Thursday, March 29, 2012

Yet another set of VNA calibration standards

This time - Type-N female. One of the most critical components in DIY calibration standard is the connector used to build it. The market is flooded with all kinds of connectors but most of them are of really shoddy quality (sorry to bash on China but that's where all the junk comes from). After playing with and testing different connectors, my conclusion is that one is shooting in the dark ordering connectors from China to use for DIY cal standards. Very rare there will be good enough connector from China, even for 60 MHz VNA cal standard. It is unbelievable the amount of sub-standard junk they make!
The best way to go about is to find high-quality brand name connectors on the surplus market. Ebay is also a good source but it takes time and patience to find what you need.
I've mentioned before that having a few sets of cal standard -different connector type and gender is very useful! I needed a good set of female Type-N OPEN and SHORT (I have a commercial LOAD).
One can buy on eBay a used female Type-N to female-SMA adapter of REALLY high quality for about $10/pc - I am talking about Amphenol 131-445 / HP-Agilent #1250-1404. Even at $30 this adapter is still a bargain. It is actually a bulkhead female Type-N used for commercial VNAs, Spectrum analyzer and other lab equipment. The body is machined from stainless steel, the pin receptacle is gold-plated copper-beryllium and it is using very interesting dielectric.
The center dielectric is made of a small plastic disc with 6 dead holes - 3 on each side of the disc. In each set the holes are at 120 degrees and the two sets are offset by 60 degrees from each other. As a result - the center pin receptacle is supported only by 6 very little plastic spokes. The holes are fairly large leaving mostly air around the center. This is as close as it gets to DIY air dielectric while still have strong mechanical support for the pin receptacle. I think the reason for the "dead holes" is to stop dust and contaminants getting inside.

For the OPEN I just cut the center pin flush to the dielectric. The dielectric disc has a brass sleeve in the center. A dab of solder fixes the pin receptacle to the sleeve. The disc is not made from teflon and soldering time and temperature should be minimal. One can easily measure the distance to the reference plane and calculate the offset. The stray capacitance and losses are minimal due to the dielectric's construction.
The SHORT is made by soldering a "washer" of tin-plated brass to the center pin. When the back shell (the gold plated part) of the connector is screwed in place it presses onto the outer part of the washer and makes the electrical connection to the connector's body. I left the rest of the center pin intact - it goes inside the teflon dielectric of the back SMA part for additional mechanical support.

Thursday, March 1, 2012

Crystal Test Fixture

Jack Smith, K8ZOA posted an interesting paper on "Measurement approaches of Crystal Motional Parameters" and designed a test fixture (based on the "classic" IEC 444 pub. fixture) to aid such measurements. The fixture is rather simple - it is comprised of two impedance matching attenuator pads, presenting the crystal resonator with a low, accurate impedance (12.5 Ohms, the center of the 5-20 ohm range of crystal's series resistance) during transmission measurements and transforming it from and to the 50 ohms impedance of the source and detector. These attenuators act as sort of a "buffer" - because of the transformation, the crystal "sees" only the accurate 12.5 Ohm input and output impedance regardless of the source's and detector's return loss.
Each attenuator is in Pi-pad configuration with attenuation of approx. ~15 dB (due to standard resistor values, the actual calculated attenuation for each pad is 14.81 dB). Input shunt is 158.0 Ohms, series resistor is 66.50 Ohms and Output shunt is 14.30 Ohm (ideal calculated value is 14.20) - all in 1206 packages 1% tolerance.
It occurred to me that Ivan Makarov's N2PK VNA Reflection Bridge PCB can be modified very easily to accommodate both pads and I can use the board to build the entire test fixture. The modification is very simple!

I used a Dremel Tool to drill (Carbide PCB drill bit #71, 0.0260") 3 holes (0.1" spacing) for the crystal socket / header and a small end mill bit to make to cut two trace cuts.

The socket is a gold-plated machined type - it can be cut from an IC machined socket. The HC-49 crystal package fits perfectly in the socket. An adapter can be made if the fixture is to be used with other components with larger diameter leads.
The socket is installed on the reverse (solid ground plane) side of the board. The middle pin is soldered to the ground plane on the component side. Copper is removed around the two side holes (on the socket side) so it will clear the bottom part of the socket pins. The ground plane should clear each socket pin by at least 1 mm around (the diameter of the no-copper area around each hole is about 0.1"). Again, a round end-mill bit or a big round engraving diamond file bit on the Dremel Tool will do the job.

This picture shows the configuration of the resistors in each Pi-pad. The ideal value for the output shunt is 14.2 ohms, so I installed 2.2k (1%) resistors in parallel with each 14.3 resistor to bring the resistance down. I used my LCR meter to select all attenuator resistors in order to have values as close as possible to the ideal calculated values. RF screen (cut from tin-plated brass sheet) is installed across the board, on the component side to shield the output from the input. The unnecessary PCB pads are flooded with solder and connected to the common ground plane.

Two edge mount SMA (f) are installed 2" apart. The 2" spacing matches my N2PK VNA port spacing so I can install the fixture directly on the VNA ports without the use of any cables.

Two copper-beryllium contact clips - one on each side of the socket are soldered to the ground plane. These spring clips connect the resonator's can to ground when the crystal is inserted and provide additional mechanical support for the crystal package.
The picture also shows the THRU calibration jumper and 50 ohm load test jumper.