Building a prototype Dual Detector VNA

Getting closer to completion of the VNA board

Oh, boy! What a nice feeling when after many late night testing and tweaking, desoldering the components and soldering them back onto the board, making thousands of Google queries, time comes to move on. All removed during multiple tests components are now back where they belong to, the shack desk is clean and ready to receive a fresh pile of jumper wires, bobbins of solder and flux bottles, and the DigiKey guys started to forget who you were.

What I noticed a while ago, was that working on a project is much easier, and the project has more chances to be finished if you build it the way that gives fast and easy access to its functional parts. Normally this translates to installation of peripheral connectors to get the signals in and out of the boards instead of soldering the wires permanently. I now try to follow this rule even when I build circuits on the pegboards. Takes a few seconds to get the board out for the parts replacement, to clean the flux, drill holes, or show to your colleague.

Same approach was used with the N2PK VNA dual detector board. In addition to the SIP row connectors existed in the original design, I wanted to interface the RF signals by terminating them on the board with the SMA or SMB type RF connectors. The benefits received were hard to overestimate, especially when working around the enclosure. The board now is pretty much finished, and I posted a few pictures of it below. You may find a few extra components installed, and a few of them missing, but that were the ones needed only for the experimentation, or which were not critical to my use, such as the master oscillator buffer IC and its output connector, which I did not install at all.

 

Download a bigger picture of the board done with a higher resolution. Sorry, that was the most my old crappy Sony camera was capable of.

Observations and changes to the design

The differential circuit design used by the author, Paul N2PK, provides good common noise suppression even if no enclosure was used. The first tests I did with the board showed noise level at a -100dB mark. Mounting the board in the enclosure improved this figure by 10dB dropping it down to -110dB.

It appears the board and ambient temperature changes contribute the most to the detectors offset drift. The offset drift affects calibration. If you calibrate the device shortly after it was powered on, in a few minutes your calibration will be gone. This mainly affects low signal level measurements and may not be that critical for most of measurements done in transmission mode, but you have to be aware of that. My experiments showed that heat removal, such as usage of a big heatsink or a fan, may not be the best method to fight thermal drift. In this case the drift still exists, slowly rising with time (for hours). Instead, use of a small to moderate size heatsinks for the AD9851 ICs and the master oscillator gives much better drift curve, which rises fast during first 20 minutes, and then the change stabilizes within a microvolt or two. The idea here is to raise the board temperature above the room temperature so that changes in the room temperature do not have much effect.

Specifically because of the temperature conditions given above, I decided to give up the idea of using a single power circuit for both detectors. The updated layout will be rolled back to using separate power supply circuits for each detector. That is because my current design has the power ICs installed on the first detector only, and they heat the first detector, whereas the second detector stays as cool as a cucumber. Additionally, the first detector is closer to the main source of the heat, which is the Valpey-Fisher master oscillator. This causes the temperature gradient to rise unevenly along the board length, and the offset drift of the detector 1 and 2 change in a different proportion (Detector 2 change is slower, and it takes more time to stabilize). Though all this is not critical and the board I've built is going to stay as it is, the new boards will have their power circuits separated. The idea is to cool the detector 1 down a bit by taking some load off its power ICs, and warm the detector 2 up a bit by introducing some heat generated by its own power ICs.

In order to help the builders to secure the DDS and master oscillator heatsinks, I will update the layout with two mounting holes. They will be located to the left of the DDS ICs in the ground plane area. The particular methods of mounting of the heatsinks is left to the builders engineering skills.

The onboard traces between the LO DDS and the detectors seem to behave OK, and I did not notice any difference compare to using the coaxial jumpers. You can use the onboard traces for all your initial testing, and move to SMA connectors and coaxial jumpers later for the actual device use, where the LO DDS signal may need to be fed to the detectors through an attenuator. I have to tell you that the onboard traces were actually designed to be 50 Ohm transmission lines, at least theoretically. The purists having a TDR (Time Domain Reflectometer) in possession, are requested to measure the onboard traces impedance and then beat me up if this is not so.

When sniffing around with the scope probe, it was found the level of RF on the ground plane may differ from location to location. To provide a lower ground plane resistance to RF, the number of vias along the board edges, around the DDSs, and under the master oscillator will be increased in the next board revision.

Credits

I would like to thank the people who helped me in numerous ways to get this prototype board running. Here is the names in alphabetical order:

• Edgar Brown, N6OU - for believing in my skills and sharing the risk of building a prototype-based dual detector board.
• Milosh Rankovic, VA3VEF - for lending me his Minicurcuit transformer for my reflection bridge;
• Paul Kiciak, N2PK - for providing valuable comments, definitions of the tests, software, documentation, and results of his tests for comparison;
• Tim O'Rourke, W4YN - for lending me his master oscillator and providing a set of Minicurcuit transformers for the VNA prototype board.

Contact: miv@makarov.ca