I received a question today but when I tried to answer it the question does not show so I will just post the answer and hope that the person who asked the question will see it. The question asked why sometimes certain capacitors are used and other times they are not needed. The design of the filter with the way I did it needs the crystals to have slightly different frequencies but expects that you have crystals that are all the same frequency. If you had access to exactly the correct frequencies the series capacitors that slightly adjust the effective crystal frequencies could be eliminated. So after I measure a group of crystals I try to use them so that they don't need to be "pulled" as much which means that usually a capacitor or 2 can be eliminated. See the comment that Alexej Svirid made. Larry, W0QE

I've watched the video 3 times now. There is a lot to absorb from the multiple interactive parameters, and of course the weights attached to the importance of each of them. Then include the locations of the various crystals in the filter itself and well, there's a lot to juggle. But I'm getting there. Thank you for creating this video so I can keep watching it until I fully 'grok' what's going on, what matters, when and where. That's something I never had in school. It was 'once and done' with everything. I see people on KZread trying to build crystal filters with little care or attention to detail, with the expected disastrous results. Then I see people going through an excruciatingly painful degree of effort to perfectly match all possible parameters of all of the crystals, leaving them with mountains of extra 'useless' crystals. What I see from watching this video is what matters, and what does not, and how to design a very good crystal filter with a modest assortment of crystals, a real-world situation, something none of the other videos actually presented. Yes, I have a box of over a thousand similar crystals, but no, I don't want to have to sort through ALL of them in order to build 1 good bandpass filter. So thank you.

@w0qe

2 жыл бұрын

John, I have taken apart several commercial crystal filters over the years and measured the crystals and often they are not the same. For hobbyist being able to "intelligently" use crystals that are slightly off frequency is quite satisfying. Glad you enjoyed the video. Larry, W0QE

@johnwest7993

2 жыл бұрын

Larry, What immediately came to my mind as I watched you select crystals and capacitor values was curiosity about how crystal filter manufacturers do it. Because I was certain that they didn't sort 4 to 8 near identical crystals from a bag of 100 and toss out the rest. I suspect it's a more automated version of what you were doing, but of course that would be highly proprietary company information. That's OK. Just as long as I get up to speed on how to do it manually, I'm happy. Like you, I think crystal filters are a valuable tool for the lab or even out and about with the ham gear. John - ad0wo

Thank you Larry. Great Introduction and simulation.

Thank you for the comprehensive video EP2ADC

@w0qe

2 жыл бұрын

Glad the video was useful to you. Larry, W0QE

I like the decision tree you made for choosing the crystals. I just got a bunch for cheap and need to do this, want to build an IF crystal ladder filter

Fantastic information Larry, well explained and comprehensive. What are your thoughts on running the crystals in for a period before carrying out matching?

@w0qe

7 жыл бұрын

Glad you liked the video. I don't have a lot of data on this topic but I have made some measurements that might help. My usual measurement setup (AIM 4170) measures the crystals at -16dBm (25uW). If I make repeated measurements of the same crystal I see the series resonant frequency change by maybe 3Hz at 14MHz. I assume this tiny change is due to heating and if I wait 10 minutes the frequency will be within 1Hz of the initial measurement. I have also measured a crystal, then driven it at the series resonant frequency with 10mW, and then re-measured it and saw a change in frequency in the order of 30-50Hz. Again waiting for the crystal to "cool" down returns the series resonant frequency to withing 1-2Hz of the original. A 30-50Hz ambiguity in the crystal frequencies will have a very minimal effect on the overall crystal performance. For some notch filters I built I wanted to notch a +20dBm signal so I paralleled the first 4 crystals on the generator side of the filter. Each of these crystals could be expected to dissipate about 15mW and I the filter worked fine and any heating that the crystals experienced did not cause long term changes. The only data on crystal aging that I have is that the overall response of band-pass filters that are more than 5 years old has not really changed based on network analyzer sweeps. If it makes you more comfortable then I see nothing wrong with "running the crystals in" for a period but remember that the goal is actually drive the crystal where some small amount of heating will occur so you will need to set the generator specifically for each crystal and be careful not to apply too much power. An easier way might be to just put the crystals in a tray in the oven and heat them up to 50-60 deg. C for an hour and let them cool down for a day before measuring them.

Thank you Larry. These videos are really illustrating and helpful for a RF beginner like me. I am wondering if you could post the links for how to measure the crystal parameters. Thanks!

@w0qe

7 жыл бұрын

I use the AIM 4170 or AIM 4300 analyzers to measure the crystals. The software automates the crystal parameter measurements as I mentioned. Also www.w0qe.com/Projects/crystal_bandpass_filters.html has more info. The crystal parameters can be measured manually with a network or impedance analyzer. Making an impedance measurement at about 75% of the series resonant frequency gives the parallel capacitance. Measuring where the impedance is the lowest and is resistive will give the crystal series resistance and the series resonant frequency. By knowing the series resonant frequency you know the product of the motional inductance times the motional capacitance. Then measure the impedance about 1kHz above the series resonance and you get an impedance that is inductive and from this inductive reactance you can calculate motional inductance and capacitance. When you are done the parallel capacitance is almost always about 220 to 225 times the motiional capacitance.

@wbai1173

7 жыл бұрын

Thanks Larry. It helps!

Larry, is there a method to move the center frequency of the filter? I have built several ladder filters in the past. The problem I run into is the CF is different for each filter when built from the same groups of crystals. If I build one with a ~2.8 kHz BW, it will have a certain CF. If I then build one with a 700 Hz BW, the CF shifts by 2 or 3 kHz. That means if I am tuned to a CW signal using the 2.8 kHz filter and switch to the 700 Hz filter, the signal disappears. This can be handled in software but it would be better if the CW and SSB filters had the same CF like the commercial filters do. Thank you for the videos. They are all very well presented.

@w0qe

2 жыл бұрын

Ray, If you look at the Crystal Ladder Filter Calculator you will see that for any filter that is designed with identical crystals that some of the stages have frequency offsets that are set by the tuning (series) capacitances. These capacitors effectively change the frequency of the crystal but they lower the effective Q of the crystal. You could just use crystals for slightly different frequencies for the offset stages but this is not practical unless you are having the crystals manufactured specially for you. When you measure crystals you will get a range of frequencies and it is wise to pick what crystal to use for each stage to minimize the need to tune the crystals since the Q is reduced more for larger frequency shifts. However all the shifts are in the same direction and this is why the narrower band filter filter has a slightly different center frequency that a slightly wider band one does. Knowing this in advance and having a large group of crystals to sort thru you might be able to get the 2 center frequencies to be the same but I think it is probably to just shift the oscillator frequency slightly which will accomplish the same effect. Does this make sense? Larry, W0QE

@raybuck9675

2 жыл бұрын

@@w0qe Yes, it does. Mouser sells an IQD crystal at 9.000 MHz, The also sell one that is 1500 Hz above and one that is 1500 Hz below the 9.000 MHz one. Obviously these crystals are made for SSB use. If I remember correctly, with the last filters I built, the narrow 700 Hz one was about 2 kHz below the wider 2.7 kHz one. That was with crystals marked 13.500 MHz. I think I will order 20 of the 9.000 MHz one and 20 of the 9.0015 MHz one for testing purposes. If I can find 5 or 6 of each that are close to those two frequencies and use the 9.0015 one for the narrow filter I may be able to get both filters very close to each other. The shift with 9 MHz crystals should be slightly less than the 13.5 MHz ones. Shifting the oscillator works and that is what I had to do with the 13.5 filters. It just involves a little extra circuitry to let the micro know whether you are in SSB or CW mode.

I noticed that loading base-trk.ss in SimSmith 18.4 does not show the graphs. It does work in SimSmith 18.3 though. Is there a change which can be made to get it working in newer versions of SimSmith?

@w0qe

2 жыл бұрын

Rob, There is bug in SimSmith ver 18.5 which happens if the value of the Load is zero. Notice the V,I=NaN,NaN above the components where NaN means "not a number". Until this is fixed just change the Load value to 1n or some other extremely non-zero value and all will be fine. Larry, W0QE

Bla, bla, bla !!! Quit wasting my time. Make your point. Present the filter results and shut up.

@w0qe

2 жыл бұрын

If I'm wasting your time then you are a fool to keep watching. I make absolutely $0 from KZread so why not just Larry, W0QE

@johnwest7993

9 ай бұрын

@@w0qe, Larry, I cannot believe the offensive idiots who sometimes pop up in KZread comment sections. I hope you can ignore them. I can only think they are very drunk, but not very bright. They are clearly incapable of following a learning process or even a train of thought. Their abilities begin and end at clicking and grunting. This video has been the most helpful, complete, and insightful crystal filter design video on all of KZread. I know. I've watched many. This video is better than a university class in such design, and it's free, thanks to you. Yet some fool still makes such a comment. I am reminded of the story about "casting pearls..."