|
I recently bought an Icom 705 and have been stunned by the performance of twin tunable pass band filters. I want to duplicate that using analog filters. As it happens, 5 MHz HC-49/S crystals are less than a dime with other frequencies available at similar prices. But 5 MHz provides some interesting options in radio design. Particularly in the self diagnose and repair department.
Key concepts:
a 5 MHz IF allows using WWV for very precise calibration and alignment in conjunction with a noise source
the required frequency shifts are very small relative to Fp & Fs
varactors should easily perform the shifts
a nanoVNA will measure the xtal to provide the values needed to match that xtal
a program will calculate the optimal choice of xtals and capacitors & varactors from small pools (10-20) of parts to meet design goals with manual trimming as needed. A pair of 4 pole varactor tuned filters shouldn't be that hard.
Why go to all the problem of matching filters only to have a fixed frequency filter? Worst case use a latching DAC and MSP430 to set the varactors. All a question of current drain.
Have Fun!
Reg
|
Hi! I am puzzled about the fact that using 2 cascaded, overlapping filters results in less ringing than a single CW filter with same resulting bandwidth. Have you experienced that or is it a guess?
I suppose a good RF simulator could give an idea of the results making a time domain simulation.
BTW, I don't think 5.0 MHz is a particular option, one doesn't need having WWV accuracy to align a filter, in special a shiftable filter. 9 or 10MHz should be easier to shift because the percent change is smaller.
Have success!
Daniel Perez LW1ECP
El viernes, 12 de mayo de 2023 19:19:23 ART, Reginald Beardsley via groups.io <pulaskite@...> escribió:
I recently bought an Icom 705 and have been stunned by the performance of twin tunable pass band filters.? I want to duplicate that using analog filters.? As it happens, 5 MHz HC-49/S crystals are less than a dime with other frequencies available at similar prices.? But 5 MHz provides some interesting options in radio design.? Particularly in the self diagnose and repair department.
Key concepts:
a 5 MHz IF allows using WWV for very precise calibration and alignment in conjunction with a noise source
the required frequency shifts are very small relative to Fp & Fs
varactors should easily perform the shifts
a nanoVNA will measure the xtal to provide the values needed to match that xtal
a program will calculate the optimal choice of xtals and capacitors & varactors from small pools (10-20) of parts to meet design goals with manual trimming as needed.? ? A pair of 4 pole? varactor tuned filters shouldn't be that hard.
Why go to all the problem of matching filters only to have a fixed frequency filter?? Worst case use a latching DAC and MSP430 to set the varactors.? All a question of current drain.
Have Fun!
Reg
|
"Ah, Grasshopper, It is good you asked this question." ( If you'll forgive a KungFu paraphrase.)
All the equations governing the time and frequency behavior of filters are well defined.
The experience is the IC-705. All the rest is 40 years of DSP in the oil industry and an active interest in low cost, high performance QRP CW rigs.
A bell is a narrow band resonator. A concrete block is a broadband resonator. Bell goes "bong......". Block goes "clunk".
This is intrinsic to the mathematical relationship between time and frequency. For most of my life I thought that was the end of the story. And I worked at the reflection seismology research level. I could have made a *killing* if I'f figured this out. But so far as I know, no one in the community that invented DSP thought of it.
The fundamentals are simple.
The time domain response of a filter is dependent upon the width of the filter, not the input signal. If you output from the first filter a segment that only has 50 Hz of overlap with the 2nd filter yo will get a 50 Hz filter without any ringing. It's very fundamental and I am embarrassed neither I nor, to my knowledge, any of my coworkers thought of doing it.
Have Fun!
Reg
On Friday, May 12, 2023, 05:34:30 PM CDT, Daniel Ricardo Perez LW1ECP via groups.io <danyperez1@...> wrote:
Hi! I am puzzled about the fact that using 2 cascaded, overlapping filters results in less ringing than a single CW filter with same resulting bandwidth. Have you experienced that or is it a guess?
I suppose a good RF simulator could give an idea of the results making a time domain simulation.
BTW, I don't think 5.0 MHz is a particular option, one doesn't need having WWV accuracy to align a filter, in special a shiftable filter. 9 or 10MHz should be easier to shift because the percent change is smaller.
Have success!
Daniel Perez LW1ECP
El viernes, 12 de mayo de 2023 19:19:23 ART, Reginald Beardsley via groups.io <pulaskite@...> escribió:
I recently bought an Icom 705 and have been stunned by the performance of twin tunable pass band filters.? I want to duplicate that using analog filters.? As it happens, 5 MHz HC-49/S crystals are less than a dime with other frequencies available at similar prices.? But 5 MHz provides some interesting options in radio design.? Particularly in the self diagnose and repair department.
Key concepts:
a 5 MHz IF allows using WWV for very precise calibration and alignment in conjunction with a noise source
the required frequency shifts are very small relative to Fp & Fs
varactors should easily perform the shifts
a nanoVNA will measure the xtal to provide the values needed to match that xtal
a program will calculate the optimal choice of xtals and capacitors & varactors from small pools (10-20) of parts to meet design goals with manual trimming as needed.? ? A pair of 4 pole? varactor tuned filters shouldn't be that hard.
Why go to all the problem of matching filters only to have a fixed frequency filter?? Worst case use a latching DAC and MSP430 to set the varactors.? All a question of current drain.
Have Fun!
Reg
|
It's probably worth noting that the filter passband shape affects group delay. This was noted and illustrated in EMRFD.? The rapid transition from a flat passband to sharp skirt rolloff yields some impressively bad group-delay values.
I don't have the mathematical wherewithal to evaluate this, but it was worth a try on the workbench. Instead of tuning a filter for the fastest transition, I opted instead for a gentle skirt transition.? I could clearly hear the difference in the recovered audio. It seems that hobbyists have been 'building in' crummy audio with the sharp crystal filters.? The authors of EMRFD suggested that the so-called Gaussian filter mitigates the unwanted group delay peaking. Analog Design's 'Analog Filter Wizard' provides group delay plots that illustrate the perils of sharp transitions .? Eye-opening, and 'no free lunch' for this hobbyist.
- K1SWL
On Fri, May 12, 2023 at 8:27?PM Reginald Beardsley via <pulaskite= [email protected]> wrote: "Ah, Grasshopper, It is good you asked this question." ( If you'll forgive a KungFu paraphrase.)
All the equations governing the time and frequency behavior of filters are well defined.
The experience is the IC-705. All the rest is 40 years of DSP in the oil industry and an active interest in low cost, high performance QRP CW rigs.
A bell is a narrow band resonator. A concrete block is a broadband resonator. Bell goes "bong......". Block goes "clunk".
This is intrinsic to the mathematical relationship between time and frequency. For most of my life I thought that was the end of the story. And I worked at the reflection seismology research level. I could have made a *killing* if I'f figured this out. But so far as I know, no one in the community that invented DSP thought of it.
The fundamentals are simple.
The time domain response of a filter is dependent upon the width of the filter, not the input signal. If you output from the first filter a segment that only has 50 Hz of overlap with the 2nd filter yo will get a 50 Hz filter without any ringing. It's very fundamental and I am embarrassed neither I nor, to my knowledge, any of my coworkers thought of doing it.
Have Fun!
Reg
On Friday, May 12, 2023, 05:34:30 PM CDT, Daniel Ricardo Perez LW1ECP via <danyperez1= [email protected]> wrote:
Hi! I am puzzled about the fact that using 2 cascaded, overlapping filters results in less ringing than a single CW filter with same resulting bandwidth. Have you experienced that or is it a guess? I suppose a good RF simulator could give an idea of the results making a time domain simulation. BTW, I don't think 5.0 MHz is a particular option, one doesn't need having WWV accuracy to align a filter, in special a shiftable filter. 9 or 10MHz should be easier to shift because the percent change is smaller. Have success! Daniel Perez LW1ECP El viernes, 12 de mayo de 2023 19:19:23 ART, Reginald Beardsley via <pulaskite= [email protected]> escribió: I recently bought an Icom 705 and have been stunned by the performance of twin tunable pass band filters.? I want to duplicate that using analog filters.? As it happens, 5 MHz HC-49/S crystals are less than a dime with other frequencies available at similar prices.? But 5 MHz provides some interesting options in radio design.? Particularly in the self diagnose and repair department. Key concepts: a 5 MHz IF allows using WWV for very precise calibration and alignment in conjunction with a noise source the required frequency shifts are very small relative to Fp & Fs varactors should easily perform the shifts a nanoVNA will measure the xtal to provide the values needed to match that xtal a program will calculate the optimal choice of xtals and capacitors & varactors from small pools (10-20) of parts to meet design goals with manual trimming as needed.? ? A pair of 4 pole? varactor tuned filters shouldn't be that hard. Why go to all the problem of matching filters only to have a fixed frequency filter?? Worst case use a latching DAC and MSP430 to set the varactors.? All a question of current drain. Have Fun! Reg
|
*Everything* about the shape is reflected in every other aspect of the filter. TANSTAFL! One should never design a filter without examining the behavior in both time and frequency very closely.
My biggest concern at present is electrically tuning the center frequency. For a start I've designed some 5-7 order Butterworth BP LC filters. Now I need to sort and bin a large tote of RF chokes and build them.
I plan to use fixed caps for the first couple to test the shift and shape and then start fiddling with varactors and crystals once my 5 MHz xtals arrive.
Reg
On Saturday, May 13, 2023, 05:34:24 AM CDT, Dave Benson <davek1swl@...> wrote:
It's probably worth noting that the filter passband shape affects group delay. This was noted and illustrated in EMRFD.? The rapid transition from a flat passband to sharp skirt rolloff yields some impressively bad group-delay values.
I don't have the mathematical wherewithal to evaluate this, but it was worth a try on the workbench. Instead of tuning a filter for the fastest transition, I opted instead for a gentle skirt transition.? I could clearly hear the difference in the recovered audio. It seems that hobbyists have been 'building in' crummy audio with the sharp crystal filters.? The authors of EMRFD suggested that the so-called Gaussian filter mitigates the unwanted group delay peaking. Analog Design's 'Analog Filter Wizard' provides group delay plots that illustrate the perils of sharp transitions .? Eye-opening, and 'no free lunch' for this hobbyist.
- K1SWL
On Fri, May 12, 2023 at 8:27?PM Reginald Beardsley via <pulaskite= [email protected]> wrote: "Ah, Grasshopper, It is good you asked this question." ( If you'll forgive a KungFu paraphrase.)
All the equations governing the time and frequency behavior of filters are well defined.
The experience is the IC-705. All the rest is 40 years of DSP in the oil industry and an active interest in low cost, high performance QRP CW rigs.
A bell is a narrow band resonator. A concrete block is a broadband resonator. Bell goes "bong......". Block goes "clunk".
This is intrinsic to the mathematical relationship between time and frequency. For most of my life I thought that was the end of the story. And I worked at the reflection seismology research level. I could have made a *killing* if I'f figured this out. But so far as I know, no one in the community that invented DSP thought of it.
The fundamentals are simple.
The time domain response of a filter is dependent upon the width of the filter, not the input signal. If you output from the first filter a segment that only has 50 Hz of overlap with the 2nd filter yo will get a 50 Hz filter without any ringing. It's very fundamental and I am embarrassed neither I nor, to my knowledge, any of my coworkers thought of doing it.
Have Fun!
Reg
On Friday, May 12, 2023, 05:34:30 PM CDT, Daniel Ricardo Perez LW1ECP via <danyperez1= [email protected]> wrote:
Hi! I am puzzled about the fact that using 2 cascaded, overlapping filters results in less ringing than a single CW filter with same resulting bandwidth. Have you experienced that or is it a guess? I suppose a good RF simulator could give an idea of the results making a time domain simulation. BTW, I don't think 5.0 MHz is a particular option, one doesn't need having WWV accuracy to align a filter, in special a shiftable filter. 9 or 10MHz should be easier to shift because the percent change is smaller. Have success! Daniel Perez LW1ECP El viernes, 12 de mayo de 2023 19:19:23 ART, Reginald Beardsley via <pulaskite= [email protected]> escribió: I recently bought an Icom 705 and have been stunned by the performance of twin tunable pass band filters.? I want to duplicate that using analog filters.? As it happens, 5 MHz HC-49/S crystals are less than a dime with other frequencies available at similar prices.? But 5 MHz provides some interesting options in radio design.? Particularly in the self diagnose and repair department. Key concepts: a 5 MHz IF allows using WWV for very precise calibration and alignment in conjunction with a noise source the required frequency shifts are very small relative to Fp & Fs varactors should easily perform the shifts a nanoVNA will measure the xtal to provide the values needed to match that xtal a program will calculate the optimal choice of xtals and capacitors & varactors from small pools (10-20) of parts to meet design goals with manual trimming as needed.? ? A pair of 4 pole? varactor tuned filters shouldn't be that hard. Why go to all the problem of matching filters only to have a fixed frequency filter?? Worst case use a latching DAC and MSP430 to set the varactors.? All a question of current drain. Have Fun! Reg
|
I worked with high end Telemetry receivers at the factory. All tuned circuits were Lc or RC. The only crystals were for LO or system clock applications. The LC were aligned on HP Network analyzers. RC filters were Sallen-KKey, used to shape the baseband output. For testing these, I used a HP3325A and a Fluke 8921 True RMS Voltmeter that worked out to 20 Mhz. I could measure the -3dB point with ease, and quickly to test each of 16 computer selected filters
Telemetry needs very low phase delay, especially on very weak signals. We used a 70 MHz IF? which could be recorded to instrumentation recorders to allow recovery of multiple channels of frequency multiplexed data.
|
I'm a little confused. How is this related to implementing very narrow filters by using isolated sequential, partially overlapping filters? The Icom 705 will give single signal reception with stronger stations within 50 Hz. That blew my mind and I've been doing this stuff a very long time.
In modern professional work you use DSP, but for portable ham use, analog is much better if power is a concern.
Digital filters are my forte. I just need to know what you want done. Analog is a bit more ambiguous.
Have Fun!
Reg On Saturday, May 13, 2023, 06:45:54 PM CDT, Michael A. Terrell <terrell.michael.a@...> wrote:
I worked with high end Telemetry receivers at the factory. All tuned circuits were Lc or RC. The only crystals were for LO or system clock applications. The LC were aligned on HP Network analyzers. RC filters were Sallen-KKey, used to shape the baseband output. For testing these, I used a HP3325A and a Fluke 8921 True RMS Voltmeter that worked out to 20 Mhz. I could measure the -3dB point with ease, and quickly to test each of 16 computer selected filters
Telemetry needs very low phase delay, especially on very weak signals. We used a 70 MHz IF? which could be recorded to instrumentation recorders to allow recovery of multiple channels of frequency multiplexed data.
|
We introduced the first SDR based Telemetry system as well. Our analog designs had bandwidths down to 1 KHz, without crystal filters. Have fun, but between my work with Telemetry and Microwave TV relays, I take group delay quite seriously. I also dealt with it in designing Cable TV return systems.
I do have one question. How will you prevent the IF signal from modulating the tuning voltage on your Varicaps?
On Sat, May 13, 2023 at 8:31?PM Reginald Beardsley via <pulaskite= [email protected]> wrote: I'm a little confused. How is this related to implementing very narrow filters by using isolated sequential, partially overlapping filters? The Icom 705 will give single signal reception with stronger stations within 50 Hz. That blew my mind and I've been doing this stuff a very long time.
In modern professional work you use DSP, but for portable ham use, analog is much better if power is a concern.
Digital filters are my forte. I just need to know what you want done. Analog is a bit more ambiguous.
Have Fun!
Reg
I worked with high end Telemetry receivers at the factory. All tuned circuits were Lc or RC. The only crystals were for LO or system clock applications. The LC were aligned on HP Network analyzers. RC filters were Sallen-KKey, used to shape the baseband output. For testing these, I used a HP3325A and a Fluke 8921 True RMS Voltmeter that worked out to 20 Mhz. I could measure the -3dB point with ease, and quickly to test each of 16 computer selected filters
Telemetry needs very low phase delay, especially on very weak signals. We used a 70 MHz IF? which could be recorded to instrumentation recorders to allow recovery of multiple channels of frequency multiplexed data.
|
Sorry about the slow response.
I'm not yet aware that is an issue. I'm still working on the basics, e.g. measuring a Kenwood TS-520 CW filter and construct fixtures. But a tuned choke or two should do it. I need a 25 femptoFarad change to . A series arrangement should be fairly insensitive to small variations in capacitance of the varactor. The first prototype will be manual air variables.
It's entirely possible that it can't be done reliably. But the only way to find out is try doing it. Unfortunately, my 8593A is misbehaving so I seem to have had a repair job take priority. I now have 500x crystals in hand.
Have Fun!
Reg
|
OK a bit of math. I'm using conventional operational mathematics notation where * is convolution and . is multiplication. d(w-w1) is a delta operator in frequency.
The normal representation is:
S1(w) = S0(w) . F1(w) . F2(w)
If F1 and F2 have the same BW but different center frequencies we can rewrite as:
S(1) = S0(w) . (F0*d(w-w1)) . (F0*d(w-w2))
which simplifies to
S(1(w) = S0(w) . (F0(w)**2 ) * d(w-w3)
This is the best explanation of why it works that I've been able to devise. It says that the pass band of the individual filters sets the time domain response, but the overlap sets the frequency domain response. The delta in frequency makes the time domain impedance complex.
Have Fun!
Reg
|
Daniel Ricardo Perez LW1ECP
Michael A. Terrell