The case of the Ding Dong Distraction. My XYL is in poor health and daily she takes an enormous amount of medications. She also has difficulty swallowing, so, the process is to grind up the medications and mix them with applesauce.
That concoction smells bad let alone put it in your mouth. Lately she is refusing to take the medications and so I had to come up with something to have her take and swallow the pills. Enter D^3 the Ding Dong Distraction. I now have a Ding Dong in my hand and tell her take the pills and you get the Ding Dong. So far it has been working.
But we also have the D^3 in our hobby only there is a slant to this that is sort of a reverse. That Distraction is the Tiny SA and the Nano VNA.
Many hams (new and old) have purchased these tools as they were distracted into believing that having those tools was a workaround to having detailed knowledge of how things work.
In a very recent email exchange with a ham, he shared he is building a receiver with two NE602's (like yesterdays post) with a 3 pole Cohn filter tucked in between the two chips. So far with a Signal Generator and a Scope he has plotted a curve across the filter pass band and in short he says it looks like crap.
He did match the three crystals to within 6 Hertz but has not done the other characterizations like the motional inductance and capacitance, Rs and the Q. Thus he is missing some key data to build the filter.
The very first step should be to get smart about crystal filters and here is a really good place to start. Search the Internet by using the parameters WA5BDU + Crystal Filters. It is a about 45Pages
There are several notable pieces of software like Dishal and AADE that take these factors and output two important pieces of data. The first is the value of the filter capacitors and the second is the impedance. Wes Hayward also has a program for building filters and has a very recent publication on building a 9 MHz Filter.
You will find there are many filter configurations starting with the simple ladder type filter and moving on to the now popular QER (Quasi Equal Ripple) configuration. Search on QER Crystal Filters and that too will yield source material from several hams. One of those chaps reads this blog.
Some of the differences in the software and configurations manifest themselves where in one form uses the same capacitors at each pole while others generate a listing of (often custom values) specific capacitors at each pole. This is all about trade offs such as filter skirts, ringing and suitability for use in digital applications (like square waves).
Worth a mention is the filter skirts, essentially how steep the rise to the pass band. As a rule more crystals in the filter the more vertical the rise which is important in keeping signals out of the pass band from sneaking through the pass band.
You may hear terms like shape factor which is a ratio of how wide a filter is say -60dB down in comparison to the pass band width. Often causally thrown around is the term Brick Wall Filter i.e. the shape is near vertical. Regrettably the brick wall is hard to achieve with crystals "but easy to do" with digital filtering using software like the Iowa Hills code.
Often without extraordinary effort, homebrew crystal filters are not symmetrical and you will hear that a specific configuration favors LSB. I may be wrong on this but I also think the Dishal array can favor USB.
Note WA5BDU's work predates the Nano VNA and Tiny SA devices so you will not see them mentioned. This also is a key point -- high quality filters can be built without those currently popular tools.
The capacitors in ladder type filters set the filter bandwidth with smaller values yielding wider bandwidths. Somewhere between 47pF and 100pF are typical SSB values and something around 150 to 470pF are in the CW filters with the higher value giving a very narrow band pass.
The impedance of the filter needs matching to your circuit. If you have an itch in your twitch to use a 50 Ohm TIA (termination insensitive amplifier), that my friend must be matched to the filter.
Do not overlook matching circuitry like an "L Match" (Red Cores below) for handling the impedance transformation. I have a K7TFC QER Filter in a radio and used an L Match to go from 137 Ohms to 50 Ohms. This module uses the QER Filter with the L Match/BFR106 and a twizzle with a pot in the Emitter leg to adjust the gain of each BFR106 amp. This is a single pass amp withe relay steering. Thus, the 1st BFR106 can be set to not over drive the filter and the second BFR106 gain is set for the best signal quality output. I can see that on a scope!
Typically Filter Impedance may range from 100 to 500 Ohms. The matching transformer and how close you get to the match determines the pass band ripple of the filter. Selection of capacitors and the matching transformer build cannot be done haphazardly that is if you want a great filter.
Now the distraction. This ham who emailed me, stated that today he was to receive his Nano VNA and was anxious to measure his filter. The Distraction Meter is pegged!
If he does nothing else and assuming there was some care when he made his scope measurements the Nano VNA will not fix the filter problem. It will not tell him anything he does not already know (the filter sucks). It will only affirm the sucking.
The real effort is to go back and study the "how" to build a proper Crystal Filter and to build devices like the G3URR test oscillator and take the data.
Next is to study how to use the software and once you have the two critical pieces of data -- Build To Print. Once built I would also suggest taking a manual run through with his signal generator and scope and then and only then get out the VNA --assuming he did calibrate the unit and knows how to do the test. You then can compare the "old way" with the new devices and satisfy yourself that they indeed match.
The typical distraction is to rip open the Nano VNA box, hook up a few wires, skip the calibration and just punch buttons. Then you might declare this filter sucks. Are you sure you have the right resonating caps and the proper matching transformer? -- Do you have to look at that stuff is suddenly a question. It is not a question, it is something you have to do first!
Do not be distracted by thinking some cheap tool is a substitute for a lengthy and detailed process that involves test, measurement and software calculations. Yeah that can be boring but necessary.
For those new to homebrewing who contact me regarding building one of my SSB Transceiver designs, I always advise buy your filter and build the rig. This way you have removed the filter from the critical path. I always envision my designs as experimenters platforms. Once you have the rig working then go build a homebrew filter for now you have a benchmark of how the rig should perform and the only thing that changed was the filter. Start small and start wisely! A four pole filter is a great place to start! A 12 pole first time to build any filter is not a good idea.
Filter measurement aside from bandwidth and ripple is also on the list. The big issue often ignored with many new filter builders is that the filter has some attenuation. So an IF Module needs to consider that you might have two amplifiers in the module which is +dB but you have a filter that is a -dB. As a rule of thumb I typically put a -5 to -6 dB in for the filter loss when I think of the module gain. Friend N2CQR built a 10 pole filter that he measured with only a - 2dB loss across the filter. That is something that I have not seen in the filters I have built. But indeed he saw it.
Also critical is how you slap the crystals together to form a filter. Ugly construction with tack soldering is a BAD idea! Many filter tutorials describe building the filter over a ground plane and that each crystal can should be grounded. The idea is to force the signal through the filter and not around it! One tutorial even goes so far as to suggest shielding between sections like you see in this photo. Yes I was short a core so stole one from this 6 pole Dishal and the capacitors are of different values.
Some notes about this homebrew filter. The crystals are 11.5 MHz (below the suggested 12 MHz top end) and the filter was to be used in a dual conversion radio with the 1st IF at 45MHz using a commercial filter. Note the numbers on the crystals. I started with a bag of 50 crystals and numbered each one. I had to go through all 50 before I could find 6 that all were within a 50Hz spread high to low. Many are only 5 or so hertz apart. The Crystal cans were grounded and I did install a shield between the sections. Not sure that shield was necessary but it makes it look upscale.
A bit of a gulp here. When I built this jewel I grumbled a lot about how to build the filter took a lot of fiddling with the Dishal software and nicknamed it Dishal Dystopia. At about this same time I was making a Skype presentation to a Radio Club in the UK and after the slide show there was a Q&A. One chap asks how come you don't like the software I designed? Luckily this filter was sitting on the desk and I picked it up and put it in front of the camera. His comment -- you did a very nice build. You never know who is in the audience!
Todd K7TFC sells QER filters in several different frequency ranges. In essence using this approach you are setting the stage for success. The filter is the heart of the rig. Screw that up and you have nothing and possibly a failure at this point may suddenly make stamp collecting the alternative choice.
Don't be distracted but do the upfront work needed to successfully build a crystal filter. The last thing you do is run your Nano VNA -- not the 1st thing! Todd offers a complete filter kit or just the PC board. In the event you are anal retentive and want to proceed with doing it "your way" -- at least get his board as that puts you 50% up the success ladder.
73's
Pete N6QW