SSB6 Design

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

• Response curve

Build

• SSB6 filter is too big for Kits and Parts board (5 crystals limit)
• 2500 Hz bandwidth

Schematic

Dishal Values

• Impedance = 195.9

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

• NanoVNA
• 2.5 KHz / div
• Initial measurement shows issues (solved later on this page)

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 (Version 2)
• No change

Incremental Build and Test

• Start with unpopulated board
• Add SMA connectors

• Added transformers
• Installed 200 ohm resistors at input caps to ground
  • At C1, C11

• 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

• Add machine socket pins to install parts

• Install caps in socket pins

08	8.997809	10.3	10.3	
30	8.997810	11.3	11.3	
48	8.997812	13.3	13.3	
50	8.997814	15.3	15.3	
31	8.997818	19.3	19.3	
46	8.997822	23.3	23.3

• Parts installed on pinned card

Fixed First Card

• Compared pinned board to first board
• Noted C5 and C6 were swapped
• Repaired
• Sweep

• Excellent results
• 2.625KHz bandwidth - close to 2.5KHz design goal
• 2.5 dB insertion loss
• Low ripple.
• Fcenter = 8.998892 MHz

SSB4 vs SSB6

• Input complex impedance across passband

References

• Crystal ladder filters – Software
• Crystal Ladder Filters for All - ARRL QEX
• DISHAL software by DJ6EV
• A Practical Test Set for Comprehensive Crystal Testing
• Crystal characterization and crystal filter design
• Crystal parameters Written by Hans Summers (QRP Labs)

Sorted crystals table

• Sorted crystals table

Assembly Sheet

• SSB6 Assembly Sheet