Saturday, April 29, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.10

How to Build a 11.5 MHz Crystal Filter

Method #1:

Simply Purchase four 11.5 MHz Crystals at the cheapest price you can. Next build a Ladder Filter using five coupling caps of the same value. For a SSB Filter use 68PF and for a CW Filter use 470 PF. A guess at the in/out impedance would be in the neighborhood of 150 to 200 Ohms. Use 200 Ohms, as a 50:200 match is just a 4:1 transformer.
With this method you rely mostly on luck. It will probably not work too well. However if it does then you should immediately go out a buy a single lottery ticket as you are on a path to striking it rich.
Here are the shortcomings of Method #1. With only four crystals and making no measurement of their actual frequency you will never know: 1) how well matched they are in frequency 2) what is the filter center frequency and 3) the actual Zin/out. Did I also mention that if you don't test the crystals in an oscillator circuit prior to just installing them in a filter you may not know that one or several are inoperative (you did buy bargain crystals). But many "good enough" filters have been made this way. That said very likely there are substantially more poor filters than good ones that were built using this approach. But hey you built a crystal filter.

Method #2:

You purchase four crystals at the cheapest possible price and you make a measurement of the crystal frequencies using the G3URR test oscillator. You dutifully note the "loaded" frequency and the spread of each crystal as related to all of the crystals. (A goal is no more than about a 50 Hz spread across all of the crystals). After obtaining this data you simply ignore the information and follow the Method #1 approach. Again if it is perfect, then buy more lottery tickets. But more than likely it will not be. Oh by the way --you will probably need the center frequency info in your Arduino Sketch so you know how much to shift the USB LSB BFO frequencies -- but hey close is close enough. So you get a few dings from the SDR police on 40 Meters --who cares?

Method #3:

This is where you find some ham who really knows what they are doing and after an enticement of the standard B^3 (Booze, Bucks and Babes) have them build you the filter. Just sit back and relax and wait for the unit to arrive. This is a lot less stressful and all you need to do is install it in your rig. Now wasn't that easy? Never let the XYL find out you spent $250 for a crystal filter is the real issue.

Method #4:

This is where you get serious about homebrewing a crystal filter. The process involves the following:
  • Collecting information on how to actually build a crystal filter. There are several really good sources. First do an Internet Search on Nick Kennedy WA5BDU, as he has prepared an exhaustive tutorial on the steps needed. Also search on Almost All Digital Electronics as they have a computer program that is very handy to design a filter. I also believe that EMRFD has a program on the DVD that is located in the back jacket. Do not overlook You Tube Videos on how it is done. Bottom line you need a disciplined process and resource information.
  • You will have to build some test hardware including the G3URR test oscillator. Basically this oscillator enables you to measure the frequency of a crystal and then by loading that crystal with a small capacitance shifts the crystal frequency. That amount of shift is an important parameter in the final calculations. (It has something to do with pole zero spacing) This is where you need to buy one of those $13 TV SDR Dongles! Get one and modify it so it will work on HF. You should also download the free software program HDSDR. This $13 device will let you precisely measure the crystal frequencies with and without the load. Almost better than a frequency counter. All you do is power up the oscillator and with a short "antenna lead" bring the oscillator near the Dongle and look for the output.
  • Purchase twenty five 11.5 MHz crystal (about $0.30 each at this quantity) from Mouser. When they arrive use a Brother tape label machine on the smallest print size to label every crystal from 1 through 25. Open up a Excel Spreadsheet on your computer and record the loaded and unloaded frequencies for the crystals marked 1 - 25. The first thing that should amaze you is that the crystals while nominally 11.5 MHz are all over the map. Now you can either do it by visual inspection or have Excel do it but rank order the crystals from high to low frequency. Once you do that you should now look for groupings of crystals that are within 50 Hz total spread from low to high. You might get lucky and find five or six that meet this criteria but you need a minimum of four. You will most likely find out of the batch of 25 that you will have several groupings of at least four crystals that are close in frequency. Once you have at least four then you need to follow the process outlined by Kennedy. Look to the manufacturers specifications as you might be lucky to find the "Average" series resistance as this number is needed for the calculation.
  • Most likely the final filter (for SSB) will have low value coupling capacitors (around 100PF not all the same) and the Zin/out would be in the 170 Ohm range. But unlike Method #1, you are being precise in the measuring process and WILL have four crystals that are close in frequency.
  • Method #4 is not a 1 or 2 hour process -- it might take several sessions to complete the data analysis and calculations before you start soldering crystals to a circuit board. Be sure to connect all of the cans together and ground that connection. Build the filter over a large ground plane area.
  • At this point Method #3 has a lot of appeal.
Have fun.
Pete N6QW