Howdy Fellow QRP elitists! ============= Introduction: ============= Contained herein is the information on a very nice RF Notch filter. I have built many of these and they are absolutely amazing. My design is slightly modified from the original author, Jean Dickensen, working for Microwave Filter in East Syracruse, NY. The reference article is old, yellowed, and only has the identification "CATJ" on the binding edge. No date given, but I know I built my design from it back in the 1980-1981 time frame. I also do not subscribe to their description of a "Enhanced Inductor" as the secret of the design. My derived filter is better than theirs and operation is easily described -- there are no secrets. ========= Overview: ========= I wanted a television interference filter. But I needed to guarantee it had minimum loss well up into the UHF spectrum. The undesired signal was an FM station which affected my VHF Channel 6. A 1/4 wave wire stub hanging off the back of the TV was cut and I had a very hi-Q short at that frequency, but the stub would get bumped during the course of everyday life and I had to keep messing with it. Also, some reflections interfered with optimum picture quality, and I wanted better control of the stub 'Q'. Besides, maybe an improved filter could find other uses. After much research, I found that most RF filters could do a good job at the VHF frequency, but impedance mismatches and response up to 1 GHz was disappointing. That's when goal number one materialized: 1) The filter must be a throughline at all frequencies EXCEPT the filter frequency, where I wanted loss. Later, I found the article. They had two RF connectors, one at each end of a metal box enclosure. A wire connected both center pins. At the very center of the box was a Hi-Q, fractional PicoFarad Mylar/Ceramic capacitor -- Value=.31 pF. In fine print they mention that a short section of coax can act as the small capacitor. After reviewing their complicated formulas and their description of "Enhanced Inductor" resonating with the .31 pF capacitor and the Parallel capacitor/Inductor attached below making the inductor appear 6 times larger than it really is, well, I decided there must be a better explanation. They pretty much lost me. =========================== Description: (Ed Theory :) =========================== So I calculated Reactance values and sure enough, that .31 pF was more than TEN times the Load Impedance! In fact 70 times the load impedance. Yup, rule of thumb is for 10X reactance, approx. 90 degree phase shift occurs at this point. So a signal entering the coax, bleeds across it and is subsequently re-reflected back, it will undergo 180 degree transformation and combine with the input signal. And if it weren't for dielectric losses, we'd get 100 pct. cancellation. Now at 88 MHz, that capacitor had more than 5 K-Ohm reactance. Wouldn't I get more signal to the parallel tank circuit if I made it 750 Ohms or slightly more? Still get 89.5 degree phase shift, and more signal to cancel the input with. It worked! I got a much deeper null and only minimally wider bandwidth. This is a 2.2 pF capacitance. Checking the .141 semi-rigid hardline I had, I decided 3 inches or so would be the maximum length. =========================================== Here's an ascii DWG to help save some words: =========================================== (Coax Shield Floats) BNC >---O---><-----O=================O-----><---O---< BNC | | | //// +---+----+ /// --- L1 Trimmer C1 --- | //// /// (Figure 1 ) ---------------------- Parts List for Fig. 1: ---------------------- C1 approx 1.7 to 10 pF Trimmer. (Ceramic) L1 3/4 inch Dia, .4 inch tall, 5 turns single spaced. Coax is .141 Semi Rigid Coax with shield Floating just Above the ground plane by 3/8 inch or so. ---------------------- Operation of circuit: (Ed Theory...) ---------------------- Basically C1/L1 are parallel tuned Tank circuit. At all frequencies except the tank resonance, the shield is adequately grounded thru C1 or L1 depending if the signal is above or below the resonant tank frequency. At Fo, the Tank isolates the coax from ground thereby disrupting its function as a transmission line at that frequency and becoming a Hi-Q capacitor causing approx 90 degreee phase shift as the incoming signal attempts to couple to the tank circuit. The tank circuit re-reflects this signal, undergoing additional 90 degree phase shift resulting in 180 degree shift. coupled with the incoming signal, yet somewhat weaker, the signals cancel, resulting in the Notch rejection. You can build one in less than an hour, even if you have to find the parts! ============= Construction: ============= OK. You'll need a 4 inch by 2 inch by 2 inch solid nickel tinned aluminum can. These are common in the electronics field. Barring this, you can make an enclosure completely out of double sided PC Board and soldering Beads along each edge until only the top is open. I recommend you install the BNC connectors at the ends and punch the holes before soldering. OK. Parts list: 2 BNC chassis, female connectors with one end solderable. Length of .141 inch diameter semi rigid coax prepared: Note #1:(Prepare Coax after studying directions Carefully!) ** Total length to reach from centerpin of each BNC (Note:#1) ** Strip 1/2 inch off each end of coax. The shield will NOT be Grounded to anything! Length of Tinned, 16 or 18 AWG wire, enough for 4 to 6 turns, 3/4 inch diameter coil, spaced 1 turn (Compress as needed after soldering). ONE Ceramic variable capacitor (Glass Piston Recommended) but not necessary to be Glass Piston. Capacitor from 1.7 to 18 PF should be optimum. The best Capacitor has two bottom legs, and one higher up, like upside-down letter 'Y'. The adjustment slot is up by the single leg. Construction. Ensure Coax is above the ground plane and the capacitor can be soldered to the ground plane and the top leg of the capacitor reaches the shield of the coax. This sets the height the coax is above the ground plane (About 3/8 inch). Shape the ends of the center wire of the coax to reach the BNC center pins on each side of the box, yet still have the coax shield FLOATING across the inside bottom of the metal can. Solder the Capacitor to the center of the coax shield and other end of capacitor to ground plane of the can. Wind the Coil and fit in box best you can, clearing the bottom and the top turn soldered to the capacitor/Coax shield. Connect to Rf signal and terminate in 50 Ohm load with RF Detector. Tune up and measure passband with Capacitor in Center Range. HINTS: IF Bandwidth is too wide, remove 1/8 inch increments of the coax shield from each end (Leaving Center wire intact). You should get between 15 and 25 dB attenuation for wider bandwidths, and about 13 dB notch for very narrow bandwidths. For very high frequencies, you may end up with only 1 inch of Coax shield floating and the center wire running about 3/8 inch above the ground plane. If you have a lot of center wire from the coax, you would be better off making the box smaller to minimize Blow-by from coupling to the coil. Also, the coil 'Q' is fairly important. Remember Highest 'Q' coils are Taller than they are long! i.e. A .8 inch long coil should be 1 inch in diameter and turns spaced only ONE Wire diameter. (General Rule of thumb.) If you get 16 dB attenuation and 200 KHz bandwidth, you did good. Now build a second trap. You cannot connect the second trap DIRECTLY!! You must get 1/4 wave of semi-rigid coax, coil it up and connect between the two traps. This eliminates interraction between them -- It is amazing!!! The best way to coil semirigid cable is to find a Tomato Paste can or similar hard object (Wooden Dowel?) and wrap tightly, remove. Final tune up is with the top soldered on the cans and access to the trim cap via a small hole in the top. I built one of these with small metal can enclosures, and insulating wall, 1/4 wave coax, insulating wall, and second trap. Fabulous. 45 dB rejection of the FM signal and I recall only 250 KHz bandwidth. Set and forget tuning. Try it out. P.S. For extra good results, you can use .141 Coax for the inductor or Icemaker tubing, but then you end up needing a HUGE box. Experiment on a PC Board and when you get the correct size coil, build your finished product. ========================================================================= Formulas and Misc Info: ========================================================================= Cc=(Df/2.98)*(159.15/Fo)^2 <== Cc is Coax Capacitance. C1=((2.98 + 1.5 * Df)/(Fo)^2) * 5665 L1=2.98/((119 * C1 * Fo^2) * 10^-6 - Df) Note: C1 value in PicoFarads. L1 value in MicroHenrys. Df is Delta Frequency, i.e. 200 E+3 for 200 KHz. ========================== Homebrew Inductor Formula: ========================== LuH=(R * N)^2/(9R + (10N/P)) where: R=Coil Radius (inches) N=Number of turns. P=Turns Per inch of windings. ========== Reactance: ========== Inductive ==> XL=2 * PI * F * L Capacitive==> XC=1/(2 * PI * F * C) (All in Base Units) -Ed Loranger Sept. 28th, 1998