Low Pass Chebyshev Filter and 7MHz oscillator
I thought I will give a try to building a 8MHz Chebyshev filter to use together with a 7MHz oscillator. After checking the design for a band pass filter, it looked pretty complex to me and I decided to see if I can build something simpler.
The Low Pass Chebyshev filter seemed easy enough to start the project at 9:30PM, on a monday evening. I’ve chosen a filter with a good enough rejection so that people will not complain, so a 8-th order filter will it be.
Here you can see the screen from the AADE filter software:
Note that the filter analysis by the same software actually indicates the bandpass ripple of over 2dB, despite of my original selection a 1dB bandpass ripple. Probably ripple is defined as half of the maximum variation of the level over the frequency range.
The filter consists of 4 capacitors and 4 inductors. As usual, at least for me, it is much easier to deal with capacitors rather than inductors. Checking the values, I decided I will use a 680pF together with a 180pF for C2. For all the others, I will use styroflex 1nF, which I have plenty, together with whatever I can find around. The 1nF styroflex capacitors were manufactured in Hungary and they have an excellent accuracy and look very good.
First thing, since I am a bit stingy, I didn’t want to use my precious PCB board for building such a trivial piece of electronics, so looking around, I found my failed attempt for a PCB for the oscillator in the BITX20 project. The copper was missing in some places and it was running very thin in other places, but I thought it will fit well in the quick and dirty category.
Next, it was time to deal with the inductors. I considered using some ferrite beads, one turn through it is enough to give 0.5uH or so, while two turns is over 2uH, so it is already too much. I decided to build them on air. I found some enamelled wire, which I used in a coil attempt for a metal detector. It has a thickness enough to build coils on air. I used a thick Philips screwdriver to wind the coils. I’ve done it all by hand. First I had to see what length of wire do I need for building coils around 1.2uH mark. In a few attempts, I found that from a piece of 60cm in length, I get a coil of 1.49uH, while a 50 cm piece of wire gives me 1.18uH (with this occasion, I already got my L5 coil).
After building all the coils, I thought I will solder just the 680pF and three of 1nF capacitors and the coils and see how the shape looks like. I didn’t expect to see any decent results, but I was curious to see how bad it will look like.
Well, in my opinion, it looked surprizingly good. Since it is the first Chebyshev filter I start building, I didn’t know how much better can it get after I add all the other capacitors.
After that, I soldered all the other caps and the result is here:
Next it is an image for wider frequency span. It doesn’t look so good at higher frequencies, which probably travel from a coil to the next one. The last part improves significantly when I touch some of the coils, probably when my hand acts like a shield.
For a closer look at the linearity of the filter, I changed the level range. I would be more pleased to have a 1dB nonlinearity at the top of the filter, but maybe I shouldn’t have such high expectations for a job done in such a hurry.
After all, a 2dB nonlinearity doesn’t look so bad!
Next, I decided to test the latest oscillator I built with this filter.
Here is the spectrogram. I was careful this time to adjust the resolution bandwidth (RBW) and video bandwidth (VBW) to lower the noise floor of the SA.
The harmonics or nonharmonics are attenuated by over 75dBc, which is probably enough to satisty the US radioamateur broadcast standards 🙂
It’s now 12:10AM. Since start to end, building the filter, taking the photos, processing the photos and creating this page, it took 2 hours 40 minutes. I wonder what did I miss on TV? 🙂
Update: since then, I was curious to see what capacitor and inductor values a 8MHz Low Pass Filter with a 0.5dB ripple would have. Well, it turns out that most capacitors which I initially soldered (680pF and 1nF) were almost spot on for a 0.5dB ripple filter. This also explains the surprisingly good ripple I obtained during the preliminary testing, with capacitors of lower values than required. Of course, a lower ripple also implies a softer roll-off.
Next, I wanted to see how LTSpice simulates the Chebyshev Low Pass Filter 8MHz, 1dB ripple.
Not surprizingly, the results are very similar.