Design
- 6 element filter
- W7ZOI Crystal Filter paper has several crystal filter designs
- Narrow bandwidth bandpass filters for CW/SSB
- This is the 6 crystal SSB6
- Used for transmit/receive
- 9 MHz nominal frequency
- 50 Ohms in/out impedance
- Bin crystals using Colpitts Crystal Oscillator
- Hand select close values from a pile of crystals
- Can get much closer than 10% of bandwidth with 50 crystals
- W7ZOI Filters on Kits and Parts Boards
- Other filter builds
- 5 or less crystals
W7ZOI SSB6 Schematic
Build
Schematic
Dishal Values
RefDes Dishal RefDes W7ZOI
Ck12 87.3 C3/C9 82
Ck23 110.9 C4/C8 100
Ck34 114.4 C6 100
Cs1 110 C2/C10 120
Cs3 369.5 C5/C7 470
Measured Crystal Values
- 6 crystals within 7 Hz of each other
- Selected from 50 purchased
Number Freq Delta from average
09 8.997791 -7.7
12 8.997791 -7.7
41 8.997793 -5.7
06 8.997794 -4.7
21 8.997798 -0.7
38 8.997798 -0.7
Enclosure 80x50x20
- Fits in 80x50 extruded enclosure
- Material: Aluminum
- Color: black
- Size: approx.80*50*20mm
- Thickness: approx.1mm
Performance Measurements
Overall
Detailed
- 1 KHz/div
- Bandpass looks about 2500 KHz = good
- Left freq = 8.99762 MHz
- Right freq = 9.00005 MHz
- Delta = 2.43 KHz
- Center frequency = 8.99885 MHz
- Implies crystals are properly matched?
- Insertion loss at "best" point (red marker) -5.14 dB = OK
- Deep skirts = good
- Ripple looks bad
- Right side low
Compared to SSB4
- Fcenter = 8.998MHz
- Span = 10KHz
- 1 KHz/div
- Skirts much deeper on SSB6 than SSB4
- -90 dB on SSB6 at window edges
- -60 dB on SSB4 at window edges
- Passband much flatter on SSB4
- Insertion loss lower on SSB4
- SSB4
Improvements
- Investigate changes to remove ripple and clean up right side of passband
Clean Board
- Cleaned board
- Removed wires on transformer that were not clipped flush-ish
- No effect - same ripple, same drop at the right side, same insertion loss
- No change
Add grounds to cases of crystals
- Had to scrape away solder mask to solder
- Through hole vias were plated through
- Would be better with ground pads with cleared solder mask and thermal relief
- No change
Incremental Build and Test
- Start with unpopulated board
- Add SMA connectors
- Added transformers
- Installed 200 ohm resistors at input caps to ground
- Measure transformer impedances from NanoVNA
- Input
- 48.9-j0.900 Ohms
- VSWR 1:1.031
- Output
- 48.7-j0.714 Ohms
- VSWR 1:1.031
- Impedance is flat over passband
- Transformer working as expected
- 1-30 MHz scan
- Toroid flat to board or tilted up = same performance
- Not flat over band, but OK at 9 MHz
References
Sorted crystals table
Assembly Sheet