RGB/Mono 625 SPG

[Steve Anderson]

Chaps,

This RGB-Y,Pb,Pf matrix is now complete, it was no trivial task either! The resultant circuit diagram is shown below. It's about as simple as it can be while still driving 75Ω cables at standard voltages and only using NPN transistors. It uses 10 transistors, all the same, here I used BF199's as had a few...in fact only nine, that's why one is different, but if I had one more I would have used another BF199.

Any transistor you may find in a video IF strip will do, so that's quite a choice, I would have bought some more BF199s but my supplier has ceased stocking them. So the next batch will be KSP10s (Fairchild).

It's pretty straightforward, the resistors at the input produce a Y signal which is buffered and inverted by the differential pair TR202/3 which have an emitter current-source from TR201. The Y signal at TR203s collector then has the composite syncs glued on and buffered to a 75Ω source via TR204/5. The RGB input source is straight out of a PIC micro, no 'HC541 is needed in this application.

The -Y signal at TR202s collector is buffered by TR206 and fed to both colour-differencing circuits. These resistively do the subtraction and are again buffered to 0.7V by a pair of transistors.

This quite a wideband design, the channels are not filtered but if filtering is required for feeding a PAL, NTSC or SECAM coder then TR204/231/241 could easily be made filters with the addition of two resistors and two capacitors each.

There is a small glitch on the Pr channel at the rising edge of the white bar, it's very narrow, around 15ns and probably due to building this sort of wideband circuit on veroboard. This would not be visible on an analogue scope but these digital ones show them up, sometimes to my annoyance!

If this were to be used to turn an RGB NBTV source into Y,Pb,Pr then standard general purpose trannies can be used, any of the BC54x series for example.

The screen-shot below shows how sharp the colour transitions are, the most telling is the green to magenta one. Some of the colours look a bit 'off', I put this down to the screen-camera combination. Looking at the screen live they look spot-on.

This is in no way a matrix for professional applications, but it's good enough for ATV, domestic and hobby use.

Steve A.

An afterthought...the total +12V current is around 80mA including the PIC and its +5V regulator.

[Steve Anderson]

One photo I omitted yesterday was the following...

One chip, that's it, a 625 SPG, or it could be a 405 variant, or 819, or 441....

Think of 1950, how would you have done it then? Seems impossible...

Steve A.

[AncientBrit]

Excellent Steve, well done.

Regards,

Graham

[Steve Anderson]

Work hasn't stopped on this but has been drastically slowed down by 'real work' that keeps the bills paid.

The SPG PIC has been upgraded to a PIC18F2620 which is larger in pin-count (28 pins) and program memory (64kB, 32k instructions). I very much doubt that not even half the program space will actually be required so will fit into an 18F2525 (48kB) which is slightly cheaper but otherwise identical.

The reason for the change was a requirement for a TeleText type of display on the TV at home showing various parameters around the house. Temperature, humidity, power, and keeping various records (logging).

At this stage I only plan to have one page, filling that up isn't going to be easy! It's fed into a dedicated component video input on the TV, not encoded into a video signal like real TeleText.

As the TV is wide-screen I have increased the number of columns to 64 from the original 40. Depending on the font I use the number of rows could be from 22 to maybe 28.

One of the first things I had to ascertain was the degree of overscan (if any) on the screen, I quickly made a simple test-pattern (photo below) which starts and finishes right on the line and frame blanking, I was surprised that there is no overscan in the horizontal direction but a small amount vertically. Hence the reason for 22 rows rather than my original 24.

The overscan on a CRT display would be much greater.

Things have moved on from here but there's not much to show at this stage...more anon.

Steve A.

[Steve Anderson]

Work continues (slowly) on this wide-screen TeleText interface. The Read processor/SPG is top-left (see photo below) and is basically complete. It can display the full character set on the screen, any character in any row of 27 and any column of 64.

At this stage it's monochrome, the upgrade to colour will come later, hence the reason for two missing RAM chips (lower center) and one MUX (upper centre).

Today it gained its second processor to handle the asynchronous arrival of data from the server (upper right). I have yet to start on the software for it.

The two empty sockets (lower right) are for comms, a MAX232 to communicate with a PC during development and a MAX487 (RS485) to interface with the house network...i.e. it's not Ethernet. The MAX232 will be removed once complete (I hope!!).

The RGB output of this device will feed the matrix described earlier.

Note that I practice what I preach regarding bypassing/decoupling of supplies and fat grounds. 11 disc-ceramic caps, all to keep the supplies clean. With the clock oscillator running at 40Mhz you have to. (The metal can at the top).

Steve A.

[Steve Anderson]

Continuing...the two MUX chips have been deleted by the cunning use of software gating of write data, this reduces the chip-count to eight (excluding the MAX232).

In time the Read/SPG µC will be shrunk to an 18F1320 (18-pins) as the '2620 is total overkill in this application. The Write µC will stay as is due the amount of internal RAM required.

Steve A.