EMRFD Message Archive 4105
Message Date From Subject 4105 2010-01-31 04:59:46 ha5rxz Ladder Filter Construction As it is thick snow here in Budapest I have spent the weekend constructing ladder filters. I am now puzzled.
Each filter is constructed using 4.9152 MHz crystals, I used these because I have about 500 of them and it makes getting a matched set easier. The 500 Hz CW filter has a measured center frequency of 4.915455 MHz and the 2.7 KHz SSB filter has a measured center frequency of 4.916263 MHz, a difference of 808 Hz. So.....
1) I expected to see a center frequency of 4.9152 MHz in both filters, why is this not the case?
2) Both filters are the same type (Cohn minimum loss) although the capacitor and termination values are different because of the width requirements. Why do I have that 808 Hz difference in center frequency?
Note that all crystals were matched to +/- 10Hz before I started. What am I doing wrong here?
HA5RXZ4107 2010-01-31 06:02:20 Roelof Bakker Re: Ladder Filter Construction Hi,
When you simulate ladder filters with GPLA, you will find that the
frequency of the lower edge of the filter will stay the same. In a wider
filter the frequency of the upper filter edge will move up.
So your findings are correct and nothing is wrong.
This phenomeneon will call for a different CIO frequency per bandwidth in
regards to the lower sideband though.
73,
Roelof Bakker, pa0rdt4108 2010-01-31 06:05:49 Chris Trask Re: Ladder Filter Construction >Do you have a URL that secribes the Cohn ladder filter? I'm not familiar with this, though I should be. I've run across the same problem with my approach to symmetrical crystal filters described by Zverev, and it may be related.
> As it is thick snow here in Budapest I have spent the weekend
> constructing ladder filters. I am now puzzled.
>
> Each filter is constructed using 4.9152 MHz crystals, I used
> these because I have about 500 of them and it makes getting a
> matched set easier. The 500 Hz CW filter has a measured center
> frequency of 4.915455 MHz and the 2.7 KHz SSB filter has a
> measured center frequency of 4.916263 MHz, a difference of
> 808 Hz. So.....
>
> 1) I expected to see a center frequency of 4.9152 MHz in both
> filters, why is this not the case?
>
> 2) Both filters are the same type (Cohn minimum loss) although
> the capacitor and termination values are different because of
> the width requirements. Why do I have that 808 Hz difference
> in center frequency?
>
> Note that all crystals were matched to +/- 10Hz before I started.
> What am I doing wrong here?
>
Chris Trask
N7ZWY
WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/4109 2010-01-31 06:21:12 KB9BVN Re: Ladder Filter Construction http://www.cliftonlaboratories.com/cohn_crystal_filter.htm
Looks like a good place to start.
73 de KB9BVN
Brian Murrey
----- Original Message -----
4110 2010-01-31 06:47:02 Chris Trask Re: Ladder Filter Construction ><
> >
> > > Each filter is constructed using 4.9152 MHz crystals, I used
> > > these because I have about 500 of them and it makes getting a
> > > matched set easier. The 500 Hz CW filter has a measured center
> > > frequency of 4.915455 MHz and the 2.7 KHz SSB filter has a
> > > measured center frequency of 4.916263 MHz, a difference of
> > > 808 Hz. So.....
> > >
> > > 1) I expected to see a center frequency of 4.9152 MHz in both
> > > filters, why is this not the case?
> > >
> > > 2) Both filters are the same type (Cohn minimum loss) although
> > > the capacitor and termination values are different because of
> > > the width requirements. Why do I have that 808 Hz difference
> > > in center frequency?
> >
>
> >Okay, I thought that might be the style of crystal filter. Someone else earlier posted that the problem is due to the edge of the passband rather than the centre is defined by the crystal frequency, which is correct. I ran into this problem with my filters, and I found that I could correct it slightly by using parallel vs. series resonant crystals.
> > Do you have a URL that secribes the Cohn ladder filter? I'm not
> > familiar with this, though I should be. I've run across the same
> > problem with my approach to symmetrical crystal filters described
> > by Zverev, and it may be related.
>
>
> http://www.cliftonlaboratories.com/cohn_crystal_filter.htm
>
> Looks like a good place to start.
>
My symmetrical filters use half-lattice sections with identical crystals. The centre frequency is always above the crystal series resonant frequency, unless I want to go to the trouble of adding variable inductors to pull them, and this thread has given me an idea that will simplify that.
Chris Trask
N7ZWY
WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/4121 2010-01-31 09:32:54 ha5rxz Re: Ladder Filter Construction I think I see my error. I assumed that the center frequency of the filter passband would be the series resonant frequency of the crystal. My biggest problem now is that because of the two different center frequencies of the filters my SSB/CW board outputs different frequencies depending4123 2010-01-31 11:40:40 Chris Trask Re: Ladder Filter Construction >mode.
> I think I see my error. I assumed that the center frequency of the filter
> passband would be the series resonant frequency of the crystal. My biggest
> problem now is that because of the two different center frequencies of the
> filters my SSB/CW board outputs different frequencies depending on the
>That could be remedied by pulling the BFO frequency. I use a single
crystal filter for LSB/USB (not very sharp) and then pull the BFO for one or
the other. Cuts down the number of filters.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/4135 2010-02-01 09:01:51 Harold Smith Re: Ladder Filter Construction The only time I tried making multiple bandwidth filters using nominally
identical crystals, I got around the different center frequencies by
measuring lots of crystals and then using the matched set with the lowest
frequency to make the wide filter, and that with the highest frequency to
make the narrow filter. As I recall, I ended up with a pair of filters that
could make use of the same BFO. It was a lot of work, sorting that many
crystals (a couple hundred, I think) and if I were doing it over, I would
settle for shifting the oscillator frequency as needed.
de KE6TI, Harold Smith
4142 2010-02-02 02:11:19 drmail377 Re: Ladder Filter Construction I have simulated three Cohn capacatively coupled 8-pole crystal filters. One each at 500, 2000, and 6000 Hz design center frequency. The crystals are 5.0688MHz and are assumed to be identical. The bandwidths mentioned above are as entered into the AADE Filter Designer software package. I did not increase them to tweak the simulated bandwidth as is typically required.
Findings: As expected, the wider the filter bandwidth, the lower the filter center frequency. The 500Hz filter center is 2.80kHz below the 6kHz filter. The 500Hz filter center is 1.09kHz below the 2kHz filter.
According to the help file in AADE Filter Designer, using untuned inductive instead of capacitive coupling gives a more symmetrical response and shifts the frequency response down slightly. So in theory one might use inductive coupling instead of capacitive for the narrow filter and vice-versa for the wider filter. I have not simulated this. Plus I have three filters in my design, not two.
Another technique I've heard about but not tried is putting a very small variable capacitance in parallel with the crystal. This will allow you to shift up the holder capacitance, commonly referred to as Cp. Cp normally ranges in the ones of pF, so we're talking about a small gimmick here.
Increasing the value of Cp is supposed to shift the crystal filter frequency up. I would think that at some point with this method the the passband and skirts may become asymmetrical though. Also, I'm not sure how far you can shift if you're only increasing the value of Cp. It seems the filter shifts down much better if you can reduce Cp somehow. If you read the help file in AADE Filter Designer you can see an example where real vs. simulated results are shifted in frequency due to a "fudged" Cp value; in the help-file case the filter response is shifted down when Cp is reduced.
Another approach may be to use Dishal USB and LSB desgns and some other design for CW. AADE Filter Designer can handle the Dishal case as well. The subject of Dishal filters is covered in the AADE Filter Designer help file.
As for shifting the carrier BFO. You might want to design your radio such that when you switch modes (CW/USB/LSB), the mode switch flips relays to drop in preset trimmer caps that pull your BFO. There are techniques for switching the BFO pulling caps with PIN diodes or switching bias voltages to a varactor diode. I suggest you stick with relays, anytime voltages get around a PIN or especially a varactor in an oscillator, unless the voltage is very clean and stable, you might mess up the stability and/or phase noise performance of the oscillator. Besides the mode switch isn't changed that often, so relays won't be subject to too many cycles.
Sometimes you see BFO's that are made variable with a varactor. Again, I worry about this technique without ensuring your power supply is stable and noise free. A better approach might be to panel mount the BFO and use an small air-variable capacitor. Air-variables are getting harder to find, but it seems there are still plenty of small value parts around.
Best 73's, David WB4ONA