14355	2017-09-27 10:57:00	rcbuckiii	10 KHz wide filter
Message	Date	From	Subject
If I have an IF frequency of 9 MHz, what would be the most practical way to obtain a 10 KHz bandpass (or at least 8 KHz) for reception of AM broadcast signals? I would think a monolithic crystal filter would be the only way to do this. However, the companies who make these want to sell at least 100 pieces. I see Mouser is now selling 9.000, 8.9985, and 9.0015 MHz crystals. Could a half lattice filter using the 8.9985 and 9.0015 MHz crystals possibly be built that would approach 6 KHz bandwidth? Ray AB7HE
14356	2017-09-27 11:20:22	cwfingertalker	Re: 10 KHz wide filter
Ray, Inrad makes two filters for AM at 6000 hz wide. One on 9 MHz and one 8.830 MHz. Check them out. Bill N7EU
14357	2017-09-27 11:20:26	cwfingertalker	Re: 10 KHz wide filter
Ray, Inrad makes two filters for AM at 6000 hz wide. One on 9 MHz and one 8.830 MHz. Check them out. Bill N7EU
14358	2017-09-27 11:31:40	Thomas S. Knutsen	Re: 10 KHz wide filter
Hmm, lets do some math: For a BW of 10KHz at 9MHz you need a loaded Q of 900. Thats above what you can expect from a LC filter. Using regular crystals should be possible, A quick back of the envelope calculation for Cohn gives capacitors around 10pF for some 9MHz crystals I measured some years back. Butterworth and Chebychew should be possible, and if you can accept some filter loss, shunting the crystals with some resistors for a ladder network should be possible. Using a crystal filter calculator like the AADE or Xlad should give you some pointer to do some experiments. Start by measuring the crystals you want to use, and running them in a simulator. Half a hour in front of the PC can tell you if your approach is viable or not, before going to the bench to protype it. As a alternative, you could try some ceramic resonators, either in a ladder or lattice approach. Those should have lower Q than crystals. Your suggestions for a Lattice filter should be fine, but you may find that it has some attenuation between the 2 crystal frequencies. In my experience, the termination impedances is what defines how wide the bandwith in a filter is, more so in a lattice than a ladder filter. 73 de Thomas LA3PNA. 2017-09-27 19:56 GMT+02:00 rcbuckiii@yahoo.com [emrfd] <emrfd@yahoogroups.com>: If I have an IF frequency of 9 MHz, what would be the most practical way to obtain a 10 KHz bandpass (or at least 8 KHz) for reception of AM broadcast signals? I would think a monolithic crystal filter would be the only way to do this. However, the companies who make these want to sell at least 100 pieces. I see Mouser is now selling 9.000, 8.9985, and 9.0015 MHz crystals. Could a half lattice filter using the 8.9985 and 9.0015 MHz crystals possibly be built that would approach 6 KHz bandwidth? Ray AB7HE
14359	2017-09-27 14:05:47	rcbuckiii	Re: 10 KHz wide filter
Bill, This is a home brew receiver. The Inrad filters cost more than the rest of the components combined. Thomas, AADE and Xlad seem to indicate that 6 KHz bandwidth may be the upper limit. I'll have to wire up the crystals based on the XLAD data. I can then scan them with my DSA-815TG and see what the results are. I tried using the Dishal203 program but 3.8KHz was the max I could get. Another idea I have had is to use one of the standard 10.7 MHz monolithic parts. Since my LO is a Si5351, I could program it so when AM mode was selected the LO would be 10.7 above the frequency. Unfortunately my mixer stage has a DTC filter following the ADE-1. I would have to add a couple of relays to handle switching between the 9 MHz DTC and a 10.7 DTC. The SSB and CW designs were fairly easy to design. The bandwidth of the SSB one was less than calculated but I think it will be OK. Ray AB7HE
14360	2017-09-27 19:05:59	Bill Carver	Re: 10 KHz wide filter
The max bandwidth limitation in a ladder filter is caused by the parallel resonance of the crystal on the low side with "holder" capacitance. You can't make a LSB (series xtals) ladder filter wider than the pole-zero (parallel-series) resonant frequencies of the crystal itself. There are ways to get around this, to "cancel" the parallel capacitance of the crystal. Wes Hayward presented one method a few decades ago (or maybe 3 decades!): put an inductor in parallel with the crystal such that it's parallel resonant at the filter center frequency. This does introduce some spurious resonances far from the desired filter frequency which could be a problem. Another approach that can be used for xtals at the input and output ends of the filter: use a center tapped transformer, one side of the center tap driving the xtal, the opposite phase/side driving a "neutralizing" capacitor throught a capacitor equal to the holder capacitance (say a 1-6 pF trimmer). I've found the parallel resonating the interior xtals with inductors, then "neutralizing" the holder capacitance of the xtals at the input and output end of the filter helped reduce those spurious resonances of the coils and interior crystals. When either/both of these techniques has been done, you do the desired bandwidth design using a value of zero for holder/parallel capacitance and should be able to get the 10 KHz bandwidth you want. An alternative is to use crystals with much more than 10 KHz between the series and parallel resonant frequencies in the first place (it needs to me more than just slightly more or the notch at the parallel resonant frequency will "tilt" the filter top, and the filter will be seriously narrower than designed). Just use a higher frequency crystal. I'm not sure how high you might have to go, if you have 16 MHz, 20 MHz, or other higher frequency crystals laying around you can put them in your crystal test fixture to get the series and parallel frequencies. Bill W7AAZ
14361	2017-09-27 20:22:52	bwbangerter	Re: 10 KHz wide filter
Search ebay for "crystal filter" and look at the board with three KVG 9 MHz filters attached, XF-9 S42, 43 and 44. Designed for 9.000 MHz bfo, one AM, one USB, one LSB. All for under $40 US. Ben, K0IKR
14362	2017-09-27 21:17:58	Ashhar Farhan	Re: 10 KHz wide filter
The easier option is to move up the crystal (IF) frequency. At 20 MHz, 24 Mhz and 25 Mhz, using the microprocessor crystals, it is trivial to get that bandwidth. Using the Cohn configuration, experiment with various capacitor values (start with 22pf) and termination values (putting resistors of various low values) between the 50 ohm ports at either ends. It will show high loss because of the resistors but they are a quick way to experiment with different values. I got 15 KHz with 25 MHz at 50 ohms (lucky me) using five crystals. - f
14363	2017-09-28 08:57:34	Jim Strohm	Re: 10 KHz wide filter
@Farhan, moving the xtal IF may not be a practical option if the OP is trying to fit some older gear with a 9 MHz IF. Certainly if there are no restrictions on where the IF _could_ be, it makes excellent technical sense to put it where parts are cheap and readily available, and require the least amount of tweaking. It's kind of how hydrogen-fueled hybrid cars with both a hydrogen fuel cell and a hydrogen-fueled internal combustion motor would make for great driving power and clean technology -- but we're still on a gasoline and lead-acid battery infrastructure. Clear-field development allows for great technological innovation. You are blessed to live in a nation with better opportunities for clear-field development. Here, we're saddled with century-old infrastructure in d****d near everything. About the only true innovation in the last century I can think of is -- gypsum wallboard, which replaced plaster-and-lath interior wall construction. 73 Jim N6OTQ
14364	2017-09-28 09:05:41	rcbuckiii	Re: 10 KHz wide filter
Thanks everyone for the comments. I measured a dozen crystals and entered the average results into Dishal. It appears I can get the 10 KHz BW with 6 crystals if I can accept about 2 dB ripple. I think that will be fine for AM reception. Ben, I found the Ebay listing. There are a lot of them available from several sellers. I guess the manufacturer of that board must have dumped tons of them on the surplus market. I think I will order a board just for experimental purposes. The SSB filters are a little wide but would be good enough for casual listening. Ashhar, the higher IF is a good idea. Unfortunately I already have several stages already built based on the 9 MHz frequency. An IF of 21.4 would be good because there are standard crystals and monolithic filters available for that frequency.
14365	2017-09-28 11:33:49	Ashhar Farhan	Re: 10 KHz wide filter
jim, you would be aghast at what goes on in my country. the electric cars are actually taxed more than the carbon fuel cars, 'to protect the industry'! the good thing about my country, (make it 'about my city') is that i can drive down for 30 minutes, walk into a store and buy off a reel of MMBT3904s or an arduino and a magnetometer and be back home in the next 30. there is crazy amount of regular electronic stuff that is available over the counter at really ridiculous prices. i bought a 50 watt peltier cooler for three dollars, an arduino nano board for two dollars, just this evening. - f
14367	2017-10-01 15:48:24	rcbuckiii	Re: 10 KHz wide filter
OK, I finally got around to building the 10KHz filter. I used Dishal to design and tune the filter. I designed for a 10 KHz passband and 1 dB ripple. The center design frequency was 9.000000 Mhz. I used 6 crystals with a frequency spread from 8.999973 to 9.000006 MHz. The filter impedance is 1256 ohms which I matched at both ends with a L-network. Sweeping the filter with the DSA-815 shows it to not be as designed. There is more ripple in the passband. The maximum ripple is -4.1 dB at one point. There are 3 other points where the ripple is -1.28, -2.18, and -2.3 dB. The low side -3 dB point is at 9.000561 Mhz and the high side -3 dB point is at 9.010277 MHz resulting in a 9716 KHz spread. The -6 dB points are 9.000272 and 9.010527 MHz. Maximum attenuation is around -45 dB on both the low side and high side. The difference between design and actual performance I'm sure is due to component tolerances. I could probably clean it up somewhat by replacing the fixed caps with small trimmers and adjusting them while watching the SA. But is it worth the trouble? Will the ripple be acceptable for AM reception? One of the bands in the receiver will receive WWV at 10 MHz. I'm thinking that if the one second tones sound normal from WWV, it will be OK for broadcast reception. It will be a couple of more weeks before I have a working receiver to test that theory. Comments welcome. Ray, AB7HE
14372	2017-10-02 00:41:12	Ashhar Farhan	Re: 10 KHz wide filter
While desiging the filters, i use series.resistors.to.get strong driving and terminating impedances. The ripple.is controlled by the terminal impedance at the cost of shallower skirts. - f
14374	2017-10-02 08:25:24	rcbuckiii	Re: 10 KHz wide filter
Ashhar, I am going to temporarily put a couple of trimmer caps in place of the series elements. I'm interested in what effect it will have on the center frequency of the filter. It was designed for a 9.000 CF, but is over 5 KHz higher when swept. I also noticed the coax shield wire had come loose from the filter input connector. When I soldered that back in place, attenuation dropped to over 70 dB. That is about the limit of the DSA-815. Ray AB7HE
14375	2017-10-02 08:42:09	Ashhar Farhan	Re: 10 KHz wide filter
If you reduce the resolution bandwidth, the noise floor will drop. That will give a slower and a more detailed scan. I scan the crystal filters at the lowest possible rbw. - f
14377	2017-10-02 16:04:12	rcbuckiii	Re: 10 KHz wide filter
After several hours of playing with trim caps I have settled on this design: - 3dB low side 8.999706 MHz - high side 9.006380 MHz This is 6674 Hz wide - CF 9.003043 MHz -6dB low side 8.999356 MHz - high side 9.006893 MHz The filter is essentially flat except for a -2.53dB dip at 9.004233 MHz. I was trying to get a wider response on the high side but it was a losing battle. Expanding the high side bandwidth increased the ripple a couple of dB at 3 other points toward the low side. I'm sure the 6674 Hz band width will provide good AM reception. This has been a learning experience. Now I have to design the 2.7 KHz SSB filter and 600 Hz CW filter. They "should" be easier than the AM one. Ray AB7HE
14378	2017-10-03 09:18:41	peter_dl8ov	Re: 10 KHz wide filter
All of this makes me appreciate the work done by KK7B on the R2pro / T2pro. I abandoned using crystal filters years ago and I've never looked back. Peter DL8OV