Electrically tunable crystal band pass filters

[rhb]

It's common practice to use a capacitor to vary the frequency of a crystal over a narrow range in a VCXO or to trim the frequency in a Clapp or Colpitts design.

The same physics applies applies to  using a crystal as a filter, but I've not seen mention  of tunable crystal IF filters.

Consider a 5 MHz IF.  The spec on ebay 10-15 cent HC-49S crystals is +/- 50 ppm Fc.  That works out to 250 Hz which is irrelevant if you are moving the frequency by 1 kHz

I've attached the section of Zverer's "Handbook of Filter Synthesis".  The last page shows the filter I want to build as a tunable center frequency, constant BW filter which I can cascade with a similar filter with the opposite shift using an emitter follower to isolate the 2 filters.

It seems to me that if the caps in parallel with the xtals are implemented with varactors one can electrically tune the filter center frequency.  The BW needs to be 1500 Hz  to get good time domain response and ideally symmetric pass bands or anti-symmetric so that the intersection pass band is symmetric.

The Icom 705 does this in DSP, but this should be a lower power drain method of implementing the feature.

Does anyone have any experience with this concept or know of prior art?

Have  Fun!
reg

[David Hess]

More commonly a variable passband crystal filter is implemented with one or two fixed crystal filters, and varying the frequency of the local oscillator to the mixers.

[rhb]

Would you cite some examples?  I'm having a hard time grasping how it works.

Edit:  How would this work for a NorCal 40 style radio?

[David Hess]

It is called variable passband tuning.

For one IF filter, if the local oscillator for the first mixer is shifted, and the local oscillator for the second mixer is also shifted, then it is as if the IF filter is moving in frequency.

If this is done with two IF filters, then the combined width can also be changed because one filter will cutoff the low frequency, and the other will cutoff the high frequencies, producing a filter response narrower than either filter alone.

[T3sl4co1l]

Would it not be better to construct a standard crystal resonator filter and vary the coupling factor (thus bandwidth) instead?  That's simply varying the shunt capacitances.

Or are you looking to make two independently tunable filters, which might be used in cascade but could also be independent therefore should be independently tunable?

Tim

[rhb]

Please forgive my being especially dense today, but I'm  still baffled.

How does varying the frequency vary the width of the pass band?  It will shift the frequency of the input signal within the pass band, but doesn't change the combined pass band of the two filters.  So white noise input wouldn't change however you shifted the LO. 

The only way I can see what you describe  working is with another IF.  With which you get more IMD products.

The Icom 705 uses DSP to allow CW filter BWs from 50 to 1500 Hz in 50Hz steps.  And I've experienced complete rejection of a stronger signal  30 Hz away during a contest.  Icom calls it "Twin PBT".  You can hear the noise level drop as you narrow the filter overlap.  It's a cascade of a pair of 1500 Hz filters BW which can be shifted +/- 750 Hz.  The SSB filters are different.

It blew my mind.  I've been doing DSP for 40 years and had never heard of the idea.  You can have narrow filters but they ring badly.  So we never used them.  We never thought of running two broad filters with variable overlap. I'm still amazed. 

My plan is to code up Zverer's calculation for the 2 xtal filter and then solve for the same crystal at multiple center frequencies.  Then use that to develop a design to split a batch of xtals into high and low and then combine them to achieve the objectives using matched pairs.  At least until I figure out the mathematics of different xtal properties.

Have Fun!
Reg

[rhb]

Would it not be better to construct a standard crystal resonator filter and vary the coupling factor (thus bandwidth) instead?  That's simply varying the shunt capacitances.

Or are you looking to make two independently tunable filters, which might be used in cascade but could also be independent therefore should be independently tunable?

Tim

You can have narrow bandwidth or short time domain response, but not both from a single filter.

The time domain response of a BP filter is dominated by the BW.  But a broad band filter presented with a pure tone is going to have the time domain response matching its BW. 

I've attached the page from the Icom manual.

Reg

[T3sl4co1l]

You can have narrow bandwidth or short time domain response, but not both from a single filter.

The time domain response of a BP filter is dominated by the BW.  But a broad band filter presented with a pure tone is going to have the time domain response matching its BW. 

I've attached the page from the Icom manual.

Reg

Implying, what, that two filters together can violate Heisenberg* conditions?  If it's looks and smells like a free lunch, I got news for 'ya...

*The uncertainty principle in QM is just the time-frequency tradeoff mentioned above.  I... gosh, I forget offhand if there's a word/name/phrase associated with the pure Fourier version? But its physics application is well known, in any case.

If you capture the signals from such a receiver, or synthesize the filter yourself through whatever means, decomposable however you like -- you will find exactly such a duration of transient response.  The easiest way to see this, I suppose, is that the time-domain response of both filters is convolved, so the FIR kernel duration is the sum of the two for starters; and if each filter is sharp enough (say, high order Chebyshev) to give a usefully narrow bandwidth when so combined, then their transient responses will reflect that, being similarly long -- some times the reciprocal bandwidth.  In fact, that scale factor will be the Q factor of the highest-Q pole pair, or something similar to that.

Tim

[mtwieg]

TSP has a good video explaining different tunable filter types, including frequency-translating filters (that part starts at 16:50):

[David Hess]

Please forgive my being especially dense today, but I'm  still baffled.

How does varying the frequency vary the width of the pass band?  It will shift the frequency of the input signal within the pass band, but doesn't change the combined pass band of the two filters.  So white noise input wouldn't change however you shifted the LO.

Each filter is preceded by its own mixer and tunable local oscillator.  The combined response of the two filters is their overlap.  So if one filter is shifted lower in frequency, and the other filter is shifted higher in frequency, the combined passband becomes narrower.

This was a high end feature of triple conversion receivers which required precision tuning of each local oscillator, which became feasible with DDS.

[rhb]

@Tim  So how did I get complete rejection of a stronger CW signal 30 Hz offset from the station I was listening to without a trace of ringing?  DSP is not magic.  I could open the BW to 100 Hz and hear both or reduce it to 50 and here only one station. No change in time domain response from the full 1500 Hz to 50 Hz

@Dave what multi-conversion radio provided a 50 Hz CW BW filter option?

Most things which can be done using DSP can be done by analog means.  This is my best guess at present as to how to implement an analog equivalent.  If no one is interested in xtal filters which can be tuned electrically over a narrow range it's probably best I wander off.

Have  Fun!
Reg

[mawyatt]

Most things which can be done using DSP can be done by analog means.  This is my best guess at present as to how to implement an analog equivalent.  If no one is interested in xtal filters which can be tuned electrically over a narrow range it's probably best I wander off.

Have  Fun!
Reg

The fundamental difference between analog and digital filters, is analog filters are constrained as IIR types and digital filters can be FIR types.

Best,

[David Hess]

@Dave what multi-conversion radio provided a 50 Hz CW BW filter option?

Was that directed to me?  The narrowest filter option I commonly see is 250 Hz.  With passband tuning, it could potentially be narrower, but probably not with slopes as good as a dedicated narrow filter.

Most things which can be done using DSP can be done by analog means.  This is my best guess at present as to how to implement an analog equivalent.  If no one is interested in xtal filters which can be tuned electrically over a narrow range it's probably best I wander off.

There are still some differences.  While a practically ideal filter can be implemented using DSP, analog filtering has the advantage of removing interfering signals before they are further amplified and potentially cause overload, so dynamic range is higher.

[mawyatt]

A analog filter (IIR) is constrained such that if one wishes a maximally flat Amplitude response, then the Phase/Group Delay is already defined (Butterworth Type), or if a maximally flat Phase/Group Delay is required, then the Amplitude response is defined (Bessel). All the analog filters reside between these two extremes.

Whereas in the FIR digital filter domain the Amplitude and Phase/Group Delay can be independent. It's also usually much better to design a FIR digital filter for a given task rather than implement a classic analog type in the digital domain because of the analog limitations.

Best,

[mawyatt]

There are still some differences.  While a practically ideal filter can be implemented using DSP, analog filtering has the advantage of removing interfering signals before they are further amplified and potentially cause overload, so dynamic range is higher.

Good point about the pre analog filtering before the ADC conversion. We often implemented a typical front end analog filter with intent of limiting out-of-band signals sufficiently to not cause issues with the ADC conversion, then the high performance channel, signal, waveform selecting and such filtering in the digital domain, sometimes even "equalizing" the front end analog filter characteristics.

Best,

[rhb]

This thread is about tuning the two crystal band pass filter design in the attachment to my first post.

Got that?

My plan is to write a program to implement the design procedure given by Zverer  Then using the same crystal parameters solve for the capacitors required for the same BW and shifted center frequencies.  Then use a varactor to implement the values.

Anyone with something to say about the design process given in Zverer I'd love to hear from you.  The rest can have fun together.  I'm not responding further to anything outside the scope stated in bold.

Reg

[mawyatt]

Zverev is indeed the classic filter book for serious analog filter work and has a large section on just Crystal Filters alone from pg 414 to 498, and a host of different design approaches and procedures to consider.

The method you've shown in Fig 8.2 seems reasonable for the narrow band approach @ 5MHz, and maybe note the "Limitations on the Method Based..." regarding the transformer & other effects pg 452 and the note pg 449 about the shunt mutual transformer inductance not being considered.

Disclaimer we've not designed this type of narrowband crystal filter, so take our comments considering such, but "feel" the biggest obstacles ahead will be accurately characterizing the crystals eqv circuit models, not sure how well the cheap HC-49S are in this regard and require very high Q (see pg 452), and the actual transformers and their env models. As I'm sure you are aware small capacitance variations make a big difference in the filter overall response, especially so for a high "Q" narrowband type, so employing "trimmer" types to "home in" on the desired performance might be wise before introducing the varactors.

A simple terminated Emitter Follower (2N3904) should allow acceptable isolation between filter stages, altho a BUF34 or unity gain HS op-amp might prove better.

Anyway, good luck with your filter design!!

Best,

[T3sl4co1l]

@Tim  So how did I get complete rejection of a stronger CW signal 30 Hz offset from the station I was listening to without a trace of ringing?  DSP is not magic.

Beats me.  30Hz notch sounds pretty ringy to me.  I'm not up on the perceptual features of notch filters, if for example a flatter (Bessel?) but higher order type (= even longer transient response!) is more innocuous.  You would certainly see much ringing on a scope with an impulse response, but you might not hear it that way because it's the lack of frequencies, not the presence thereof.  And if it's surrounded by noise or tones rather than impulses, you might not notice it at all -- it's a lack of noise or tones.

Not sure if you're just not aware -- ears are notoriously poor oscilloscopes, and not even that good of signal analyzers, due to myriad perceptual biases.  Ears are the last thing you should trust; measure always!

Tim

[rhb]

@mawyatt Thank you for some excellent comments.  My fastest op amps are TL082s at 3 MHz.  Probably should do something about that ;-)  I'll use a 2N3904 for now.

Zverev presumes identical crystals.  I'd really like to simply adjust other circuit values to meet filter requirements.

In the presumption that if the crystals are not identical I *think* I can treat it as a coupled LC filter.  Thus with the  presence of a low power MCU (e.g. MSP430) to control the varactors what I thought to do was assemble parts,  derive the motional parameters of the xtals by sweeping the filter at several varactor settings.  Then explicitly calculate the required capacitance in parallel with the filter for a range of frequencies (2 kHz max). Then program the MCU to set a static DAC to produce that voltage when you set that frequency.  Add fixed and variable caps as needed based on varactor range. 

The nanoVNA should handle all the data collection and if you're selecting the best 4 from a batch of 10 it seems attractive to me.  But use L0 optimization to determine the optimal values using a PC program.

What I'm after is a "Not a NorCal 40" with twin pass band tuning in the analog domain at under 200 mA @ 12 V in receive.  That's 30% of the 705 drain.

BTW I gave 500 5 MHz xtals coming.

[rhb]

@Tim  So how did I get complete rejection of a stronger CW signal 30 Hz offset from the station I was listening to without a trace of ringing?  DSP is not magic.

Beats me.  30Hz notch sounds pretty ringy to me.  I'm not up on the perceptual features of notch filters, if for example a flatter (Bessel?) but higher order type (= even longer transient response!) is more innocuous.  You would certainly see much ringing on a scope with an impulse response, but you might not hear it that way because it's the lack of frequencies, not the presence thereof.  And if it's surrounded by noise or tones rather than impulses, you might not notice it at all -- it's a lack of noise or tones.

Not sure if you're just not aware -- ears are notoriously poor oscilloscopes, and not even that good of signal analyzers, due to myriad perceptual biases.  Ears are the last thing you should trust; measure always!

Tim

50 Hz BP centered on signal  with adjacent signal offset 5 Hz > than edge of filter. BTW I'm watching all this on the waterfall and changing the BW.  No notch filter involved.  Read the brochure page it's an accurate description.

I'd have thought I'd said this enough.  I was introduced to DSP in 1982 when I took a job as a geophysicist with Amoco.  I'd never heard of it or seen a seismic section. I then spent 4 years working under Milo Backus, one of Norbert Wiener's GAG students at UT Austin.  I then worked for most of the rest of my career as a contract scientist/programmer for big oil companies including super majors.  I attended 25 of 28 annual meetings, presented one paper and generally hung out with the A list in reflection seismology.

What I observed during the contest the weekend after I got the Icom 705 blew my mind.  You can't make 50 Hz BW filters!!!!!!!!!  But there I was confronted by the unarguable evidence.  So I had to figure out how this was possible.  If someone with my background goes, "WOW!!!" you might want to pay attention.  I immediately understood how it worked and why it was possible.  And astounded that I'd never heard of the technique before.

The time and frequency domain characteristics of a filter  are bound together by the Fourier transform and the minimum phase constraint of physics.

Reg

[mawyatt]

BTW Zverev indicates that if the bandwidth is 1% or lower, then the crystal "Q" becomes the limiting factor and things don't follow the design equations well, think was on pg 452. Your bandwidth is more like 0.03%, which might suggest that you can't achieve the desired narrow band results because of the limiting crystal "Q".

With 500 crystals you'll have a good range to select from, hopefully some very high "Q" ones in the batch. Do you have access to a quality LCR meter to extract the crystal parameters? I think a VNA might not give as good a result since the crystal parameters are in the high impedance range and far from the normalized 50 ohms, but maybe you can get a decent estimate of such with a VNA.

Regarding the emitter follower, run the 2N3904 at a high current so the emitter impedance is very low, then create an effective 50 ohm source with a ~50 ohm series R which should look reasonable symmetrical with sink/source dynamic currents as "seen" from the load side. You want the reflected energy from the 2nd filter stage input to "see" a 50 real source impedance for both + and - signal swings and absorb that energy, otherwise the bandpass and stopband will get distorted.

Anyway, we'll offer to evaluate a few crystals (not all 500 tho!!) with our Hioki IM3536 Lab LCR meter for you if you want, but shipping might be expensive.

Best,

[mawyatt]

What I observed during the contest the weekend after I got the Icom 705 blew my mind.  You can't make 50 Hz BW filters!!!!!!!!!  But there I was confronted by the unarguable evidence.  So I had to figure out how this was possible.  If someone with my background goes, "WOW!!!" you might want to pay attention.  I immediately understood how it worked and why it was possible.  And astounded that I'd never heard of the technique before.

The time and frequency domain characteristics of a filter  are bound together by the Fourier transform and the minimum phase constraint of physics.

Reg

Have you considered a Commutating Filter? We utilized this concept back in ~1982 to extract narrowband signals, where narrowband as in 1ppm! Bandwidth and Center Frequency completely tunable over decades of frequency and spot on because it's clock derived!

This is an interesting concept that straddles analog and digital domains with discrete time continuous amplitude signal processing.

Here's a note from memory, you can make the filter bilateral (add another series R on output side of C), the center frequency tracks the clock, the switches are just standard CMOS altho one could use (we did) much faster switches. You can experiment with just using Rin and Rout without the series R, and also you can make a bandreject (notch) by swapping R for series C altho we never did the notch type, maybe a good candidate for an experiment 

This Commutating Filter concept lay dormant for decades until it was realized that cutting the filter in half, and weighted summing the stored charge across each shunt capacitor created a new type RF/MW mixer that violated conventional physics and produced biphase passive mixers that had well below theoretical noise figures, we called these "Polyphase Mixer". Here's some notes on such, be sure to follow all the IEEE papers especially Andrews and Molnar who were the principle academic researchers at Cornell (our efforts like most weren't published due to the nature of our work), when this got out it lit up the entire academic research world overnight

Caution for those interested this is a deep technical dive, and one must spend considerable time and effort to truly understand and appreciate what's going on with this apparently simple circuit following the discrete time continuous amplitude path. Please follow all the IEEE papers, and discussions. So interesting and intriguing was this concept that DARPA created a special workshop on this very subject

https://www.eevblog.com/forum/rf-microwave/polyphase-or-n-path-mixer/msg3381802/#msg3381802

Best,

[rhb]

Ooh!  You're fun!  Lots of reading to do.  Thanks!

I'm fairly familiar (i.e. read an MS thesis) with commutating mixers, but never heard of commutating filters.  Same or different?  Who developed  this?  I can see lots of opportunity for a commutating mixer being a filter in the right hands.  A commutating mixer followed by a BP filter is a tunable BP filter.  Is that what you are describing?

The overall concept I'm pursuing is a 5 MHz "Twin Pass Band Tuning" IF followed by a quadrature detector (e.g. Tayloe)  with possibly additional filtering at audio using SCAFs or DSP on low power MCUs if a pair of 2 xtal filters don't provide adequate rejection. 

Edit: I neglected to look at the jpg.  Yes, that is exactly what I have in mind for the 2nd mixer.  And what I would describe as a commutating mixer followed by a BP filter.  I want to have analog with the option of DSP

[mawyatt]

The Polyphase or as some prefer the N-Path Mixer is not a traditional commutating mixer as you've studied.

First off the PPM achieves well below the theoretical 3.92dB NF of a biphase mixer as we were taught and as Fourier Analysis predicts (can explain if interested, and there's no LNA, this is completely passively), has very high TOI, bidirectional, complex (R +-jX) impedance matches at the input antenna port with baseband components and no inductors, input match tracks clock over decades of frequency without tunable elements (no varactors and such), directly down converts to I and Q at baseband and so on!! There's probably 50 or more research papers related to the PPM and it's derivatives now.

If you read and follow the mentioned Polyphase Mixer threads and details you can forgo the Tayloe Detector and Commutating Mixer all together, and just use a PPM. You can even create the narrow bandpass (1500Hz at 5MHz, and tune to 1MHz or 10MHz with constant bandwidth) at the Antenna Input as seen from RF side so out of band signals get squashed right at the antenna port!!

The research work dates back almost couple decades now but has trickled down to some of the higher end SDRs. It's complelely integratabtle in advanced CMOS, heck we were using TSMC 65nm CMOS decades ago for early renditions. Some time ago Apple acquired a small company (Dr Caroline Andrews was a principle) that was developing these devices and believe the Apple EarPods now utilize the PPM, likely in the iPhones as well. 

Anyway, fun stuff indeed

Best,

[mawyatt]

I'm fairly familiar (i.e. read an MS thesis) with commutating mixers, but never heard of commutating filters.  Same or different?  Who developed  this?  I can see lots of opportunity for a commutating mixer being a filter in the right hands.  A commutating mixer followed by a BP filter is a tunable BP filter.  Is that what you are describing?

Recall the 1st papers on the Commutating Filter were somewhere back in 1950~60s, they used a pair of automotive distributors (for V8) and drove the rotors on a common shaft with a synchronous motor.

BTW Biphase mixing, ie. multiplying by +-1, is the most efficient means of frequency mixing as taught and Fourier Analysis shows, and the translated frequency has a 2/pi amplitude loss, thus the theoretical 3.92dB Noise Figure. Since you can't get an ideal lossless conventional passive mixer, most high performance passive mixers when driven by a high power LO (approximating a +-1 multiplication) show NF of ~6dB. Active mixers (i.e. Gilbert style) are much worse since the also modulate a DC bias which introduces additional noise and have NF greater than ~9dB. This was all true until the discovery of the PPM which completely displaces this mixer theoretical NF limit, and believe some have achieved (measured) ~1dB NF to date.

The physics behind this PPM that can trump mixer noise theory is fascinating and difficult to get ones mind wrapped around, but nevertheless does so wether one understands such or not, the PPM knows

The overall concept I'm pursuing is a 5 MHz "Twin Pass Band Tuning" IF followed by a quadrature detector (e.g. Tayloe)  with possibly additional filtering at audio using SCAFs or DSP on low power MCUs if a pair of 2 xtal filters don't provide adequate rejection.

You can do all this and much more with a single PPM, and likely even eliminate the entire RF front end as the PPM is a Direct Down Conversion Mixer with Direct I and Q outputs!!

Edit: I neglected to look at the jpg.  Yes, that is exactly what I have in mind for the 2nd mixer.  And what I would describe as a commutating mixer followed by a BP filter.  I want to have analog with the option of DSP

The PPM is a case, as is the Commutating Filter, of what we coined back in 70s as Discrete Time Continuous Amplitude Signal Processing utilizing the benefits of both analog and digital signal processing. We developed a custom handheld Real Time SA back in ~80 utilizing this signal processing based upon the Chirp Z Transform with custom CCD chips as the real time convolvers operating with discrete time (clocked) continuous amplitude (charge domain).

Anyway, fascinating subject

Best,

[gf]

I've attached the page from the Icom manual.

The two Icom filters obviously have an ultra-wide flat top and a very steep transition band. Almost brickwall. Then it is certainly possible to obtain a narrow bandwidth by choosing a small overlap.

I've attached the section of Zverer's "Handbook of Filter Synthesis".  The last page shows the filter I want to build as a tunable center frequency, constant BW filter which I can cascade with a similar filter with the opposite shift using an emitter follower to isolate the 2 filters.

I cannot imagine that a filter with only two crystals and a -3dB bandwidth of 1500Hz can have such a steep transition band that a cascade of two frequency-shifted filters can achieve a stopband bandwidth of only (say) 50Hz at -40dB (or whatever stop band attenuation you desire). IMO you need a very high filter order in order to make that possible. I guess even so high that it ends up being impractical.

[mawyatt]

I cannot imagine that a filter with only two crystals and a -3dB bandwidth of 1500Hz can have such a steep transition band that a cascade of two frequency-shifted filters can achieve a stopband bandwidth of only (say) 50Hz at -40dB (or whatever stop band attenuation you desire). IMO you need a very high filter order in order to make that possible. I guess even so high that it ends up being impractical.

This is where a Commutating Filter can likely help, achieving extremely high effective "Qs" (BW/CF), while stable with temp and time, tunable over decades with constant BW, and utilizing available inexpensive components.

Best,   

[rhb]

You must learn to walk before you run.  I just want to demonstrate that one can replicate the behavior of the DSP implementation in analog form.

The competition is the NorCal 40 IF filter.  Just a different configuration of 4 crystals with some emitter followers added,

I'll be using commutation in the final mixer.

Have Fun!
Reg

[mawyatt]

Or replace the entire RF/IF/Filter/Demod chain with a PPM 

Best

[rhb]

You can't replace the TPBT structure with PPM without much more complexity and spurious outputs.

How does your PPM implement a 50 Hz PB without either using 2 PPMs or a 50 Hz filter? The former is more complex and the latter not usable.

The TPBT is one of the most impressive accomplishments in filter theory I've ever seen and I've seen some really amazing stuff.   I 'd like to emphasize that the Icom display is just that, a conceptual display.  Not fact,    The BWs are 250, 500 and 1200 Hz for CW, 1.8, 2.4 and 3.0 kHz for SSB and  250, 500 and 2400 Hz for RTTY.  I'll skip the rest
With regard to the dual crystal filter, I think I have everything sorted except the transformer.  What core and how many turns?  On what basis does one decide?

Reg
Reg

[vk4ffab]

You must learn to walk before you run.  I just want to demonstrate that one can replicate the behavior of the DSP implementation in analog form.

The competition is the NorCal 40 IF filter.  Just a different configuration of 4 crystals with some emitter followers added,

I'll be using commutation in the final mixer.

Have Fun!
Reg

Ok, Cohn Min Loss xtal filter with varactor diodes is something I have done, the biggest issue is finding varactors with a suitable capacitance spread to cover bandwidth needs. With the Cohn Min Loss design, the caps set the bandwidth and impedance matching sets the ripple. By making the IF double conversion you could implement Twin Bandpass Tuning quite easily, with some though to overall receiver gain. Each filter will have 3 to 10db loss, depending on how well you match the crystals, their Q and the bandwidth of the filter, narrower filters have more loss so you wont be using buffers to isolate one stage from the other, but rather gain stages in the order of 10 to 20db gain.



Input and output impedance is typically about 200 ohms and the xtals need to be matched to withing 10Hz of each other. Buy lots of xtals and have fun sorting them LOL and start with a xtal with lots of decimal places in its value, like 4.19543Mhz rather than 5.00Mhz, less variance, greater Q.

[rhb]

I am interested in electrically tuning the 2 xtal circuit on the last page of my the excerpt from Zverev..  I'll use higher order filters when I get that working.

Walk before you run.

reg

[mawyatt]

You can't replace the TPBT structure with PPM without much more complexity and spurious outputs.

Sorry not sure what TPBT means?

How does your PPM implement a 50 Hz PB without either using 2 PPMs or a 50 Hz filter? The former is more complex and the latter not usable.

If you spend some serious time reading (warning it's a deep dive) the details we've offered, you should realize the Commutating Filter and PPM both implement a narrow bandpass filter at the commutating clock frequency, so no need for 2 of either.

The PPM achieve this BPF because the shunt C and "leak" effective R form a simple 1st order LPF which is mirrored around the effective clock, thus "looks" like at the RF port, a narrow BPF centered at the clock. We reviewed a IEEE paper for a PhD student at USC which won the best student paper at the ISSCC back in ~2012, this was about utilizing the Bilateral transformation property of the PPM to upconvert a multi-order LPF (added for clarity) baseband I & Q filter, thus from the RF antenna port it "looked" like a much steeper slope BPF filter centered at the clock.

Back in ~80 we implemented a Commutating Filter that could extract tones at RF & Microwave frequencies. We had to detect tones in 4KHz sequential replicating "bands" in Supergroups, for example Fo +-n*4KHz, where n is incrementing by 1 integer and Fo is the center frequency. The tones were displaced at 4KHz -150Hz (3850Hz)  and 0 +150Hz (150Hz) relative to the 4KHz repeated "bands" and recall this filter had something like 10~20Hz -3dB bandwidth at RF & Microwave frequencies!!

The TPBT is one of the most impressive accomplishments in filter theory I've ever seen and I've seen some really amazing stuff.   I 'd like to emphasize that the Icom display is just that, a conceptual display.  Not fact,    The BWs are 250, 500 and 1200 Hz for CW, 1.8, 2.4 and 3.0 kHz for SSB and  250, 500 and 2400 Hz for RTTY.  I'll skip the rest
With regard to the dual crystal filter, I think I have everything sorted except the transformer.  What core and how many turns?  On what basis does one decide?

Reg
Reg

Please enlighten us, don't know what this TPBT actually is??

BTW long ago recall there was some concern amongst researchers about the harmonic response of the PPM. It was discovered the harmonic response follows a N +-1 where N is the number of clock phases, so with an 8 phase clock one would experience a response at 7 and 9 times the center frequency (think we got this correct, you can check).

Anyway, interesting topics.

Best,

[vk4ffab]

You can't replace the TPBT structure with PPM without much more complexity and spurious outputs.

Sorry not sure what TPBT means?

How does your PPM implement a 50 Hz PB without either using 2 PPMs or a 50 Hz filter? The former is more complex and the latter not usable.

If you spend some serious time reading (warning it's a deep dive) the details we've offered, you should realize the Commutating Filter and PPM both implement a narrow bandpass filter at the commutating clock frequency, so no need for 2 of either.

The PPM achieve this BPF because the shunt C and "leak" effective R form a simple 1st order LPF which is mirrored around the effective clock, thus "looks" like at the RF port, a narrow BPF centered at the clock. We reviewed a IEEE paper for a PhD student at USC which won the best student paper at the ISSCC back in ~2012, this was about utilizing the Bilateral transformation property of the PPM to upconvert a multi-order baseband I & Q filter, thus from the RF antenna port it "looked" like a much steeper slope BPF filter centered at the clock.

Back in ~80 we implemented a Commutating Filter that could extract tones at RF & Microwave frequencies. We had to detect tones in 4KHz sequential replicating "bands" in Supergroups, for example Fo +-n*4KHz, where n is incrementing by 1 integer and Fo is the center frequency. The tones were displaced at 4KHz -150Hz (3850Hz)  and 0 +150Hz (150Hz) relative to the 4KHz repeated "bands" and recall this filter had something like 10~20Hz -3dB bandwidth at RF & Microwave frequencies!!

The TPBT is one of the most impressive accomplishments in filter theory I've ever seen and I've seen some really amazing stuff.   I 'd like to emphasize that the Icom display is just that, a conceptual display.  Not fact,    The BWs are 250, 500 and 1200 Hz for CW, 1.8, 2.4 and 3.0 kHz for SSB and  250, 500 and 2400 Hz for RTTY.  I'll skip the rest
With regard to the dual crystal filter, I think I have everything sorted except the transformer.  What core and how many turns?  On what basis does one decide?

Reg
Reg

Please enlighten us, don't know what this TPBT actually is??

BTW long ago recall there was some concern amongst researchers about the harmonic response of the PPM. It was discovered the harmonic response follows a N +-1 where N is the number of clock phases, so with an 8 phase clock one would experience a response at 7 and 9 times the center frequency (think we got this correct, you can check).

Anyway, interesting topics.

Best,

TPBT == Twin Pass Band Tuning. Essentially it allows one to move 2 filters in relation to each other to get overlapping of parts of the bandpass. The OP wants to do this analog.

I am interested in electrically tuning the 2 xtal circuit on the last page of my the excerpt from Zverev..  I'll use higher order filters when I get that working.

Walk before you run.

reg

2 Xtals in that configuration will not provide a usable filter, its slope factor will be poor. You need to cascade 3 of those stages before you start to get a reasonable shape, GQRP Club had a filter kit using that design but the transformer cans are unobtanium now days. Its a lot of math for not very much gain, which is why I recommended the Cohn Min Loss, its simple and does not require a lot of math or complex transformers to get a good result, its what is in the nocal 40, sans the series pairs on the in and output.

Also you wont be able to vary the bandpass width with that design using varactors, which I thought was part of your design parameters, being variable width filters with twin pass band tuning.

Cohn works well with crappy computer grade xtals which you dont know or do not have to measure the motional parameters for. it has been shown over and over again, by better builders than I, that these mass produced xtals do not produce better filters than the cohn by using any other method like Dishal et al.

Which is why I recommend cohn filters. Select 4 or more xtals within 10hz of each other, caps are all the same value, transformers use Mix-43 ferrite and and the filter in and out impedance is typically 200 ohm. No math, no stress. Cap sets the filter bandwidth, less cap == wider bandpass, transformers control the ripple. Nothing can be easier. Replacing the caps with varactors that give 30 to 400pF should give filter width variable from 200 to 3000Hz.

https://www.gqrp.com/club_ssb_filter.pdf

[mawyatt]

Thanks!!!

Best,

[rhb]

Twin Pass Band Tuning

The 705 brochure explains it very well at the layman level.

You *CANNOT*  implement a 50 Hz wide BP filter that doesn't ring for about 20-40 ms.  You *CAN* with a pair of partially overlapping,  broader filters achieve the same result..  Major education for me. 

[T3sl4co1l]

Twin Pass Band Tuning

The 705 brochure explains it very well at the layman level.

You *CANNOT*  implement a 50 Hz wide BP filter that doesn't ring for about 20-40 ms.  You *CAN* with a pair of partially overlapping,  broader filters achieve the same result..  Major education for me.

You've mentioned this several times, but haven't provided measurements illustrating the claim.  Surely you can hook up a simple tone burst or ASK signal and measure the radio's output under these conditions?

Tim

[mawyatt]

Twin Pass Band Tuning

The 705 brochure explains it very well at the layman level.

You *CANNOT*  implement a 50 Hz wide BP filter that doesn't ring for about 20-40 ms.  You *CAN* with a pair of partially overlapping,  broader filters achieve the same result..  Major education for me.

We'll also challenge that statement, as "ring" implies the classic underdamped response. Since the Commutating Filter can be implemented as we've shown earlier with an array of simple RC low pass filters which by definition are well over damped and have no "ring" response whatsoever, whether standalone or commutating into a BPF. However they do have a well defined settling time, which is 3.91*R*C for 2% of final value (they are IIR filters wether commutated or not). The PPM behaves in the same manner.

Edit: We checked the Icom IC-705 Basic Manual and found Digital Twin PBT page 4-4, even this layman can understand!! With this understanding, seems nothing more than just moving the upper and lower steep skirts of LP and HP filter sections to achieve a steep skirt bandpass response, is this supposed to be some kind of new filter magic or something? BTW this IC-705 is loaded with features and such and seems to be a classic SDR implementation by moving the ADC right up to the antenna proceeded by a preselection filter, nice radio

Best,

[rhb]

I'm waiting for 500x 5 MHz xtals to arrive from China.  I've been using 40 MHz 3rd overtone xtals for my basic experiments.

The TPBT concept is completely new one on me.  In fact, had I not observed it,  I'd have said it was impossible.

The time domain response of a filter is determined by the type and the BW and order.  Narrow BWs inherently have long time domain responses.

I'll continue further when I have some reswults to show, e.g. the variation in capacitor values for a constant BW filter being tuned over the BW or some 5 MHz xtal filters.

Reg

[vk4ffab]

Twin Pass Band Tuning

The 705 brochure explains it very well at the layman level.

 is this supposed to be some kind of new filter magic or something? BTW this IC-705 is loaded with features and such and seems to be a classic SDR implementation by moving the ADC right up to the antenna proceeded by a preselection filter, nice radio

Best,

Not really magic, but useful in certain situations HAHA. The real question is, how narrow do you actually need ones filters to be. For CW/Morse code operating, I have never used a filter narrower than 200Hz. So the 50Hz ringing issue, seems to me to be a non issue as filters that narrow for CW/Morse are irrelevant and if close in stations still remain a problem, switching sidebands and using the filters in ones head are good enough. In my home brew tranceivers the CW xtal filter in the IF is 500Hz and the audio filters about 300Hz. The only time I actually use TPBT is on SSB where moving one skirt up or down can be helpful to remove close in interference.

[rhb]

This is about electrically tuning a crystal band pass filter,  it is not about a TPBT.

Reg

[pienari]

Digi cap will do the job.

[rhb]

I finally got the maths sorted.

The attached PDF shows the variation in capacitance for the 4th order Butterworth design example in EMRFD.  I should like to point out that the variation for a 400 Hz shift in both direction is tens of femptoFarads.  This is a constant 400 Hz BW shifting the center frequencies of the filters.

To keep the crystal matching effort low, I'm going to do a 2 xtal filer first and test tuning it before taking on higher order filters.

BTW What's a "digi cap"?

Have Fun!
Reg

[RFDx]

I've attached the section of Zverer's "Handbook of Filter Synthesis".  The last page shows the filter I want to build as a tunable center frequency, constant BW filter which I can cascade with a similar filter with the opposite shift using an emitter follower to isolate the 2 filters.
It seems to me that if the caps in parallel with the xtals are implemented with varactors one can electrically tune the filter center frequency.  The BW needs to be 1500 Hz  to get good time domain response and ideally symmetric pass bands or anti-symmetric so that the intersection pass band is symmetric.

Changing the parallel capacitance of the crystals will barely move the center frequency and will ruin the passband ripple. A Lattice filter imho isn't the right filter topology for a tunable filter. A low order ladder type filter is easier to tune. The crystals have the same series resonance. Attached is a second order Chebychev ladder filter with 0.1 dB ripple and a BW=1.2kHz. The parallel capacitance of both crystals is compensated. Dercreasing  the crystal series capacitors from 100pF down to 20pF, moves the center frequency up at least 1.2kHz with no impact on filter bandwidth.

Crystal data:
Fs = 5MHz
Ls =  77.93937 mH
Cs = 13 fF
Rs = 15 Ohm
Cp = 4 pF

[RoGeorge]

Lots of reading to do [about Poly-Phase Mixer, AKA N-Path Filter]

Note the name might vary, found it once in a paper called 'Harmonic rejection mixer', 'Network transfer functions', 'Passive mixer-first receiver', 'Sampled data filters', etc.  You may want to watch these first:

N-Path Filters
ISSCC Videos
https://youtu.be/MP7m5OjXWUg

N path filters: basics & demo
icdutwentenl
https://youtu.be/L3wJ1XedpSo

The Realization of Narrow Band-Pass characteristics Using Sampled Data Filters (PDF thesis free to download, has all the math:  https://macsphere.mcmaster.ca/handle/11375/15508)


Back to the tunable filter, I think there is a tacit assumption in this:

Twin Pass Band Tuning

The 705 brochure explains it very well at the layman level.

You *CANNOT*  implement a 50 Hz wide BP filter that doesn't ring for about 20-40 ms.  You *CAN* with a pair of partially overlapping,  broader filters achieve the same result..  Major education for me.

You've mentioned this several times, but haven't provided measurements illustrating the claim.  Surely you can hook up a simple tone burst or ASK signal and measure the radio's output under these conditions?

Tim

rhb, this is about the sinx/x lobes, right?  (the "inescapable ringing", is it about a brick-shape band pass filter when the filtered signal is translated back, from frequency domain to time domain?)

If so, this can be avoided by choosing another shape for the bandpass filter (instead of the assumed brick that will ring in time domain).  For example, I've read a Gaussian shape translates also into a Gaussian shape (when transforming from t to F domains, or the other way around).  So a very narrow bandpass filter of a Gaussian shape will produce a time domain pulse with no ringing (the pulse will be also Gaussian in shape).  Or at least that's how I remember it.

This idea (of a Gaussian-shape bandpass filter in order to avoid ringing in time domain) was inspired to me after reading an old Tektronix service manual, where they were explaining that the Y amplifier with a Gaussian response will not ring (the raise time = 0.35/band is also from there, derived for a Gaussian response amplifier, not true for nowadays oscilloscopes).  I'm not a ham, never tested how well the Gaussian filter idea will perform to implement a, say 50Hz wide or so, CW filter that will not ring at all.


The idea of two crystal resonators with variable coupling between them (to vary the bandpass) seems very attractive (I think it was T3sl4co1l who brought it in this topic).  It's the same phenomena as the "line-splitting" from physics.

The main idea is that two resonators coupled together will influence each other.  Frequency and/or band can be changed by changing the frequency of only one of the resonator, or by changing the coupling factor.  No matter if the resonators are crystal, LC, orbitals in atoms, etc, if they are coupled, the resonance bandwidth can be changed by changing the coupling factor between them.

Did once a small simulation to convince myself this is true, here with 2 LC resonators and "k" as variable coupling factor of the coils, see how the bandpass widens while coupling them tighter, or how they "push" each other frequency:



My guess is, that for very narrow band signals as a single tone CW, the resulting bandpass will be flat enough, or would not matter.  A flat response might matter for wider bandpass filter, but won't matter much for a single tone CW.

So, a simple variable capacitor between two crystal resonators (or maybe a variable resistor? - not sure about the best way to couple the two crystals) might just do the job:   obtaining a variable bandpass crystal filter, with adjustable band by changing the coupling factor.

[rhb]

Let's suppose for convenience that the pass band of each filter is a 500 Hz trapezoid.  As a consequence the effective passband if they have a small overlap is a narrow trapezoid.

Naive examination would assume that the effective pass band governed the time domain response.  Narrow filters are wide in the time domain.  Doesn't matter if it's a Gaussian.  Narrow Gaussian in frequency is a wide Gaussian impulse response.  I would have told you it couldn't work if you'd asked over lunch at work.  I was wrong.

Some definitions:

d(w) - delta function in frequency

B(w) - transfer function in frequency of the individual 500 Hz filters.

* - convolution

. multiplication

D = B(w) * d(w-w1) . B(w)*d(w+w2)  // shifted pass band


When you compute D the transfer function is B(w)**2*d(w-w1+w2).  The delta term in frequency is a complex exponential in time.

I invite you to create a 50 Hz wide Gaussian filter centered on 600 Hz and transform it to the time domain.

Have Fun!
Reg

[RoGeorge]

Gaussian shape was proposed as a measure against pulse ringing in the time domain, not against pulse widening in time.  The Gaussian pulse will be wide in time, just that it won't ring.

Never mind, I've read again the OP, and it seems I have misunderstood the request.  You need a constant width band pass filter with adjustable center frequency, while I was offering with those coupled resonators an adjustable bandwidth filter with constant center frequency, sorry.

[rhb]

I'm trying to make the analog equivalent of the DSP in the Icom 705.

BTW  Gaussian filters very definitely ring in the time domain.  You're thinking of the symmetric zero phase time domain response.  But reality is minimum phase. I suspect the DSP version is zero phase, but an analog version can't be.

[T3sl4co1l]

But this is still strictly based on what your ears think "ringing" sounds like (or the lack thereof, as the case may be)?

Tim

[RoGeorge]

I suspect there might be one of those interdisciplinary naming issues, where the same word is used to denote different things.

For me, ringing in time domain means the oscillation that appears after an edge.  It takes some time before the signal settles to the correct value, as in this figure where, instead of a clean step signal, there is an edge followed by a dampened oscillation: 

Image from https://en.wikipedia.org/wiki/Ringing_(signal)

That oscillation right after the edge, I call it 'ringing', and it was not part of the original signal.  The original signal was a clean step.

Now, (and here I might be wrong, didn't have the time to simulate yet) I expect that if we pass a brick pulse through a brick-shape band pass filter, the output after the filter (in time domain) will show ringing, or at least some overshooting after each edge of the pulse.  However, if we pass the same signal through a Gaussian-shape band pass filter, the output pulse will show no ringing, only mellowed edges but no oscillations.  Am I wrong in these expectations?

[T3sl4co1l]

And, if we're talking bandpass filters (which in context of radio, means some offset between desired signal, IF passband, and BFO), then the impulse response is that of the envelope, not the carrier (which is obviously ringing quite a tremendous amount in absolute terms if it's a center frequency of say 4MHz and a bandwidth of 50Hz!).

We haven't heard any mention (read: I don't recall seeing it mentioned) what the received signal was, nor how it was determined to "not ring".  So far it sounds like it was just by ear, which I've tried to remind is a NOTORIOUSLY bad metric.  I don't recall any response on that, so I assume it's a point that's been ignored, presumably because the reader disagrees with what they presume is an insult to their perceptions, without even asking if it was intended as an insult, nor objectively gauging whether such a position is reasonable in the first place.

But I'd be glad to be corrected on those assumptions.

Tim

[rhb]

That is the classic definition of "ringing".  It's origins are a very messy bit of mathematics.

Your graphic shows the response to a Heaviside step function.  This is the classic test of oscilloscope AFEs.
ringinThe main cause of ringing is too steep a filter edge, though at a refined level the nature of the transitions matter.  Gaussian and Bessel filters behave the best, but it's a rare scope that looks good in the face of one of Leo Bodnar's pulsers.

Gaussian and Bessel filters don't ring, but the minimum phase constraint broadens the impulse response considerably c.f. Zverev. and the impulse response become symmetric with significant delay. Narrow BW conventional filters will muddy a fast CW signal a lot.


Model it in Octave or MATLAB.  Too much like a homework exercise from 305 years ago for me.

Reg

[mawyatt]

I'm trying to make the analog equivalent of the DSP in the Icom 705.

BTW  Gaussian filters very definitely ring in the time domain.  You're thinking of the symmetric zero phase time domain response.  But reality is minimum phase. I suspect the DSP version is zero phase, but an analog version can't be.

Completely wrong!! A Gaussian Filter has no time domain ring response period!! It can't exceed unity magnitude in a normalized step response and can't go negative in a normalized magnitude impulse response!!

The Impulse Response of a Gaussian Filter is in the form of e^-(t^2) which is bounded between 0 and 1 for all t.

The Step Response of a Gaussian Filter is in the form of (1+erf(t))/2 which is also bounded between 0 and 1 for all t, where erf(x) is the Error Function x and has limits of +-1 for all x.

BTW you can't implement an Ideal Analog Gaussian Filter, it requires and infinite number of elements to implement the exponential transfer function.

Best,

[rhb]

In the case of operators which are symmetric in both domains, the phase delay significantly lengthens the time domain response.

If 50  Hz wide filters are old hat, who made them?

[gf]

Gaussian and Bessel filters don't ring, but the minimum phase constraint broadens the impulse response considerably

A Gaussian filter still gives you the shortest possible rise time w/o overshoot in the step response. Any other filter that gives you the same rise time either has a wider bandwidth, or it overshoots.

The major drawback of a Gaussian filter is IMO the rather low selectivity. A compromise with little overshoot and much better selectivity are transitional filters like e.g. "Gaussian to 12dB", which start with an approx. Gaussian response in the passband and transition to Chebycheff in the stop band (see https://www.analog.com/media/en/training-seminars/design-handbooks/basic-linear-design/chapter8.pdf).

[rhb]

TANSTAFL

[rhb]

OK some order of magnitude calculations.

A 50 Hz wide Gaussian is 20 ms wide.  At 20 wpm each dit and space are 20 ms long.  If you try to send faster it all mashes together.  Symbols (.-) don't complete before the next one begins.  A computer can copy it, but not a human.

The thing which is so cool about twin pass band tuning is the time domain response is set by the width of the filters, not  the passband of the cascade.

Many  of the ham gear makers are doing this.  The first reference I can find is the Icom 7300, but I'm pretty sure Yaesu has a functional equivalent in the FTdx-101 from the video I watched.

I'd like to know who invented the architecture of overlapping isolated filters.  I am incredibly impressed by the digital version and very curious how well an analog version can do.  I've spent 40 years doing DSP.

Reg

[gf]

The thing which is so cool about twin pass band tuning is the time domain response is set by the width of the filters, not  the passband of the cascade.

I calculated/simulated this for two cascaded 10th order Butterworth bandpass filters, fc=10kHz, bw=500Hz (just an arbitrary example).
Take a look at the plots. The envelope's step response definitively does depend on the passband of the cascade.

If the two filters are tuned to the same center frequency, the -3dB cascade bandwidth is ~456Hz, the envelope rise time is approx. 2.3ms, but (as expected) there is some overshoot.

The narrowest -3dB cascade bandwidth I could achieve was approx. 100Hz, when the two filters are de-tuned approx. +-250Hz.

[ If I de-tune more than +-250Hz, then the overall -3dB bandwidth of the cascade becomes wider again, with a dip in the center. Even though it is a 10th order bandpass, it still has a limited roll-off in the stopband, which eventually determines the narrowest achievable bandwidth for the de-tuned cascade. With 10th order Butterworth bandpass filters as used in this example, a cascade passband narrower than approx. bw/5 is obviously not possible. ]

Now the important point: The envelope rise time of the de-tuned cascade is no longer 2.3ms, but now it is approx. 6.9ms. And it does not ring any more, because the frequency response of the cascade rolls off softly near the center (not a flat top anymore -- see plot). If the time domain response of the cascade would be determined only by the time domain response of the wide-band filters (as you claim), then we would not see a different time domain response now. But we do see a much slower response with the 100Hz cascade bandwidth than with 100% overlap.

For comparison, an ideal Gaussian bandpass with 100Hz BW would have ~6.65ms envelope rise time. So the resulting rise time of ~6.9ms comes indeed close to Gaussian, but it still does not beat a Gaussian. As you said, there is no free lunch. Regardless how you do it ("twin" or otherwise), you cannot outwit the trade-off between rise time (pulse width), overshoot and bandwidth. The frequency domain transfer function of the cascade is still the product of the transfer functions of the two filters, and the impulse response of the cascade is still the inverse Fourier transform of the cascade's transfer function. Even a filter realized as a combination of two de-tuned partially overlapping wide-band filters cannot outwit that.

EDIT: Attached Octave script

Code: [Select]

pkg load signal

fc = 10000
bw = 500
detune = 0
f1 = fc - detune
f2 = fc + detune

points=2000
fs=100000

t = [0:points-1]/fs;
signal = sin(2*pi*fc*t);

[b1,a1] = butter(5,[(f1-bw/2)/(fs/2) (f1+bw/2)/(fs/2)]);
[b2,a2] = butter(5,[(f2-bw/2)/(fs/2) (f2+bw/2)/(fs/2)]);

[H2,f2] = freqz(b2,a2,fs,fs);
[H1,f1] = freqz(b1,a1,fs,fs);

figure 1
plot(f1,20*log10(abs(H1.*H2)))
xlabel("Hz")
ylabel("dB")
grid on

figure 2
plot(1000*t,filter(b1,a1,filter(b2,a2,signal)))
xlabel("ms")
grid on


[rhb]

Edit: Complete rewrite

I'd  always said exactly the same thing until I used the TPBT feature on an Icom 705 to completely reject a stronger signal 30 Hz away on one side and another 50 Hz on the other side on the waterfall display from the station I was copying! 

I spent my career doing this.  I was introduced to DSP in 1982 and took two semesters of "Integral Transforms" taught by Bill Guy at Austin using Churchill's "Operational Mathematics".  You can't do that!   It won't work.  But I was listening to the result.

You tell me how they make it work.    Here's my best explanation:

D(w) = B(w)*d(w-w0) . B(w)*d(w-w1) =? B(w)**2*d(w-w0)*d(w-w1) = B(w)**2*d(w-w3)

where d(w) is a delta function in the frequency domain that performs the shift and all the normal Fourier theorems apply.


I thought I'd see if tunable crystal filters would work with an emitter followers  to provide isolation to recreate what they are doing in DSP for use in a portable QRP rig as a means of reducing the current drain.

Here's the work needed to properly simulate this.  It's a huge number of figures to generate :-(  Probably need Smith charts to do it right.


The filters:

a 500 Hz BW Nth order filter with an Fc of 5 MHz

a 50 Hz BW Nth order filter with an Fc of 5 MHz

a pair of N/2 order 500 Hz BW filters with F1L=4.999525 F1H=5.000025 F2L=4.99975 F2H=5.000475

Work needed:

design filters 

plot the filter operator in frequency and time showing phase and complex impedances

generate a random set of sine waves which span the BW of the filters

plot the responses in time and frequency

modulate the sine waves with a 2 Hz square wave (CW keying "5" at 25 wpm)

plot the responses

[rhb]

Go to a ham radio store during an HF CW contest in your region and ask to be shown the Icom 7300 or 707 TPBT feature in operation.

Then tell me how it works.  Because I've always been told it can't be done.

Reg

[T3sl4co1l]

Keep in mind they could be doing further audio processing, emphasis, expanding, noise gating, etc. I kind of doubt that's actually the case, but it's a possible explanation for intelligible (to a human) results when the code is otherwise digitally resolvable.  That is, it's not pulling information out of nowhere (bandwidth), it's expanding apparent bandwidth by nonlinear methods.

Tim

[rhb]

Let's consider 3 signals at 5.000020, 5.000050 and 5.000100 MHz.

Let's have 2 filters.  One has a pass band of 500 Hz from 4.999575 to 5.000075.   Filter the signal composed of the 3 sine waves with the filter.  That leaves 2 signals at 5.000020 and 5.000050.  The signal at 5.000100 has been removed from the signal.

I now pass that signal through a 2nd filter with a pass band from 5.000025 to 5.000575.  That leaves just the signal at 5.000050.

How does the 2nd filter know that the signal has been through the first filter and the upper signal removed?  How is this different from using only the 2nd filter and turning off the signal source?

Serious question.  I want to know the answer.  This confounds all I've believed for 40+ years.  It raises serious question about the statements about commutation and association Norbert Weiner set forth in the 40's.

I'm a seismic guy.  We are the lineage that originated DSP.  We could use a 125 Hz Nyquist system..  An individual modern seismic survey is a $10-30 million operation to acquire and process.

The correct equation is D(w) =[S(w).B(w)*d(w-w0)] . [B(w)*d(w-w1)]

where S is the input signal.

I am working on implementing this using crystal filters, but have 500 crystals to measure.

I'd love to have someone else do the computer simulation stuff and let me sniff solder fumes and test the results.  I did the computer stuff for a very long time.  My blood lead levels are too low ;-)

Reg

[David Hess]

Go to a ham radio store during an HF CW contest in your region and ask to be shown the Icom 7300 or 707 TPBT feature in operation.

Then tell me how it works.  Because I've always been told it can't be done.

Told that what cannot be done?

Twin passband tuning sure works.  I described it in an earlier post and have used it many times.

The implementation requires fine tuning of the local oscillator frequencies so it was a pretty late development.  Most radios just get by with IF shift tuning which does the same thing and works the same way but is limited to a fixed IF filter width.

[rhb]

It's the mathematics that is the problem.

What's the correct equation to describe the situation?

Edit:

This has really become an issue with me.  I agree with gf and David Hess. 

All my life I was taught that if you had an underdetermined system of equations there were an infinite number of solutions.  In 20004 David Donoho of Stanford proved that was rarely the case if the solution of Ax=y was sparse.  That is most of the entries in x were zero.  Thus you could solve NP-Hard problems in L1 time *most* of the time.  And if you couldn't, well you couldn't.

[rhb]

OK

@gf and I generally agree about the mathematics.   Though there are a lot of nuances to consider.

@David Hess and I agree about the filter performance.

Which leaves the question of the correct mathematical description.  Without that, there is no way to design an analog filter to replace the digital filter.  Analog filters must be causal and that implies minimum phase.

I've put forth an equation in standard transform notation.

So how do we reconcile this apparent conflict?

Reg

[gf]

Keep in mind they could be doing further audio processing, emphasis, expanding, noise gating, etc. I kind of doubt that's actually the case, but it's a possible explanation for intelligible (to a human) results when the code is otherwise digitally resolvable. That is, it's not pulling information out of nowhere (bandwidth), it's expanding apparent bandwidth by nonlinear methods.

I think of a simple comparator, which recovers rectanglular pulses from the pulse-shaped ones ¹⁾

Pulse shaping on the TX side also plays a role. If the signal is already properly band-limited, you can even send it through a brickwall filter (a little bit wider than the signal's occupied bandwidth), and the brickwall filter won't add additional ringing.

EDIT:

¹⁾ For instance, if root raised cosine pulses with alpha=0.5 are turned into rectangular pulses with a comparator at 50% threshold, the resulting edge jitter is still only (roughly) 10% of the symbol width (see timing jitter of the 50% crossings in the attached eye diagram). So I guess the "crisp" rectangular dits recovered by the comparator should be well recognizable (if this speed can be recognized by a human at all, and granted that the signal is not too noisy). At 50 baud (20ms symbol duration) and with alpha=0.5, the occupied bandwith of these RRC-shaped pulses is only +-37.5Hz from the center. Maybe even a smaller alpha (-> smaller BW) is still feasible (I just calculated for 0.5).

EDIT: Sorry, was too good to be true. Had a typo in the calculation. I'll re-calculate and update the message when I find some free time.
False Alarm.

[mawyatt]

..... We are the lineage that originated DSP. .....

That is questionable since DSP concepts were developed for telephone line digital communications back in 40s and 50s at Bell Labs, and later became known as MODEMs that were part of the US air defense system called SAGE in 50s, and AT&T offered commercial versions of these in the 50s. Pagers were also in use in the 50s.

Had the honor of working along side Dr William Acker, a brilliant DSP guru that was behind the world leading high thru-put Group Data Modems at Honeywell in the 60s & 70s with his key patents in DSP based channel equalizers, later became Paradyne and eventually acquired by AT&T.

So Digital Signal Processing has been around for a long long time, likely preceding the transistor, although not generally known in the early days specifically as DSP, just another useful means of signal processing.

Best,

[rhb]

..... We are the lineage that originated DSP. .....

That is questionable since DSP concepts were developed for telephone line digital communications back in 40s and 50s at Bell Labs, and later became known as MODEMs that were part of the US air defense system called SAGE in 50s, and AT&T offered commercial versions of these in the 50s. Pagers were also in use in the 50s.

Had the honor of working along side Dr William Acker, a brilliant DSP guru that was behind the world leading high thru-put Group Data Modems at Honeywell in the 60s & 70s with his key patents in DSP based channel equalizers, later became Paradyne and eventually acquired by AT&T.

So Digital Signal Processing has been around for a long long time, likely preceding the transistor, although not generally known in the early days specifically as DSP, just another useful means of signal processing.

Best,

Norbert Weiner developed the foundations of DSP on contract to the Navy in 1940.  It was known as "the yellow peril" during the war because of the classified color and the difficulty of the mathematics. It was published in 1949 as "The Smoothing, Interpolation and Extrapolation of Stationary Time Series"

The oil industry immediately jumped on it and funded the Geophysical Analysis Group with 8 students.  Enders Robinson and Sven Treitel wrote a series of papers on DSP which appeared in Geophysics that were reprinted  as the Robinison-Treitel  Reader and widely distributed by seismic service companies.

Here's a more detailed story from the man who did the first deconvolution by hand on paper after digitizing the traces by hand in the summer of 1952.
 
https://library.seg.org/doi/10.1190/1.2000287


My PhD supervisor at Austin was a member of GAG, Milo Backus.  So yes, reflection seismologists originated DSP.  We were the only users that could work with a 125 Hz Nyquist.

[mawyatt]

Interesting read, was not aware of GAG involvement!!

Altho should have been since a former colleague (Prof at Cornell) father, Dr Peter Molnar at Colorado, won the Crafoord Prize in Geoscience in 2014

Thanks for the info

Edit: BTW the colleague at Cornell, Dr Al Molnar, he's the academic emphasis behind the Polyphase Mixer, Mixer-First, N-Path Mixer, whatever you want to call it. If you check the IEEE papers mentioned, you'll find he's one of the original authors.

Best

[rhb]

I have by any definition a *large* DSP library.  One of the the things I find *really* annoying is the general failure by the EE community to acknowledge what they owe the oil industry.

TI didn't start by building modern high speed wide word ADCs.  They started by building 250 Sa/s 16 bit ADCs which they sold for princely sums when they started building them in the 60's.

It's an interesting counter example to industry exploiting defense research.  Though Weiner's assignment was to figure out where to point an antiaircraft gun.  So it's defense exploiting industry exploiting defense.  And there's no another layer of industry exploiting defense.

FWIW I created an example of a pair of overlapping filters today using a 1 Hz resolution frequency domain with a 10 MHz Nyquist array and trapezoidal filters.

Have Fun!
Reg

Edit:  The norm today for an offshore survey is 5-6 sets of parallel lines with stations on 6.25 m or less spacing x 100-200 m (to keep the dozen 10-20 km long cables they are towing from tangling).  It takes several months to acquire the 10-12 TB of data and typically 6-9 months to process it.  Total bill for the project will run around $30-50 million.  These are done when it's time to decide where to place the wells.

The imaging step, the digital equivalent of holography, compute is  7-10 days on 10-20 thousand cores.

[rhb]

Keep in mind they could be doing further audio processing, emphasis, expanding, noise gating, etc. I kind of doubt that's actually the case, but it's a possible explanation for intelligible (to a human) results when the code is otherwise digitally resolvable. That is, it's not pulling information out of nowhere (bandwidth), it's expanding apparent bandwidth by nonlinear methods.

I think of a simple comparator, which recovers rectanglular pulses from the pulse-shaped ones ¹⁾

Pulse shaping on the TX side also plays a role. If the signal is already properly band-limited, you can even send it through a brickwall filter (a little bit wider than the signal's occupied bandwidth), and the brickwall filter won't add additional ringing.

EDIT:

¹⁾ For instance, if root raised cosine pulses with alpha=0.5 are turned into rectangular pulses with a comparator at 50% threshold, the resulting edge jitter is still only (roughly) 10% of the symbol width (see timing jitter of the 50% crossings in the attached eye diagram). So I guess the "crisp" rectangular dits recovered by the comparator should be well recognizable (if this speed can be recognized by a human at all, and granted that the signal is not too noisy). At 50 baud (20ms symbol duration) and with alpha=0.5, the occupied bandwith of these RRC-shaped pulses is only +-37.5Hz from the center. Maybe even a smaller alpha (-> smaller BW) is still feasible (I just calculated for 0.5).

The IC-705 has the following IF BWs to choose from:

AM: 9kHz, 6 kHz & 3 kHz
SSB: 3kHz, 2.3kHz & 1.8kHz
RTTY: 2.4 kHz, 500 kHz & 250 kHz
FM: 15 kHz, 10 kHz & 7 kHz
DV:  15 kHz, 10 kHz & 7 kHz
CW:  1200 Hz, 500 Hz & 250 Hz

The Fc of both filters are individually adjustable in 50 Hz steps.  Clearly AM & SSB *must* be linear. 

FWIW The BW of a CW signal is 0.35/rise_time.  It's actually not the keying rate.  But the keying rate does set the maximum usable rise time.

[gf]

Clearly AM & SSB *must* be linear. 

Potential non-linear processing was rather meant in the baseband, after demodulation. For instance I see now way how linear processing could re-sharpen the pulse edges once they have been blurred and high frequencies have been completely eliminated. However, non-linear processing can do that, re-introducing high frequencies.

I'd  always said exactly the same thing until I used the TPBT feature on an Icom 705 to completely reject a stronger signal 30 Hz away on one side and another 50 Hz on the other side on the waterfall display from the station I was copying!

At which data rate? Still 50 baud (= 20ms symbol duration = 60 wps), or slower?

[T3sl4co1l]

Clearly AM & SSB *must* be linear. 

Potential non-linear processing was rather meant in the baseband, after demodulation. For instance I see now way how linear processing could re-sharpen the pulse edges once they have been blurred and high frequencies have been completely eliminated. However, non-linear processing can do that, re-introducing high frequencies.

Quite.  And, without measurements, we're merely left to speculate what the actual response is.

As you're [rhb] well aware, DSP can be changed at the flick of a bit; a filter doesn't need to be consistent across modes (or time or space for that matter; we're not restricted to LTI systems here!).  It might be easier that way, but it could also be that they went to the trouble of computing coefficients live, instead of literally shifting center frequencies around.  Analytic filter design might've been hard back in the day (for certain values of "day"; it's really not that new), but it's entirely understood now, and eminently computable.  It could even be that they detect the kind of program matter (not actually very hard in context of full modern computing, but again -- even less likely in an embedded context I would guess), and apply other kinds of filtering (including nonlinear effects as above) to improve legibility.

Tim

[mawyatt]

Recall a non-linear device used in very early telephone and radio called a "coher" or something similar. Evidently had almost magical properties based upon lightly compressed carbon particles as in a current/voltage dependent resistor, altho can't remember details (OK I'm old, but not that old!!).

Best,

[tggzzz]

Recall a non-linear device used in very early telephone and radio called a "coher" or something similar.

https://en.wikipedia.org/wiki/Coherer

I'm old enough to remember (when I was a kid) reading about them in very old books

[rhb]

Historically, the only way to implement TPBT was using 2 IFs with fixed filters and adjustable LOs.  I don't know of any instances of that, but would love to know what radios implemented TPBT before the Icom 7300.

I've got about 2/3 of the math coded up in Octave.  I've got signals at 5 MHz, 30 Hz below and 50 Hz above, two filters 500 Hz wide with 50 Hz of overlap and a 50 Hz wide filter.  Filters are trapezoids with 10 Hz wide skirts on each side. KISS. 

Everything is being specified in the frequency domain for convenience.  I'm not using any of the Octave packages.  Just doing the basic math by hand.  This for the simple reason I know the operations a *lot* better than I know Octave/MATLAB.  I can count 8 different DSP software packages I used over the years.  I spent far more time fixing bugs in DSP software or writing new codes than actually using it.

I am unlikely to make things minimum phase as they should be or make the huge number of figures needed to explain it all.

My goal is an analog version of TPBT using a single fixed IF.

Reg

Edit:  This is every bit as tedious as I expected.  Octave is crashing while using a 56 GB memory allocation when I'm only part way through the calculations.  So I have to go back and reorganize to create, plot, use and delete the series as I go through that nightmare task list I laid out earlier.   I'm not happy with 2 Hz for only 1 second, so I want to make the series longer.  I'd really like 10 seconds, but that will be an overnight "tape to tape" job on this machine as it's only 16 GB of DRAM and 64 GB of swap.

[RFDx]

Historically, the only way to implement TPBT was using 2 IFs with fixed filters and adjustable LOs.  I don't know of any instances of that, but would love to know what radios implemented TPBT before the Icom 7300.

An example would be the commercial R&S EK56 receiver from the end of the 60s. It uses Twin-PBT to implement 20 switchable IF-bandwidths.

[David Hess]

I found a pile of examples going back to the 60s.  I think some of the confusion, besides various manufacturers calling it different things, is that in some implementations the lower and upper filter cutoff was adjustable, and in others the filter width and filter offset were adjustable, but they were both implemented in the same way with adjustable local oscillators.

[rhb]

Major examples?  These are obviously radios outside the amateur realm at the time.

I've been working on measuring crystals and it looks challenging with 500 to measure.  In particular lab temperature control over a 2 day period. Or lack thereof :-(

So I'm considering binning the crystals by frequency using a counter and test oscillator.  Then put them in the refrigerator overnight and bin based on cold temperature frequency.

That will reduce the number I need to measure with the 8753B/85046A   in one session to find matching crystals.  So, hopefully,  the results should be more consistent with each other.

I'm beginning to appreciate that this is not going to be a quick project if done properly.  But for 8 cents each I *have* to try ;-)  Two of the 4th order filters in the EMRFD example with varactor tuning is all I need.  So 2 sets of 4 well matched crystals out of 500.

Have Fun!
Reg

[StuartA]

I hope that I have not got my wires seriously crossed here, but the Hammarlund tube receivers all had variable IF (455kHz) bandwidth using circuitry developed in 1938. I guess it could be developed for other IF frequencies. See https://www.vintage-radio.net/forum/showthread.php?t=196991. People who used those sets spoke very highly of the IF bandwidth control.

[rhb]

Thank you.  iI never knew the details.    Dad referred to that circuit, though without a schematic, 55 years ago. That's an expensive crystal. 

I've seen variants, but not with variable Q, though  I knew  how it was accomplished. I've still got  copies of QST from when Dad was in high school in 1935.

The narrower BWs of the Hammarlund circuit would inevitably tend to ring.  However, I've not gotten less than ~150 Hz -3 dB BW from a single crystal.  So it might not be a problem.  I've been fooling around with a single crystal, but not as sophisticated as the link.

FWIW I've been discussing various aspects of this on several different lists which is becoming confusing trying to keep track of what I posted where.  For my own sanity I'm going to start cross posting  to qex@groups.io.

There is a surprisingly long thread on the HPAK list on measuring motional parameters with an 8753B which is NOT where I would have expected to get so detailed.

This is an attempt to duplicate the performance of  DSP TPBT  at a single IF using crystal filters to reduce the power consumption below DSP levels.  It seems to me highly unlikely to take less than a year long project given delays for instrument repairs, etc.

I have 400x xtals from one seller and 100x from another seller.  I've been diddling around with them while I worked out a simple way to measure them quickly.

I've designed a fixture to allow using a 16047A fixture from the 4284A LCR meter on either an SA w/ TG or a VNA to measure crystals and am now waiting on some SMA-N/BNC-F bulkhead jumpers to arrive from China.

Have Fun!
Reg

The screenshot is for the 4th order Butterworth example in EMRFD.

[stellavox]

There was a construction article in QST magazine a year or two back - transceiver with a varactor tuned (as I remember) crystal filter - to vary the bandwidth

[rhb]

Thanks,  I'll look for that.  It's longer ago than you think if the ARRL really did remove construction articles from QST as was stated a couple of years ago when I requested QEX as my league publication.  I've done lots of experiments tuning a single crystal filter using both series and parallel capacitance.

Single crystal filters were very popular in Dad's day.  He told me about single crystals filters when I was a teenager.  But the real clincher was  W7ZOI's experiments in EMRFD.

My 4395A arrived yesterday.  I'm still waiting on parts for the adapter to mount the 16047A LCR fixture to the 4395A.  This is proving a more complex project logistically than I expected.  The 4395a is a VNA,SA and impedance meter all in one with 1 Hz resolution from 10 Hz to 500 MHz.  It is *sooooo* cool!

Have Fun!
Reg

[busaboy]

The biggest problem today is getting the crystals you need. Vendors of custom crystals for hobbyist all went out of business a long time ago. It's far easier to buy monolithic crystal filters of different bandwidths. Finding some as a hobbyist however is not easy.

[rhb]

I'm designing around what's cheap.  $0.08 USD is pretty cheap for 5 MHz crystals.  The real cost is the labor to measure them (and the test kit.  though there are lots of options besides what I'm using.

At the moment I'm waiting for parts for the fixture to mount my 16047A fixture on the 4395A and test some crystals.

Have Fun!
Reg