Computers Overview
Commodore PET
Sinclair ZX80
Sinclair ZX81
BBC Micro
Commodore 64
Sinclair Spectrum
Memotech MTX
      Hardware Hacks
      Legacy (1980s)
              Games ROM
              I/O Ports
              System Bus
          Design Data
      PAL Reader
    User Groups
    Video Wall
Memotech CP/M
Atari ST
DEC 3000 AXP
Raspberry Pi



"MTX Plus+" Video Board


Building the Prototypes

The majority of the components mounted on my prototype board

The three position jumper block adjacent to the GAL socket is the "Test" jumper position, depending on whether it is set high or low, the GAL decodes I/O port addresses for MTX or MTXPlus+.

With most of the wiring in place

Martin's version of the prototype board at a similar stage of construction as mine

Martin's board, partially wired, with mainly the power lines and RAM chip wiring in place.

As usual, much neater than my efforts.

Martin's board, test fit to his backplane, not fully wired at this point, but showing power on to the board

The almost completed video main board - the only components, apart from the ICs, that are missing are the oscillator and capacitors for the VDP clock, these should be arriving shortly.

You can also see the blank daughter board attached to the main board and may notice that the DB is larger than I originally intended. The DB is going to have quite a few components on it so I have slightly expanded its size and widened it to take advantage of the available free space on the main board.

The completed video main board with the blank daughter board removed, ready for testing.

The board was initially checked with a multi-meter, then, with a fuse inserted into the 5V fuse holder and no chips installed, it was connected to a standard MTX using the MTXPlus+ adapter board to check that operation of the MTX was unaffected.

Solder side of the completed main board

 . . . . Construction of the daughter board is next . . . .

Martin's main board - now fully wired

Video Conditioning Daughter Board

I have been pretty busy, so got a bit behind with the assembly of the daughter board, Martin has pressed ahead and this is his, almost complete, daughter board (DB).

We have both decided to see if we can do without the PAL clock at the moment, so those components are missing off this board.

Martin's Daughter Board

Martin's video board with the daughter board mounted.

The I/O connector on the right hand side has only has connections to a subset of the pins - enough for composite video at this point. 

Similar to the that way I tested my board, Martin used the MTX adapter board with the video board "Test" jumper set to "Test" to check that the board did not affect operation of the MTX. Components were gradually added to the bare board - VDP, DB, PAL, HC374 and 128K of DRAM before the CXA1645. (The MTX is connected to an old green screen monitor.)

Martin called this photo a "sanity test", appropriate, when he went to insert the video decoder chip, my insanity was confirmed ! . . . . . .

At this point, a faux pas with my design/layout was revealed.

When the CXA1645Ps from Lez arrived, I did not look at them, or the datasheet, closely enough - the package is 400mil or 0.4" wide, rather than the more usual 0.3" pitch, so Martin had to make a small alteration to his daughter board to cater for the extra width, obviously, I will need to do the same, but as I had not installed the socket yet, it will be easier for me.

With the CXA1645P installed, Martin was able to write to the V9958 - the MTX has sent OUT commands to I/O port 137 and the background colour of the monitor connected to the V9958 is a nice shade of blue.

The MTX monitor shows that port 137 (89h) has been written to instead of a port in the 152 to 155 (98h to 9Bh) range intended in the design. There is a small error in the GAL program which Martin has now fixed to make the V9958 respond to the intended I/O port addresses.

With the revised GAL code loaded, Martin was able to write to the correct ports and generate patterns on the V9958 monitor, although some of the data values read back after being written were incorrect.

On further investigation, Martin found a small construction error, which when corrected, unsurprisingly, gave better results.

(The display was connected to the V9958 Composite Video output, some "smearing" and over-saturation of the colours is visible.)

With the video card in "Test" mode, Martin was able to set up VDP Text 1 mode and program enough characters in the pattern name table to generate the "Ready" prompt, but this photo is much more interesting!


The photo shows the "Test" jumper in the "normal" position - the MTX is writing to both VDPs through the original MTX port addresses - i.e., driving independent screens.


The V9958 display was not brilliant, it was still using the composite video output, but so far, it's a great result.

Now how about this ?

The only difference here is that Martin did a bit of "tweaking" with the variable capacitor on the daughter board.

There is still some smearing, and the colours may still be a little over-saturated, but the display is still composite video, at NTSC frequency, though a UK specification TV.

My Daughter Board

OK - I'm catching up with Martin again and my DB has now been built. Like Martin, I have not added the PAL clock components until I see whether the NTSC output is acceptable.

I did add a socket for an 74LS04 should the PAL clock be required as it was easier to add the socket before doing some of the other point-to-point wiring.

Solder side of the daughter board - as you can see, I made use of some of the component leads and created a number of wire links to reduce the amount of Kynar wire on the board.
My daughter board added to the main board, ready for testing.

When I first tried the DB, I started getting problems with the MTX being used for testing crashing with the video board connected. After wasting quite a bit of time inspecting and retesting the DB, the fault was apparently a hardware problem with the GAL

With the GAL replaced, I have now been able to generate output from the V9958.

This poor quality iPhone photo shows the composite video output from the V9958 configured to respond to the same I/O port range (1 & 2) as the MTX VDP. The picture has not been optimised and the TV needs adjusting, as the banner shows, the video is 480i at 60Hz, i.e., is being generated using the V9958 NTSC clock signal.

For comparison, here is an equally bad photo of the normal MTX composite video output, in this case, the video is 576i at 50Hz - the normal UK PAL configuration.

Even within the limits of the poor quality photos, although the NTSC video output does work on our multi-standard TVs, there is an appreciable loss of quality between PAL and NTSC.

In this close up of the V9958 screen display, you can see that the characters appear rather more pixelated than the corresponding display from the V9929 (below).

I guess that this should not have been too surprising as PAL has some 20% better vertical resolution when compared to NTSC, with PAL having 580 visible scan lines, rather than the 480 of NTSC.

The same portion of the screen from the normal MTX video output, photographed at roughly the same resolution as the photo above

Once the boards have been fully tested and evaluated, this difference in quality may be enough of a driver to add the PAL sub-carrier clock components.

For comparison with the output from Martin's card, the output from my board running the same test program as Martin. In my case, since I didn't have two composite video monitors handy, the program is running under FDX BASIC, with the FDX driving a VGA monitor.

SCART Audio & Video Outputs

Martin got to the stage of testing the SCART connections before I did and found a couple of minor problems.

When connected via the TV SCART input, the composite video signal wasn't generating a picture on the TV. We were not aware that SCART cables have cross-overs in the connections for audio and composite video. Making the cross-over between pins 19 and 20 in the SCART connection allowed the composite video signal to be displayed on the TV. (The pin-out for a fully wired SCART cable is shown on my notes page.)

Connector Break-Out "Board"

Audio / Visual connectors fitted to TVs are typically quite large - SCART connectors being the largest and most common connector used in all but the most recent devices for the European market. Although smaller, S-Video and phono type connectors also have a relatively large footprint, so, given the limited space available on the daughter board, it would not have been possible to mount the connectors on the board itself.

The daughter board has a single 20 pin header designed to carry the audio / visual signals to the relevant connectors, both Martin and I have built small "break out" boards where the individual SCART, S-Video, and phono connectors will be installed. In my case, since a SCART socket is quite large, to reduce the stress on the break out board connector, I have used a 15-pin VGA type connector for the SCART lead which required me to make a customised 15-pin to SCART cable, but which seems to be a good solution.

The Video board connector break-out board


Connector Function

S-Video socket


Phono socket (RCA socket)

Composite Video

3.5mm stereo socket

Stereo Audio

VGA Connector



RGB vs. Composite Video Outputs

I finally got around to making up a cable to connect the RGB video outputs to my TV.

The images opposite are a couple of quick screen shots with my iPhone, other than minor cropping, they have not been retouched.

This one is using the NTSC Composite Video output and shows the boot screen as MTXPlus+ is started up.

This image is of the NTSC RGB output

If you open up the full size image, you can see the improved quality of the RGB output, the photo does not really do it justice, the difference in picture quality seen on the TV is much more obvious.

PAL vs. NTSC, Composite vs. RGB

And again . . . better late than never, I have now added the PAL components to my daughter board.

When Martin tried the composite output in PAL mode, he found that there was considerable dot crawl, so it will be interesting to see if I find the same.

There are now quite a few jumpers on the board to select the different audio and video modes, I have annotated this photo to identify the jumper positions on the board.
The small photos embedded in this webpage don't really allow for meaningful comparisons to be made, for more details, click on the image to open the raw photo. These photos were taken with an iPhone and, as above, have not been edited, other than some minor cropping to give broadly similar image sizes.

It should also be noted that I tend to use the system with a 4:3 aspect ratio on the TV, the image does not fill the widescreen of a modern TV. but the squarer image is more true to the original.

This photo is of the NTSC composite video output with a PANEL display on the screen - I use PANEL as a quick way of producing some text to be able to compare image quality.

As the TV status banner shows, this is a 480 line, 60Hz signal, confirming that it is NTSC.

In addition to the jumpers that configure the CXA1645 for NTSC or PAL, the VDP has internal registers that set its Composite Video output to NTSC or PAL and set the number of display lines to 192 or 212.

This image is the result of leaving the hardware jumpers set to NTSC and setting the VDP mode to PAL with 212 lines. You can see the absence of colour as the PAL colour burst clock is missing and the corruption at the bottom of the display as the MTX ROM only uses 192 line VDP modes.

With the board jumpers and VDP mode set to PAL, I had to make changes to the PAL clock circuit to get a stable colour output. (The PAL clock design is now based on the original MTX computer system clock.)

The PAL composite video output gives a reasonable quality image, and is on a par with the NTSC composite output. Both are significantly better than the RF output that I used BITD, but, as expected, are far inferior to the RGB output, so, MTXPlus+ will normally be configured for RGB.  


And finally . . . . . .

a comparison of the VDP RGB output in NTSC and PAL modes

This image shows text printed from MTX BASIC with the VDP in its default mode, i.e. NTSC - 480 lines at 60Hz.

This image shows exactly the same text being displayed with the VDP set to PAL mode - 576 lines at 50Hz




Roger Samdal's Spectravideo site, VDP page

GR8BIT project, Eugeny Brychkov, GR8BIT Knowledge Base article on adding Composite and S-Video output (KB0014).


mailto: Webmaster

 Terms & Conditions