Narrow-Bandwidth Television Association

This thread will pick up the continuation of the SSTV-625 converter where "Retro SSTV Anyone?" left off.

I've dubbed it "MkII" as I plan to make some hardware changes from the first version. More will appear here in due course...

Steve A.

Steve,

Look forward to seeing your progress and good luck. If you need help with prototype building email me.

Regards Chris Lewis

Chris, that's not a bad idea! First off the layout I have done previously, and intend to use again, may not suit your method of construction. It does need to be compact with minimal stray capacitance.

Before any start is made on this the proposed changes are the processors will probably be a PIC18F45K20 for the write uC, and a PIC18F25K20 for the read device. These have less program memory but it's still plenty enough. Plus they're 3.3V devices instead of 5V allowing direct interface to an SD card. But that's for the future. They should be considerably cheaper, well, they are here.

RAM will be an AS6C1008 (128k x 8b), twice as large as it needs to be but 64kB versions are rare, plus they operate at full speed on 3.3V...GBP1.72 +VAT (Farnell stock No. 1562896), say two quid each including UK VAT.. (55ns). Other changes will be a replacement of the blanking gates with 74AC08 devices instead of HC to maintain speed at the lower voltage.

The circuit diagram of the uC/RAM/logic board will probably change little as Microchip do tend to keep pin-outs very similar between devices. But that's not a certainty!

Steve A.

Taking a few minutes out of work I have confirmed that the replacement PICs (PIC18F45K20 & PIC18F25K20) are drop-in replacements for the original 5V devices - at least for this design. There may be other differences but they don't matter and/or aren't used. So if there's any justice in the world it should be a case of changing the 5V supply to 3.3V and tweaking the code here and there (he hopes).

The progress on this is unfortunately going to be slow until I can clear this pile of 'real work' away.

Steve A.

Over the past few days I've been musing over the possibility of somewhat simplifying the NBTV/SSTV to 625 up-converter. So with a gap in the 'real work' schedule I've come up with this as a starting point...

The original used the following which doesn't include either SSTV demodulation nor NBTV DC restoration/clamping. Both of which may be done in software in the future.

Original.
1x40-pin DIP PIC18F46K22 or PIC18F45K20, Write processor.
1x32-pin DIP AS6C4008 or AS6C1008, RAM.
1x28-pin DIP PIC18F26K22 or PIC18F25K20, Read processor.
1x20-pin DIP 74HC573, Octal latch.
1x16-pin DIP DAC-08, 8-bit D-A.
2x14-pin DIP 74HC08, Quad AND gates.

With a change to a serial RAM arrangement things in terms of hardware become a little more sane.

2x28-pin DIP PIC18F25K20, one each read and write processors.
1x16-pin DIP DAC-08, 8-bit D-A.
2x14-pin DIP 74AC08, Quad AND gates. (Note change to AC devices).
2x8-pin DIP 23LC1024, Serial SQI RAM chips.
...and a drop in supply volts to 3.3V, NOT 5V anymore. Though the +/-12V is still required for the analogue circuitry.

I still have to work on the timings but I'm pretty sure this is do-able. There's two RAM chips so they can be used 'broadside' much in the way of a conventional parallel RAM. With their maximum clock rate being 20MHz results in a pixel rate of the same. A bandwidth of 10MHz should you need it. We don't.

Normally I would post a couple of GIF files here for the circuit, but since I 'upgraded' to AutoCad 2016 the option to Save As a TIFF file has vanished. In the past I would then use Irfanview to convert that to a GIF before posting here. So for the time being a couple of pdfs of the initial circuits are attached until I resolve this issue. Once resolved I'll replace the posted pdfs with in-line GIFs.

Steve A.

I've added gif versions of the pdfs, but when getting them to fit on the forum the clarity suffers. So from now on I'll post both pdfs and gifs.

Well, the board layout implications of the posting above are more than I expected! In a positive manner. I may shuffle things around to reduce stray capacitance. The differences between the MkI version and this MkII version are plain to see. MkI on the left, MkII on the right. (MkI top, MkII bottom in gif version)

But before I get carried away I MUST work on the timings - something for the weekend.

Now the question is where do I place this thread? It's both NBTV and SSTV.

Steve A.

Sorry, pdf for reasons mentioned above. I'm waiting for my Acad expert to return from Malaysia, late this evening I think.

I'm still shuffling things around, both physically and electrically so don't even think of using the above information. It's there to give an overall impression of what's involved. There's no great difference but pins used for various functions have changed.

I must mark all things "Not For Construction" until they are in a valid state.

I must work out how to globally change the font used as the current one (ISO) is rubbish.

Steve A.

Currently I'm working on the SSTV timing requirements and I have a slight dilemma. If the pixel read-rate is fixed at some certain rate which gives a correct 1:1 aspect ratio on the 625 screen for 128-lines, the input of a 120-line signal will look slightly wide. Conversely correct for 120-line will result in a slightly tall aspect ratio for 128-lines. The answer is to change read pixel clock rates. That's awkward and in this level of processor, difficult.

I have to say that in the original MkI version I kept the read clock rate constant based on a 128-line format. When 120-lines were displayed they were slightly wide, but not excessively so, tolerable. At least on real world photos, not test-cards.

A compromise may be a read-rate half-way between the two...yes, no? Switching crystals or external clock sources for the different standards could be done, but it's messy and adds complexity which I don't want.

I guess it comes down to which standard was (is) more common, 120 or 128.

I don't mind making a version biased towards 120-lines, and another biased for 128.

Your thoughts gents?

Steve A.

It wouldn't bother me would it be much the same result of 7.2 sec or 8 sec viewed on a p7 monitor using the other countries line rate ...see it but aspect ratio off a touch thats ok i think its fine ...more important the over all result i am in a 50 hz country .
yours looks a touch smaller to this one ....
http://www.qsl.net/sv1uy/128lineSSTV.html

A physical switch to select crystal wouldn't be too tricky, I assume?

The advantage of a oscilloscope-type CRT-based display is that you'll usually have a vertical gain or 'height' control you can twiddle to adjust the aspect ratio. But a 625 TV or monitor is fixed unless it's very very old (1950s). Switching crystals in the converter could be done with a very small relay to keep the crystal-chip lead lengths to a minimum controlled by a front-panel switch.

An alternative is to use something like QRP Labs 'Progrock' kit (http://www.qrp-labs.com/progrock.html) but it could be viewed as a sledgehammer to crack a nut. Very useful for other things though. The GPS discipline is optional, not essential.

The two different line rates, US 15Hz/EU 16.7Hz, are dealt with a sample-rate adjustment front-panel pot. This could be also a way of overcoming this problem by setting the sample rate such that only 120 samples are taken in a 120-line picture (It is actually 240). The remaining ones would be in within the sync-pulse and ignored.

It's something that I'll put on the 'back burner' for the time being, the rest of it requires my attention right now.

Interesting site Harry, shame some of the links are dead. 80 TTL chips, nothing unusual in those days - keeps you toasty warm on those frosty mornings.

Steve A.

I've made a decision on the aspect ratio/crystal frequency for the NBTV up-converter application. Using a 16MHz crystal for the read processor will result in a +1.7% increase in the NBTV vertical height - hardly noticeable. That's assuming the source is in a 1:1.425 aspect.

The nominal aspect ratio in NBTV is 3:2 = 1.5 which includes the sync duration. If syncs are 5% of line-time the the ratio becomes 1.425. I aim to always use 5% which comes out at 125us.

But the SSTV side of this is proving awkward. Using the same crystal the error is +8.7% tall in 128-line pictures. For 120-line pictures the error is about +2%. Crystal switching may be the only way to get this better for 128-line sources. But I'm not quite defeated yet.

Steve A.

Steve, I just bought a second hand ROBOT 400 scan converter. The thing looks like new. I haven't fired it up, but I have scanned / read the documentation. It looks to me that the extra 8 lines are used only for a "hanging" grey scale. So the picture area, which should be square, is 120 lines. While receiving I should first load the grey scale for a whole picture (128 x 128 pixels) and then overwrite the top 120 lines with the incoming SSTV.

But I haven't yet seen the picture on a CCIR monitor. Then I can measure the size on the screen. I prefer to do these things step by step.

By the way, it struck me that there is no circuit diagram in the documentation. It does no go further than a block diagram and a lay out of ICs on the large PCB. I also could not find where / how the switching over from 525 lines 60 Hz to 625 lines 50 Hz is done. But they state that both modes are possible. The same is with the line frequency of 15 Hz or 16.7 Hz. This set is for 230 V 50 Hz grid voltage, so I hope both are already in the correct mode. I guess that you have a circuit diagram?

Sorry to say Klaas, I have no documentation for the Robot 400 at all. In the days it was produced to publish openly a circuit diagram would mean that anyone could copy the equipment. Today where almost all equipment has an element of software within you're to an extent protected. Building all the hardware without having the software is pointless unless you want to write it yourself. Most processors have a code protection feature, but there's a few clever guys out there who are able to bypass it. Once the code protection bit is set the chip supposedly can't be read, only erased and re-programmed.

As for the 400, I guess you may have to spend some time trawling the 'net. If you're lucky you may be able to find someone who also has a 400 along with the schematics, a bit of a long-shot though. If you are successful I sure would like a copy, even that which you have scanned already.

As for the 128/120-line matter, if the additional eight lines were only used for a grey-scale then you could say that it really doesn't matter the aspect isn't quite correct, the 120-line 'active' part would be near enough. I have seen a few screen-shots of 128-line pictures and the grey-scale (as best as I can remember) was always at the bottom. But either way, it doesn't matter.

IF the extra eight lines were only used for a grey-scale then my dilemma is solved. I had 'sort of'' resigned myself to accepting the error on 128 lines anyway.

Steve A.

This eBay italian seller is selling one which says complete with manual and schematic.
You could contact and ask for a copy - or even purchase.