New SSTV Demodulator.

Forum for discussion of SSTV topics. Slow Scan television (SSTV) is a picture transmission method used mainly by amateur radio operators, to transmit and receive static pictures via radio in monochrome or colour.

Moderators: Dave Moll, Andrew Davie, Steve Anderson

Re: New SSTV Demodulator.

Postby Steve Anderson » Tue Nov 29, 2022 11:38 am

acl wrote:...have you tried this yet with the original baseband SSTV to 625/VGA ? Regards Chris Lewis


No, there are a few incompatibilities between this demodulator and the MkI up-converter. The only way I could see this being used with the MkI is to use the analogue 'monitoring' output to feed the A-D input of the MkI. This would require adjustment of DC and signal levels - though no big deal. But you lose the advantage of the digital FIR filter with its vastly reduced overshoot on rapid (SSTV?) transitions. Even so, it probably would better than the modified Robot 70 version. It's worth investigating...

If you only require a SSTV-baseband (analogue) video the micro could be replaced with something like a 14-pin PIC16F684. That's also worth looking into, thanks for the idea...

However, the objective is to replace all the complex parallel data/address lines of the MkI with a serial arrangement. As things stand the large 32-pin RAM in the MkI would be replaced with one, maybe two, 8-pin serial RAM chips. The output D-A would also hopefully become a single 8-pin device.

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby Steve Anderson » Tue Nov 29, 2022 3:27 pm

As a result of Chris's suggestion here's an interim analogue-only version of the new SSTV demodulator. It may need gain and DC-shift adjustments downstream, conceptually the same as 'Black Level' and 'Contrast' controls. It does not have the advantage of the FIR filter mentioned previously and uses the 'zero-crossing' method of demodulation. But it should be an improvement on the Robot 70-type of demodulators...maybe, I was not impressed with my first version of a zero-crossing arrangement, but I'll give it another go.

Drawing removed as now obsolete...

As an afterthought using an 18-pin version (PIC16F628/648) could be a replacement for the PIC18F14K50 and provide both analogue and serial data output. The reason I chose the '18F14K50 is I have a dozen lying around - I can't recall what I ordered them for! Secondly, they have a bigger/more flexible instruction set, thirdly, they're 2.4 times faster allowing the FIR filtering to be implemented. If using a PIC that can use a 64MHz clock, that's 3.2x faster. There is a fourth plus-point, but memory fails me!

Note: I haven't built or written any software yet for any of these recent versions - I suggest you don't either until I get some results...

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby Steve Anderson » Wed Nov 30, 2022 12:39 pm

As things have turned out, I'll be sticking with the PIC18 version. Aside from all the hardware and software advantages it has over any PIC16 device it's also at least 33% cheaper than all the PIC16 versions mentioned previously. The difference between 14 and 20 pins is no big deal, surely?

The current drawing is that posted in this thread on the 28/11/22, V6.43. Stay tuned...

The fourth plus-point I mentioned in the previous posting is there's no 'bank switching' instructions required in PIC18 devices which uses up program memory and means an additional execution cycle (i.e. time wasted with no benefit).

I'm also considering this to be a NBTV (12.5Hz frame rate) to 625/50 or a version of VGA....but one step at a time...625 out will be first...

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby Steve Anderson » Sat Dec 03, 2022 1:28 pm

I've been thinking on the memory requirements for an up-convertor that follows on from this demodulator. The MkI version used 128 lines x 256 samples for the active video duration of 55ms (EU standard).

Using a zero-crossing method of demodulation results in a variable quantity of samples during the active video line-time of 55ms. 1500Hz x .0.055s = 82.5 samples, 2300Hz x .055s = 126.5 samples if using full-cycle sampling. Those figures double to 165 and 253 respectively if half-cycle demodulation is used.

This is somewhat worse when you come to the sync pulses, 1200Hz x .005s = 6 or 12 samples for full and half-cycles respectively. As the sub-carrier is not locked to the line rate there will be occasions when perhaps only 4 or 8 (or 5,10 and 7,9,11) samples might be resolved. This may account for the 'sync-jitter' I noticed on my first zero-crossing demodulator attempt.

I'm not sure if this issue can be resolved, suggestions and ideas welcome...all the above assumes timing and frequencies are exact, some tolerance needs to be considered too...

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby Steve Anderson » Sat Dec 03, 2022 2:18 pm

Following on from the above - I've thought about this in the past, quite a lot really, so here goes...

What if we mixed the incoming 1200-2300Hz signal (+/- sidebands) with say 20kHz? That would give us a double-sideband resultant I.F.. The LSB being 17.7-18.8kHz, the USB being 21.2kHz-22.3kHz? (Again +/- sidebands). Filter off the LSB then use the zero-crossing technique at this higher IF frequency to improve the definition, though not the bandwidth, of the demodulation? I'm straying into an area here where my knowledge and experience is limited. Is the concept valid? I realise the filtering may be tricky/difficult/impossible, but is the concept sound?...or have I made a complete idiot of myself?

This, if valid, would be a more complex arrangement, but if it gets around the shortcomings above it may be worth it...

Steve A.

My major error maybe my confusion between AM and FM mixing and the resultants.

A good contender for a double-balanced mixer would appear to be the now obsolete MC1496, though I can still get them here at US$0.50 each. Usually used at much higher frequencies than 20kHz, but I don't see why it shouldn't work. As I mentioned, filtering may be the biggest issue.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby Steve Anderson » Sun Dec 04, 2022 12:06 pm

Current iteration of the SSTV zero-crossing demodulator, without the mixing system mentioned in recent posts.

I have yet to build, write the software and test it. Note the change of logic supply voltage to 3.3V.

Steve A.

SSTV V6 Demodulator 8.gif
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby Steve Anderson » Sun Dec 04, 2022 7:34 pm

The more I dwell on the zero-crossing method of SSTV demodulation the more I'm getting dubious about its supposed plus-points. I'm not saying it's totally flawed, but I just can't see any advantages. So for the time being I'm putting it to bed unless I have a 'Eureka Moment' - or someone else does.. Therefore I'm going to carry on with the last S-curve [1] demodulator posted here in this thread on Nov. 20. It provided quite acceptable results in a previous version, the one referred to in that posting should be marginally better.

It will still be followed by a FIR filter within the micro which 'should' be better than the analogue Butterworth filter previously used, it was the only thing I felt deserved the criticism I heaped upon it! It's simply in the nature of Butterworth filters, perhaps my scorn was a little unjustified...

Steve A.

[1] I'm now going to refer to this as an 'S-curve' demodulator as it bears so little resemblance to its origin, the Robot 70.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby kd2bd » Mon Dec 05, 2022 3:53 am

I've had these conversations with myself, too, Steve. 8)

Take a look at any of the SSB Phasing Exciters designs available on-line for Amateur Radio use for ideas on how to upconvert an SCFM SSTV signal to an arbitrary RF frequency without having to resort to using mechanical or crystal filters for sideband selection. (https://www.arrl.org/files/file/Technology/tis/info/pdf/9304027.pdf) The most challenging aspect of this approach will be the development of a polyphase network capable of splitting the SSTV signal into two components (I and Q) that have equal amplitudes but a 90 degree phase difference. Here is a description of a popular approach that uses op-amps configured as ALL-PASS filters: https://www.hobbielektronika.hu/forum/getfile.php?id=367683. Any design acceptable for SSB voice will be more than adequate for an SSTV signal.

In lieu of an LM1496, a very effective doubly balanced mixer/modulator can be designed around a CD4053, CD4066 (or 74HCT equivalent) using a square-wave local oscillator and some op-amps. (Or, if you really want to stay with a sinusoidal LO, there's always the SA612 chip which is still in active production (although no longer available as a DIP) and diode ring modulators.)

At any rate, once you've upconverted the SSTV signal to RF, you can employ an inexpensive ceramic filter to limit the detection bandwidth to improve the SNR, and have a wide choice of FM demodulators at your disposal (including quadrature detectors commonly used today for narrow-band FM detection). Post-detection filtering gets real easy, too, because your cutoff frequency (being RF) is well above the baseband DC to 1 kHz or so that you're interested in processing.

73 de John, KD2BD
User avatar
kd2bd
Mad Scientist
 
Posts: 65
Joined: Wed Sep 28, 2016 5:47 am
Location: New Jersey

Re: New SSTV Demodulator.

Postby Steve Anderson » Mon Dec 05, 2022 1:15 pm

kd2bd wrote:...a very effective doubly balanced mixer/modulator can be designed around a CD4053, CD4066 (or 74HCT equivalent) using a square-wave local oscillator and some op-amps. 73 de John, KD2BD


Hi John, thanks for all the data and links you provided, though if you have any data or links to items related to the quote above I'd be interested in perusing them.

I'll be needing some time to let what you pointed to in your posting above sink in - prepare for some silly questions coming your way!

I posted the following link some years ago, you may have missed it, I'd be interested in your thoughts. The authors hardware is using 'a sledgehammer to crack a nut', and these days could be implemented with a much simpler hardware solution I think.

http://www.agurk.dk/bjarke/projects/dsp ... 20sstv.htm

The formula derived at the end shouldn't be that hard to turn into code for a micro...perhaps with the exception of the division, but it's not impossible...

Demodn={Qn*In-1-In*Qn-1}/{In2+Qn2}

The author mentions "It performed quite well...", how well one wonders? Unfortunately he didn't provide any 'scope screen-shots or results obtained. I did try, without success, to contact him some years ago.

Also he limited the demod bandwidth to 800Hz rather than say 1kHz, but that's just a case of altering a few factors.

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby kd2bd » Tue Dec 06, 2022 3:16 am

Hi Steve.

Here's the general idea of creating a balanced modulator using the approach I mentioned earlier:

balanced_modulator.png

This drawing came from the website of Max Carter, who published an abbreviated version of this article in Communications Quarterly in the 1990s. His article completely changed my way of thinking in terms of digging weak signals out of the noise, and I went on to use some of the techniques he described in some of my subsequent projects.

It would take a lot of circuitry to up-convert an SSTV video signal to RF and perform the demodulation there. But, outside of a few precision resistors and capacitors, the circuity is cheap.

Two of these balanced modulators would be required: One fed with SSTV at a 0 degree phase, and the second fed with SSTV at 90 degrees. The local oscillator ("Input B") would be a square wave at the desired suppressed carrier frequency. The phase of the local oscillator feeding each balanced modulator would need to be offset by 90 degrees as well, and there are simple techniques employing flip-flops for doing this.

After the two balanced modulator outputs are combined (using equal value resistors), a high quality SSB signal is produced, which is an up-converted replica of the original SSTV signal.


73 de John, KD2BD
User avatar
kd2bd
Mad Scientist
 
Posts: 65
Joined: Wed Sep 28, 2016 5:47 am
Location: New Jersey

Re: New SSTV Demodulator.

Postby Klaas Robers » Tue Dec 06, 2022 4:49 am

SSTV is an FM signal. To my feeling there is only information in the zero crossings of the signal. That is in the positive crossings and the negative crossings. That is why I limit the input signal to a square wave and then use both edges.

Do we think that we can generate more zero crossings by up-mixing the analogue FM signal? And that the position of the newly generated zero crossings are relevant? That is a new point of view for me.
User avatar
Klaas Robers
"Gomez!", "Oh Morticia."
 
Posts: 1656
Joined: Wed Jan 24, 2007 8:42 pm
Location: Valkenswaard, the Netherlands

Re: New SSTV Demodulator.

Postby kd2bd » Tue Dec 06, 2022 9:59 am

Hi Klaas.

Frequency up-conversion would make post-detection filtering a lot easier, and if you wanted to count zero crossings, there would be many more zero crossings available per pixel, and the number of zero-crossings would be more closely matched over the entire range of modulation frequencies (as opposed to the almost 2:1 ratio that exists at the standard SCFM audio frequencies).

73 de John, KD2BD
User avatar
kd2bd
Mad Scientist
 
Posts: 65
Joined: Wed Sep 28, 2016 5:47 am
Location: New Jersey

Re: New SSTV Demodulator.

Postby Steve Anderson » Tue Dec 06, 2022 2:44 pm

...it seems like I've 'opened a can of worms' here...but it all does raise some interesting thoughts...

One thought, or more correctly a question, why was something like this not applied to broadcast FM receivers? (88-108MHz). To my knowledge they all employed some form of analogue S-curve detector, ratio detector, Foster-Seely detector or quite a few other, though similar, variants. I'm bound to be proven wrong, but the vast majority used one of those methods rather than some form of counting/digital method, maybe cost/complexity?. What about amateur radio FM transceivers/receivers on 2m, 70cm and so on? What do they use in todays digital world for FM demodulation?

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: New SSTV Demodulator.

Postby Klaas Robers » Tue Dec 06, 2022 9:29 pm

I fear, Steve, this is because FM broadcast reception is from the era of electron tubes. Each extra tube was very expensive. I even remember that my company Philips "invented" a new tube with seven grids, the EQ40 and EQ80, a nonode, to limit and demodulate FM, because Philips did not like to pay to Forster-Seely for their patent on their FM-detector. Later on the Nat.Lab. developed the ratio detector using two germanium diodes OA85 for the same purpose. In fact this made the Forster-Seely circuit obsolete for radio and television receivers.

How nowadays in Software Defined Radios the FM detection is done, I don't know.... well, in software.
User avatar
Klaas Robers
"Gomez!", "Oh Morticia."
 
Posts: 1656
Joined: Wed Jan 24, 2007 8:42 pm
Location: Valkenswaard, the Netherlands

Re: New SSTV Demodulator.

Postby kd2bd » Wed Dec 07, 2022 3:47 am

In the late 70s and early 80s, single chip FM demodulators employing Quadrature Detectors starting becoming popular (LM3065, LM3075, LM3089, LM3189, etc.)

Quadrature detection could be used for SSTV provided a phase-shift network (likely built from a series of all-pass filters) could be produced to provide a 90 degree phase shift at mid-band (1750 Hz), a 0 (or 180) degree shift at 1200 Hz, and a 180 (or 0) degree shift at 2300 Hz. Post-detection filtering would present the same challenges as would the Travis Discriminator (Robot design).

Back around 1987, I got interested in OSCAR satellites. I built a very simple 1200 bps AFSK data demodulator to decode data and telemetry signals from the OSCAR-9 and OSCAR-11 satellites. The demodulator employed a zero-crossing detector that worked well only if the received signal had a high signal-to-noise ratio. With the satellites using only about 300 mW of transmitter power on 145.825 MHz, and my using an indoor antenna, this was a bit of a challenge.

I knew there had to be a better way. I built a second demodulator using an XR2211 PLL-based AFSK chip that provided noticeably better performance. Then I learned about the significantly more sophisticated UoSAT demodulator designed by James Miller, G3RUH, and built a stripped-down version of the circuit he published in the May 1983 issue of Wireless World. In the opening paragraph of his article, James explains why zero-crossing detectors offer such poor performance as the desired signal and any undesired noise both produce zero-crossings, it is impossible for the demodulator to distinguish between the two, and performance (under "real-world" conditions) suffers as a result.

Food for thought...

73 de John, KD2BD
User avatar
kd2bd
Mad Scientist
 
Posts: 65
Joined: Wed Sep 28, 2016 5:47 am
Location: New Jersey

PreviousNext

Return to SSTV

Who is online

Users browsing this forum: No registered users and 5 guests