Building an SMD Crystal Filter

What do you do while awaiting parts to build an SMD Bitx? You take inventory and see what you can build with what you have! I was waiting for a 1206 smd resistor assortment but had all of the other parts. I decided to duplicate the crystal filter. I would use 0805 resistors and chip caps with the smd crystals on hand. I had a kits and parts prototype board so I would use it as a base for all of the parts. The bi-directional amplifiers of the bitx were supposed to present a 200 ohm impedance between stages so I would use the 220 ohm resistors in my smd resistor assortment to represent that load. The version 3 bitx and the smd bitx filters used 2 - 100pfd and 1 - 220pfd capacitors so I would duplicate that. I would use some 0.100" molex pins that I had for the input and output connections.

After assembly, the board looked like this. The 0805 chip components fit very well on the 0.100" pads of the board. There are interconnection wires on the bottom of the board.

Next step, test it. I ohmed out the connections and everything looked good so I hooked it up to my HP 8507a network analyzer.

I was rewarded with a filter spectrum on the analyzer. I quickly noticed that the filter was not flat. That didn't surprise me as I had done no crystal selection. I had taken the first 4 crystals that were on the tape of 50 that I had.

Lets take a look at the insertion loss.

The center of the screen horizontally is 0db and the other settings are 1khz/div horizontal and 10db/div vertical. This display shows a narrow filter of a little over 1khz. That was a surprise as all of the filters I have tested in the past were wide. Lets look at it a little closer.

The settings now are still 1khz/division for the horizontal but now the vertical is 1db/division. This shows about a 1.6khz bandwidth at the -3db points with about -3.5 - -4db ripple in the middle. That's not very good. Lets look at a couple of other settings.

I've changed the horizontal display to 3khz/division. You can see the normal charateristic higher rejection on the upper frequency side of the filter. The right edge of the display is about 10khz above the filter band pass and is -58db down.The left side is about 18khz below the band pass and is down about -45db. Lets change the display one more time.

We are still at 10db/division vertical but now 10khz/division horizontal. The rejection difference is really apparent here.

What does all of this tell us? Well, I won't really won't know until I have a way of matching the crystals. I will try a basic matching for frequency once I get the bfo circuit on the smd bitx completed. It's harder removing and testing the smd devices than the standard through hole components though. You'll just have to wait for part 2.