This is a visual demonstration showing the major steps in the process of designing a filter using the new all- Windows version of S/FILSYN.
The first step, after establishing the required performance specifications, is to take a guess at the design cutoff frequencies, and enter these into the program, along with other parameters. This is accomplished using a Windows- type menu, with multiple choices and places for numeric entry, as in this screen-shot:
More details of the stopband specifications, corners or template, and specific infinite rejection frequencies (if desired), can be entered on the next screen, as "detailed parameters".
After entering all the information, the program proceeds to compute a transfer function, and optimize, moving the design parameters to equalize the "safety factors" in all stopband regions. This can be repeated and parameters changed until satisfactory performance is achieved, as seen in this screen shot:
At this point, the program is told to proceed with synthesis, and it selects a feasible topology, and synthesizes a filter. ALKDEM1 (seen below) is the filter as synthesized by S/FILSYN, with no direction or assistance. It can be analyzed, in S/FILSYN, and will be found to meet its specifications, but it is not in a form, or with element values, suitable for building. The termination is not 50 ohms, there are 'hot' nodes with no accomodation for stray capacitances, and we cannot yet determine whether the component values will be graceful.
1 |---R---| 50.000000 ohm 3 |---C---| 22.145111 pF | ,-', 4 | L C 80.057065 nH res.frequency | `-,' 21.393992 pF 121.611427 MHz 5 |---C---| 11.181378 pF | ,-', 6 | L C 707.430739 nH res.frequency | `-,' 2.783256 pF 113.423128 MHz 8 | C 807.550316 fF 9 |--L-C--| 168.771711 nH res. frequency | | 30.344368 pF 70.328484 MHz 11 |---C---| 29.691067 pF | ,-', 12 | L C 272.778047 nH res.frequency | `-,' 7.629688 pF 110.321973 MHz 14 | C 1.970483 pF 15 |---C---| 1.363709 pF 17 |---R---| 521.389450 ohm
The first step is to do a Norton transformation, and to interchange a few branches, to see what the element values become when we have 50 ohm output. In ALKDEM2, we see that the last section has an inductor of only 26 nanohenries, which we feel might be inconveniently low. No exact transformation can alleviate this deficiency.
ALKDEM2 1 |---R---| 50.000000 ohm 3 |---C---| 22.145111 pF | ,-', 4 | L C 80.057065 nH res.frequency | `-,' 21.393992 pF 121.611427 MHz 5 |---C---| 11.181378 pF | ,-', 6 | L C 707.430739 nH res.frequency | `-,' 2.783256 pF 113.423128 MHz 8 | C 807.550316 fF 9 |--L-C--| 168.771711 nH res. frequency | | 30.344368 pF 70.328484 MHz 11 |---C---| 25.298450 pF 12 | C 6.363100 pF 13 |---C---| 14.184678 pF | ,-', 14 | L C 26.158762 nH res.frequency | `-,' 79.560782 pF 110.321973 MHz 15 |---C---| 14.220471 pF 17 |---R---| 50.000000 ohm
The next schematic, ALKDEM3, shows the result of several operations, all of which are made by simple point and click strokes right on the screen. We have moved the 110 MHz peak from the end of the filter and put the 113 MHz peak there. This is an exact transformation, and results in that inductor increasing to 40 nanohenries, which we will deem to be an acceptable value, for the purposes of this example. Note that there are no longer any 'hot' nodes.
ALKDEM3 1 |---R---| 50.000000 ohm 3 |---C---| 22.145111 pF | ,-', 4 | L C 80.057065 nH res.frequency | `-,' 21.393992 pF 121.611427 MHz 5 |---C---| 6.078869 pF 6 | C 6.088803 pF 7 |---C---| 78.318170 pF 9 |--L-C--| 53.421024 nH res. frequency | | 95.866207 pF 70.328484 MHz 10 | C 27.138332 pF 11 |---C---| 22.910529 pF | ,-', 12 | L C 40.411541 nH res.frequency | `-,' 51.500425 pF 110.321973 MHz 13 |---C---| 8.357900 pF 14 | C 6.277423 pF 15 |---C---| 10.829628 pF | ,-', 16 | L C 40.411541 nH res.frequency | `-,' 48.722730 pF 113.423128 MHz 17 |---C---| 17.661198 pF 19 |---R---| 50.000000 ohm
The final schematic, ALKDEM4, shows a buildable filter. Here we see the same filter after several more transformations, all done by simple operations on the screen. We have converted the inconvenient series resonant three-element branch to one in which a capacitor appears across the coil. We have made a few Norton transformations to make the two internal coil inductances the same as the end 80 nanohenry coil, but they could be the same as the 40 nH coil, or values in between if desired.
ALKDEM4 1 |---R---| 50.000000 ohm 3 |---C---| 22.145111 pF | ,-', 4 | L C 80.057065 nH res.frequency | `-,' 21.393992 pF 121.611427 MHz 5 |---C---| 7.862239 pF 6 | C 4.305434 pF |-L-, | 80.057065 nH 7 | |-C-| 55.374117 pF res. frequency |-C-' | 8.596138 pF 191.853007 MHz 9 |---C---| 30.392276 pF 10 | C 13.633929 pF 11 |---C---| 11.629945 pF | ,-', 12 | L C 80.057065 nH res.frequency | `-,' 25.996600 pF 110.321973 MHz 13 |---C---| 2.927686 pF 14 | C 4.459994 pF 15 |---C---| 12.647057 pF | ,-', 16 | L C 40.411541 nH res.frequency | `-,' 48.722730 pF 113.423128 MHz 17 |---C---| 17.661198 pF 19 |---R---| 50.000000 ohm
Every operation shown, every result achieved, was done right at the keyboard, and screen, inside the S/FILSYN program, without the use of external programs, calculators or any other devices. Below, just as S/FILSYN would display it, is the final schematic and analysis: