second reason to lower the input voltage is to reduce the gap between it and +5 VDC. Although there are a range of devices which will turn the battery voltage into +5 VDC, most will tend to spill power as part of the conversion. When batteries are involved, this is just a waste of energy. A +5 VDC regulator will have some minimum input voltage which is required to produce an output of +5 VDC. Supplying more volts on the input (past some safety margin) only wastes power.
As it turns out, the +12 VDC input to the VNA is used to generate +5 VDC and -5 VDC voltage levels which are then used in the detector portion of the circuit. N2PK spent some time looking at the spec. sheets and determined that the +12 VDC input could fall to around 7.75 VDC, without causing a problem. This is great news for battery operation. If a diode is added to provide reverse polarity protection, the minimum input voltage rises to around 8.5 VDC. This is more than enough to drive a 5 volt regulator. Eight AA batteries in series will provide around 9.6 volts. The output will drop as the batteries discharge. It seems to me that a pack of eight rechargeable NiMH batteries can supply power for the VNA.
The power supply circuit tucked in the corner of my VNA enclosure is nothing more than a diode, followed by a 7805T regulator. The TO-220 case is bolted to the bottom of the box, which acts as a heat sink. I added filter and bypass capacitors to the input and output sides of the regulator. The input voltage can range between 9 to 14 volts DC. This is a conservative range.
The 7805 is not known for it's quality, just popularity. I will probably replace it as I get more familiar with using the VNA. At the time, I had a number of them in the junk box.
The DDS chips are specified as consuming around 100 mA. This means that the two DDS chips consume nearly 2/3 of the current from the 5 VDC supply. The chips have a power down mode, where the current drops to nearly zero. In the case of battery operation, it is highly desirable to be able to control the power down mode of the chips, so that they can be turned off when idle. On other pages I've described our VNAccess VNA driver library. This library exposes the power down mode on the DDS chips. Using this feature will help extend the life of the batteries.
I measured the current draw at the input to my VNA. Using the gnat software, I could control the power down mode of each chip. I used a stable 12 VDC supply for the test. The following table shows the current consumed by the board as a function of the power down state of the DDS chips.
VNA Current Draw @ 12 VDC Input
|Test Condition||Current Draw (mA)|
|Both DDS chips and ADC being changed/read at 7.11 Hz||290.8|
|Both DDS chips running, but not changed, no ADC readings||290.3|
|Both DDS chips reset||229.0|
|RF DDS chip in power down mode (LO DDS running)||206.6|
|RF and LO DDS chips both in power down mode||122.3|
I measured a maximum current draw of approximately 291 mA, which is very close to the value reported by N2PK, 310 mA. The last entry in the table is most interesting. If both DDS chips are placed into power down mode, the current draw drops to 122.3 mA, a 58 percent reduction in current.
This suggests that for battery operation, and to extend the life of the batteries, it is useful to consider using power down mode. This function would need to be under the control of the software.
Although soldering surface mount parts was new to me before this project, it turned out to be easy to work with them, so long as I had the right tools. I was fortunate in that I got all of the parts in the right place, and avoided any solder bridges. The VNA worked the first time I applied power. I suspect that if I would have needed to spend hours fixing assembly errors on the board it would not have been so much fun. Then, it would have been a lot more like how it felt when I went to college. There, I spent years trying to fix all of my mistakes.
Any web page where I can use the words montage, contradistinction, and ferrule, must be a good web page.
Back to my Experimentation Page
Back to my Amateur Radio Page