I’ve been away from the blog for quite awhile, mainly because I ran into several roadblocks which stymied my progress with the SW40 build and analysis.
I really wanted to re-visit the type of analysis I did when I built the Bitx20A kit, but the laptop I was using for the TAPR VNA up and died on me! I really had no time or interest in tearing apart that unit and trying to “fix” it, and it was a sluggish and rather cheap netbook style older machine anyway, so I ended up buying a new cheap laptop to replace it.
Well, unfortunately, that did not resolve matters. It’s a long and boring story. But the upshot is, the TAPR software will only run on computers running on a 32 bit OS, and the new laptop was running 64 bit Windows 8.1! How frustrating! I contacted the software developer for this tool but didn’t obtain a solution. I ended up having to partition the hard drive of my desktop and installing a 32 bit version of Windows 7 on it! That cost some dollars too, unfortunately. But I’m happy to report that I can at least use this fun and educational VNA tool again, albeit from my desktop via a lengthy USB cable.
If anyone reading this is having similar trouble with the VNA, I am working with a tech rep from Cypress (the maker of the USB chip in the VNA) to see if they can create a 64 bit version of the driver (or help me find a workaround). I’m not too optimistic. But if they do, I will create another blog specific to the TAPR VNA and the driver solution, and will also pass along to the TAPR team and the designer.
Back to the SW40!
I revisited my blog from last year on analysis of crystal parameters. Most of this information can be found by digging through books by people like Wes Hayward or some ARRL handbooks and other resources. I went through that process last year, discovered the model for the crystals used in RF filters, and discovered the algebra and techniques necessary to find the values of the parameters of that model. That previous blog has the gory details of that process and I won’t repeat it here. I will pass along the results and some plots though.
Rs = 5.5 ohms
Lm = 125 mH
Cm = 0.0127 pF
Cp = 4 pF
Series resonant frequency: 3.998944 MHz
Parallel resonant frequency: 4.006180 MHz
And a plot of the S21 for the crystal:
These numbers, and the plot, compare fairly well to the ones obtained for the Bitx20A filter crystals. The unloaded Q is higher in this crystal than it was for the other kit. And the Cm was a little different. But generally the numbers seem in line for published values typical for these crystals.
Next I built the filter on some clad and connected to the VNA. The filter is meant for 4 MHz as you can probably tell by the plot.
The resolution of these snapshots is rather poor for some reason. But the 3dB points are roughly 3.998922 and 3.999393 MHz for a bandwidth of 471 Hz. The up and down wavyness in the passband is fairly dramatic and more than 6 dB down from peaks. But I took the passband as being 3 dB down from these outermost peaks. Beyond these two points the response drops off dramatically on either side.
The 6 dB bandwidth according to the specs in the assembly instructions from Dave Benson was 500 Hz, so this is spot on it looks like (which is more reassuring from the standpoint that my test setup is accurate than it is a validation of Dave’s work, at the moment at least). This is much narrower than the Bitx20A filter, but that’s obviously by design — this is a CW rig and not an SSB rig.
This time around I was determined to use a pre-packaged filter analysis software instead of “Puff”. Puff would work but it’s very basic and I thought it was high time to use some software meant for this purpose. I tried the AADE package (V. 4.5) and after about an hour playing with it I was able to get a very nice S21-style plot out of the thing. The general shape and the bandwidth lined up very closely to the output of the VNA, which was exciting. The main discrepancy was the frequency of the passband was shifted about 4 kHz downward, but I’m not sure I’m really that concerned for now.
I was very impressed that the shape of the passband was clearly repeated by the modelling tool, with that second dip being a little deeper and wider than the first. I was also excited by the fact that it came up with a 3 dB bandwidth of 455 kHz, only 16 Hz different from measured!
I’m not sure if this is the best or most intuitive filter analysis software around. I thought there was another package made by Wes Hayward or someone. But given that I only needed about an hour to create a model and input my crystal parameters and generate a useful plot, I’d say that it did the job.
So I will continue on building this kit. I wanted to refresh my mind about these formulas and the filter analysis generally, and at least I got to play with one of the software tools out there on the net. I’m afraid I’m going to have to rapidly finish up the rest of this kit though with very little analysis, as I’ve run out of “play time” this summer and will very soon be inundated with studies and projects. I hope to post a few more blogs on progress on that front.