N2PK VNA - Not your Dad's grid dip meter

History of my implementation of the N2PK VNA

If you have not heard about N2PK VNA before, I recommend you to read through a set of comprehensive materials nicely documented and categorized, which could be found on Paul N2PK's WEB site. From there you can follow other links to get even more information. If then you decide you want to build it, you are welcome to come back here, as I will be posting my experience with my unit.

Paul's WEB site also has information how to subscribe to the VNA mailing list, so that you could keep yourself informed about possible changes, upgrades, troubleshooting techniques, part replacements etc. Once you decide to build, you can sign up for a group order to buy hard-to-get parts.

Since the VNA was first introduced, there has been some changes done to the schematics, which were not that easy to implement, because the board layout stayed unchanged. Because of this and some other reasons (Paul is currently working on extended functionality that will be coming soon in a form of a S-Parameters Test Set), I decided to take the challenge and make a board layout that would reflect all changes done to date, and allow for easier expansion of the unit once an update becomes available. As a result, a set of board revisions was produced, each having its advantages. For now I decided to stick with only one of them, which both Paul and I believe is the most appropriate. Many thanks to Paul, N2PK, for taking time to review the layouts and providing valuable comments. Note that this layout is only a rework of the original Paul's design, I only manipulated the components placement and made some (well, this is what I personally think they are) improvements.

The original VNA board, that could be ordered, if needed, from ExpressPCB, was designed to have a slow ADC and a single detector. Once Paul released an update for a faster ADC, and announced an S-parameter Test Set will probably be coming, many builders have added a second board for the fast detector to their projects. The second board is in fact the same board as the 1st one, but has only the detector and power supply components mounted. Additionally, the builders had to take care of splitting the output of the Local Oscillator DDS to feed two detectors, which required having one more board for the power splitter and additional cost. I decided to place both detectors on a same board, and add a second filter for the LO DDS. Because earlier Paul pointed out placing the second filter in close proximity to the RF DDS was found to cause higher stray coupling from the RF DDS to this output port, than to the first output port, I took additional measures, which, hopefully, will reduce or fully eliminate this problem. What was done is the RF DDS was rotated 90 degree counter-clockwise, as well as its output transformer T110, and all associated components. The output of the RF DDS anti-alias filter was diverted away from the LO DDS filters to the opposite side of the board.

The board size is 6.3'' x 2.6''. It was designed to fit standard enclosures - many of the Hammond aluminum die-cast products are sized 16cm (6.3'') in one dimension. If this dimension is the width of the front panel, then the board simply slides inside the enclosure from the front using the enclosure's internal slots. Such enclosure would still have plenty of room behind the VNA board left to mount a DB-25 parallel port connector and a power supply. And here is a 3D picture of the board:

You can see the second detector to the right from the first one. Its bus interface IC and the parallel port connector were rotated 90 degree clockwise to reduce the overall board size. Note the flip-flop IC U130 was also rotated 90 degree CCW, and together with the rotated RF DDS IC this resulted in more natural interconnections between the critical high-speed components, and shorter traces. Some other changes were also introduced, such as placing the voltage test points for each supply voltage level, and SMA type RF connectors for each port, which probably many of the builders like myself would find useful. You can see them on a close-up of the DDS section below:

The layout was made with the fast detector ADC (Analog Devices LTC2440) in mind for both detectors. All the corresponding electrical connections were added and the required parts placed on the board in a normal manner. One would not need to run onboard wires or lift IC pins - everything has been taken care of in this design. For the LO DDS circuits, because now both of its outputs are being utilized, there is no need in a lossy off-board power splitter.

Another advantage of this board design (at least this is what I think) is a lower overall components cost because of the lower parts count for the second detector as oppose to using two original boards. This is because the power for the second detector is supplied by the first detector via the on-board traces. Not only this reduces the number of components, but it also helps in keeping the output of the both detectors symmetrical because any variations in the power supply parts or ambient temperature affect both detectors in a same manner. Also the ADCs have their reference voltage supplied by a single source, which improves the channels symmetry even more.

In general, here is some of the board features that come to my mind right now (I may forget some, because I worked on it multiple times):

• Dual fast ADC
• Dual LO DDS filtered output
• SMA or SMB type RF connectors
• Much bigger ground plane area in the DDS section, which I hope also will help in better heat dissipation for the DDS ICs
• The second detector's +5D and the digital ground are to be supplied externally through J300. This was done intentionally to keep the digital and analog ground separate
• Spare bus interface inputs (4 in total) are available through U360 in the second detector to deliver more control signals to the board, if needed
• Functional layout - all needed signals aligned along the same edge of the board
• High speed PCB design techniques were applied

When the work was coming to the end, I realized I was still missing the reflection bridges. However, after careful evaluation of different options, and how the device could be used in normal activities, I decided not to place the bridges onto the VNA board. This is mainly because the device would lose its versatility. It is much more preferable, I believe, to have the RF output and Detectors inputs on the front panel, and use the external bridges. In this case they can be whatever the user wants, not necessarily the Minicurcuits T1-6T ones. Having the signals on the front panel also gives me endless possibilities in producing different testing configurations.

Nevertheless, I decided to make a reflection bridge board, too. The bridge schematics was borrowed from the VNA documentation. Here is the result:

The board has multiple vias along the edges and traces to reduce the ground plane resistance. The bottom of the board has tinned edges free of the solder mask in case I need to solder the board directly to the enclosure, or through a finger stock. The board was designed to fit the Hammond 1590A die-cast aluminum enclosure, and the idea was borrowed from Greg W8WWV's page. Refer to his nice article on the bridge construction techniques.

This board is sized 3'' x 1.3'' and can take either N-, UHF-, or SMA/SMB-type end launch receptacles. Based on Greg's experience with the 1590A enclosure, the spacing between connectors was made 2 inch. To fully equip a dual detector VNA, two of these reflection bridges would be needed.

I am glad I eventually found time to put this page together and submit my order to the manufacturer, though I still need to buy some components such as an oscillator and the Minicircuit transformers. Once I receive the boards, I will update this page with more pictures. This is not that easy to allocate a window for a manufacturer to produce a single board, so this is going to take at least a couple of weeks, maybe more. Anyways, I am looking forward to completing my build and start using this nice tool right away-too many things are lined-up on my experimentation table to be checked or tuned

Once this board layout proved to work, I'll see if I can order more boards for others. If you want one for yourself right now (as I said, I will be waiting for the assembly), drop me an email and I'll add your order to the existing one. In fact, if you want to assemble by yourself, you can finish your board well before me. Thanks for reading and have a great day!

-Update February 1st. - To fulfill multiple request from the builders, who would like to try the design, a wish list for the first group order is now open. If you want to participate in this group buy, send me an email and indicate if you are interested in a VNA board only, or you want a complete kit which is 1 VNA board and 2 bridge boards. So far the cost for a complete kit has stabilized below USD70 mark. Note the order will be made only after the prototypes proved to be working, so that the group order builder would not suffer. Rough date estimate for the 1st order is end of February, 2006. This is when the order will be placed. It will then take another 2 weeks to manufacture.

-Update February 6th. - I am preparing the Bill Of Materials (BOM) for the dual detector board. It will also include a few parts needed for two reflection bridges. The first prototype boards are scheduled to be received earlier next week.

-Update February 10th. - A hot air station is already on its way to my house. I'd need it anyways sooner or later for my other projects. This morning a trip is planned to a local components distributor to pick up a syringe of a solder paste. I decided not to go with the manufacturer for the assembly because of whole bunch of requirements they put. They also use water soluble flux, which is not acceptable for this design. The reason is the board has multiple vias under the SMD pads. If such flux gets trapped inside during soldering, it will corrode the vias with time and may lead to phantom failures later. My advice to you - do not use water soluble flux with any board that has vias under the SMD. Use a No-Clean one.

-Update February 14th. - The prototype boards have been received and look good. Both the VNA and reflection bridge boards have 2-side solder mask, and the VNA board has 2-side component print. The boards look even better then their 3D-image above. I got two sets, and one set was immediately sent to the States to a group member who kindly offered to share the risk of assembling a board from the prototype run. A little bit later I'll post a few pictures of the received boards and the enclosure ideas for both the VNA and bridge boards. Stay tuned.

-Update February 15th. - The prototype boards at a glance look OK in general. I only spotted the silk printed over the SMA center pads, and this will be fixed in the next run.

The bridge board easily fits the Hammond 1590A die-cast enclosure. The corners of the board need to be slightly shaved to make it sink down to the center of the enclosure.

This is amazing how close the actual board appearance is to one generated by the 3D software higher on this page. Just look and compare!

The board traces are 50 Ohm transmission lines, and as long as the board FR4 material allows, should be usable up to VHF. Specifically because of this the traces and the copper along the trace edge are free of the solder mask.

I have a punch that makes accurate holes of the right diameter for the N-connectors. By the next update I'll prepare a few pictures of the metal work.

My hot air station has stuck at the border waiting for clearance from the Customs. Should arrive this Thursday, if no more delays. On the weekend I can then start assembling the main board.

-Update February 17th. - Read the enclosure article here.

-Update February 19th. - My hot air station has arrived, and over the weekend I finished the board bottom - see below. Next week will be spent on assembling the top and, hopefully, powering the board up.

-Update February 20th. - No connection with drugs, simply soldering.

Overall quality of soldering is not too bad.

-Update February 22th. - The board is almost complete, needs only the header connectors. The top side took me three to four times as longer to complete as the board bottom.

The small pitch ICs (AD9851 and LTC2440 ones) were really painful to mount. Could not avoid using a desoldering wick to remove the solder bridges. All other ICs came out just fine.

As far as I can tell, there was no problem fitting the parts in (I was expecting some of them may bump into each other). Like with the original layout, the transformers leads had to be trimmed in order to fit the footprint on the board.

The final countdown to the Moment of Truth will kick off tomorrow.

-Update February 23th. - The DDS section is up and running. No problems were found and it worked from the first try. Monitoring the output on a scope, I played with the frequency settings from 0.1 to 50MHz. Both RF and LO DDS seem to be OK, and the latter produces a dual filtered signal of an equal amplitude.

-Update February 24th. - Both ADC detectors have been brought up. Detector one was easy to test with either N2PK DOS-based ADC program, or Greg Ordy's Windows Gnat application. Detector two seems to work also, but both applications crash as soon as trying to access it. I believe there is a problem with the Windows parallel port driver UserPort working under Win2000. The system generates an error log saying the application did not respond to the port status change (which is the second LTC2440 ADC output). Next time will try to find a Win98 system.

-Update February 26th. - The VNA is functional and I am testing it with Paul N2PK's test programs, which recognize both detectors. Simultaneous reflection and transmission test sweeps are available with the dual detector system. Paul has suggested a few additional tests that are specific for this board layout, and I am on my way to set them up.

-Update February 27th. - Windows software Exeter, VNA4WIN and DOS based utilities by Paul, N2PK, work with the new board. Exeter and Paul's software can utilize the second detector. Below is a screenshots of reflection and transmission tests of a 2-pole crystal filter, done close to the upper limit of the instrument, at 56MHz.

 

From now on I will be working on evaluating the detectors noise floor, as Paul advised. Shown below is a screenshot of the Detector 2 noise floor. The VNA board with no enclosure or any grounding, laying on the bench, surrounded by the switching power supplies and overhead daylight bulbs showed -110dB noise up to 50MHz. Nice.

 

-Update March 7th. - Read the Noise Test Results article here.

-Update March 8th. - Learn about building a reflection bridge for your measurements here.

-Update March 10th. - Go back to the Noise Test Results article and get yourself updated on the latest news on that.

-Update March 14th. - I have completed a set of the detectors noise floor tests and will proceed to finalizing the layout and ordering the boards for those who want to use or further test them.

-Update March 22th. - View the last prototype board pictures and learn about the things discovered during testing.

-Update March 25th. - Important information about building your dual detector VNA board. Documentation, the BOM, hints and tips.

-Update March 26th. - See some sample screen shots done at different times.

-Update April 2nd. - The first manufacturing order for the VNA kits has been placed with the delivery date April 20th or earlier. The kit includes 1 dual detector VNA board (revision 4.1 with the separate power supplies for each detector), and 2 reflection bridge boards. The kit cost is $68 if payed by PayPal, $65 if by the International Money Order. Personal checks are acceptable only from Canadian buyers - foreign checks are not cashable in Canada, sorry. This is the cost of the PCBs, and cost of shipping will be determined individually some time later. I will be working with the buyers to determine the best shipping option based on their preferences.

I will still be able to increase the order size up to Friday, April 7th, if required. If you want to order your kit now, just send me an email to the address below. If you have your VNA assembled on the classic board, but missing the reflection bridge, let me know, and I will include your order to the bridge PCB pool. The bridge boards are $11 each if payed via PayPal, $10 otherwise. You can have as many bridge boards as you need, in case you want to have in your shack a few of them built around different types of RF connectors. Since this is a generic purpose device, you could also use it for other projects. The reflection bridge board was designed to be used in the Hammond 1590A die cast aluminum enclosure. Go to the DigiKey web site, search for "Hammond 1590A", and pick one of your favorite color. All credits for this nice idea should go to Greg Ordy, W8WWV.

-Update April 4th. - Effective immediately no more orders can be taken and no payments accepted since the manufacturer will run the batch ahead of schedule. A few spare boards will still be available on "first come-first served" basis.

-Update April 5th. - All the board kits have been taken. However, I will keep maintaining the list, and if there will be enough interest, another batch could be run in a month or two. If you can tolerate waiting, you can send a request to the email address below and ask to be put on the list. Do not loose time while waiting, get your Valpey-Fisher master oscillator. If you have not done yet so, subscribe to N2PK Yahoo group, and you will find the necessary information there.

Thanks to everyone who participated in the first manufacturing run.

-Update April 18th. - The boards have arrived! I approached the package with criticism in mind, but after I inspected the boards I could finally relax because I found them to be of a high quality. My critical taste was completely satisfied. I will be pleased to deliver these nice boards to the builders who participated in the order.

In the picture below you can see two holes above the T110 transformer and below the C185a component. Their purpose is to help you secure the DDS and master oscillator heatsinks. A particular method of doing this is left to your imagination. It could be a thick narrow aluminum plate that is held on the board with two screws, and then you can go from there and mount the heatsinks to this plate. That would give more freedom in choosing the heatsinks.

I will start sending the boards out as soon as possible. Another PCB order could be placed to satisfy the demand should this need arise. Feel free to send an inquiry to the email address below if you want to be put on the list for the second batch.

-Update May 3rd. - I've completed assembly of a v4.1 board and powered it up. The board worked from the first try, that means there was no layout errors migrating from v4.0 to v4.1 layout. It was 3 AM in the morning, so I only swept both detectors for the noise floor level. See the screenshot here (scroll down to the "Update May 4th" section). The detectors noise was at about -115dB level and the board was not even in the case at the time. I will be running other tests during next few days.

-Update May 11rd. - I've done one more v4.1 board and it also doing fine. Now, after I assembled and launched two of them and became confident the layout is functional, I can offer more boards kits to the builders. I will be placing another manufacturing order with end of May delivery. The boards kit, like it was the first time, will include 1 VNA v4.1 dual detector board and 2 reflection bridge boards. Pricing stays the same. To place your order, feel free to contact me via the email address at the bottom of this page.

Check the boards kits availability here. Thanks for dropping by and reading this page.

Contact: miv@makarov.ca