Harlequin video modifications

[PokeMon]

As I am owner of a new Spectrum clone Harlequin Rev. F I got deeper into it's design and have some improvements to the video outputs.
I found a few minor design decisions, I was not happy with and show a few modifications here improving the video signals of composite and RGB and offering a Y/C video (S-Video).

The Harlequin uses the AD724R circuit to create a composite video signal and I wonder why the signals Chroma und Luma are not used as it offers a S-Video (Y/C video) output for free which give a better color signal than the composite video. Of course RGB would be even better but many modern TVs do not offer a RGB connector at the front and is hard to connect to a flat TV hanging at the wall. Most offer a composite video chinch input AND a S-Video input for fast connection of a camera to the TV at the front side. So why not connect the Harlequin via S-Video ?

That's the story about - but step for step.

[PokeMon]

First I tried to find a mini din video adapter as these are widely available for 7-pin and 9-pin used for ATI and NVIDIA graphic cards.
So the used 8-pin in the ZX Spectrum is uncommon and does not correspond to the bigger 8-pin DIN version.
Anyway I don't know any available adpater from mini din to din.

IMG_7902kk.jpg (118.32 KiB) Viewed 20617 times

There are many cable adapters available on eBay and in the internet which meets a little industry standard:
http://www.ebay.de/itm/251389762503
http://www.ebay.de/itm/190991326974

My personal favourite would be this one:
http://www.ebay.com/itm/271509149782

But all do not correspond to the 8-pin mini din of the harlequin. Anyway I am not very happy with a SCART connector (mini din to scart).

[PokeMon]

So the first goal is to solder a custom cable with single RGB outputs, so I took a VGA SUB-D to BNC cable, cut the SUB-D and soldered a mini din connector 8 pin which is a bit diffiult with such a strong cable.

This is the goal, I need BNC plugs, at least 4 different RCA plugs and RCA/BNC adapters.

IMG_7890kk.JPG (154.85 KiB) Viewed 20618 times

This is the hard part when using a big cable which is not very flexible. So not much space for putting the cable into a mini din but a hq cable with single shielded cores was my choice. Be sure not to put too much stress to the cable as the single cable cores will break fast when bended too much. I did this work two times.

IMG_7889kk.JPG (122.09 KiB) Viewed 20618 times

To be more safe I taped the connector with transparent film (original TESA ) which makes it more strong and protect the small connector and the soldered cables. Looks not perfect but does the job for a long time.

IMG_7904kk.JPG (134.75 KiB) Viewed 20618 times

[PokeMon]

When comparing the signals, I further found different output values of the video signal for the composite signal and the RGB signal.
And the RGB signal looked a bit weak/dark on my broadcast monitor.
When switching between composite video and RGB the was a remarkable difference in brightness.

As I don't want to just increase brightness/contrast which is problematich when using different devices I looked at the video levels with my oscilloscope and checked it with the schematics. The voltage levels are quite different due to resistor mismatches. So the RGB signals feeded to the AD724R are out of tolerance, which is max. 700 mV while the AD724R is feeded with 800mV signals for 100% (intense color). This additional 15% results in a overmodulation and it depends on the chipsets what they do.

They can cut the voltage simply over 700mV reducing the difference between intense and non-intense signal which is the easiest way. Anyway it is not defined how TVs or monitors behave with these signals as they are out of standard. Due to the logarithmic behaviour of the eyes 15% more signal will be recognized as 30% in intensity. On the other hand the RGB output signal is only 620mV while it should be 700mV as well for intense signal which gives only 88% of signal which will be recognized as 77% due to logarithmics.

So in comparison to the 800mV signal the 620mV signal is only 78% which will be recognized as about 60% luminance or 40% difference while assuming the other one would be 100%. This is a big difference and caused through the resistor values R19,R20,R21 (82R to 75R) and R34,R38,R43 (42R to 75R). As the signal from the transistors have a 1.3 volt level the values should be optimized to 68R (all named resistors) to divide this signal into a nearly perfect 700mV level for 100% color (intense). The non-intense level matches more or less exactly 75% at the output, resulting in about 520mV signal which will be recognized of a reduced luminance of about 50% perception.

So my advise is to replace all resistors R19,R20,R21,R34,R38,R43 with 68R values to get the same (optimal) brightness for composite video and RGB video and finally I tested it with a broadcast monitor and measured with an oscilloscope to meet the video standards.

harlequin_rmod.jpg (176.18 KiB) Viewed 20617 times

[PokeMon]

So the next mod is a bit more difficult than just replacing the resistors. I wanted a second cable for connecting the Harlequin simply to any TV or monitor with S-Video input for a better picture quality. This gives more quality than the just one-containing-all signal with reduced color information due to modulation with 4.433 MHz. There are visible artifacts which can be viewed when setting border for example to a colour other than black or white (grey). The status line is flickering in white colour - more artifacts can be seen using the test picture from the Ian Logan ROM.

With S-Video the artifacts are eliminated - anyway not as crispy as a pure RGB signal but quite better than the composite. So I took a standard 4-pin S-Video cable and simply soldered the other side to 8-pin mini din connector. I decided to use pin 2 and 3 for this purpose as luminance (pin 2, LUMA) and chrominance (pin 3, CRMA). The connector is more easy to change due the much thinner (but perfectly shielded) S-Video cable.

IMG_7906kk.JPG (92.36 KiB) Viewed 20614 times

And this is the S-Video modification:

harlequin_svmod.jpg (166 KiB) Viewed 20612 times

First C35 (3.3nF) and R49 (2.7k) are removed. These are not needed and it is unclear why the CHROMA signal is connected to the composite sync signal this way and what was the initial purpose. The CHROMA and LUMA signals from U48 (pin 9 and 11) are then feeded via 75R resistors directly to the 8-pin mini din connector pin 2 and 3. R8 is removed permanently and D11 and R12 are replaced with 75R resistors but connected directly to pin 9 and 11 instead of VCC and GND. I decided to solder one pin of the resistor into the board and the other one to wire directly to the desired pins as shown on the next pictures.

So the modification is not very superficial at the first view. I avoided long crossing wires and put the longest at the backside.

IMG_7897kk.JPG (157.21 KiB) Viewed 20612 times

[PokeMon]

Beginning at the AD724R to connect the unconnected LUMA pin 11, this can be done on top side only.
I connected it to the side of C35 which is connected to R49 and continued this connection under the board from R49 position.

IMG_7900kk.JPG (136.58 KiB) Viewed 20613 times

Here continued at the backside and for the other signal I used the other pin of C35 which is connected directly to pin 9 and feeded both through a hole which I left for a modification of R48 which was advised from the developer (team) before.

IMG_7901kk.JPG (137.43 KiB) Viewed 20613 times

And on the top side I connected it shortly to the other end of replaced R12 and the other resistor R8 placed at the place of the no more needed D11. This was more near than R8 which was placed a bit far away and is connected to pin 2 of mini din connector, too.

IMG_7898kk.JPG (154.32 KiB) Viewed 20613 times

[PokeMon]

And what is the price you can get for this work ?
A nice picture with better quality than the composite video through the S-Video input of TVs offered at the front of modern TVs.
The SCART connectors are sometimes hard to reach with wall mounted flat TVs.

Here is the picture of the S-Video - more clear and not that flurry for text displays.

IMG_7916kk.JPG (129.07 KiB) Viewed 20617 times

And here in comparison the composite video signal where you can see some artefacts, especially for text on a colored background. It is not easy to catch these moving pixels or flurry action on screen with a still camera but you can see a shadow below the text at the blue color:

IMG_7921kk.JPG (162.55 KiB) Viewed 20617 times

This is depending on the displayed colors and very remarkable when using magenta, cyan or even yellow as background.

IMG_7923kk.JPG (173.83 KiB) Viewed 20617 times

The modification removes two signals of the SCART connector which are just a bit comfort but not needed when connecting the Harlequin.
One is used to just switch to the input channel when the Spectrum clone is connected or switched on, the other is used to determine the desired aspect ratio (4:3 or 16:9). This can be controlled manually with the remote control as well so I find the S-Video output more useful than these features.
But up to you.

[superfo]

The Harlequin uses the AD724R circuit to create a composite video signal and I wonder why the signals Chroma und Luma are not used as it offers a S-Video (Y/C video) output for free which give a better color signal than the composite video.

Actually Ingo suggested me to add a S-Video output too, by using jumper to switch between RGB and S-Video. But I didn't have much time and it need more area to add another 2x220uF and 2x75 Ohm so I decided not to add it (I know you don't use 220uF, I just want to follow data sheet)

Why mini-din 8? I didn't know, it's just easy to find in my area. Actually I just need R, G, B, Sync and GND signal. If I saw mini-din 9 at that time I may use it.

My Harlequin Rev D with AY-sound, PS2 Keyboard and 128K upgrade, Also there is DivMMC interface connect to edge connector .

[PokeMon]

Thanks for clarifying. Well the mini-din is not uncommon and I just wondered why you did not use a more common 9-pin or 7-pin version and one of the standard pin outs used from graphic cards. Would make life easier.

But you could checkout the voltage levels for the signals and my proposal to use other resistor values for R19,R20,R21,R34,R38,R43 to get a more standard video signal. If you like to follow the datasheet, the max input level is 714mV pp (full range signal) for the RGB inputs of AD724 while it was in reality more than 800mV in my Harlequin. By the way is 714mV max signal is valid for NTSC, in the PAL norm it is 700mV exactly. This way you get a more harmonized output signal which has identical level for RGB and composite video.

[superfo]

But you could checkout the voltage levels for the signals and my proposal to use other resistor values for R19,R20,R21,R34,R38,R43 to get a more standard video signal. If you like to follow the datasheet, the max input level is 714mV pp (full range signal) for the RGB inputs of AD724 while it was in reality more than 800mV in my Harlequin. By the way is 714mV max signal is valid for NTSC, in the PAL norm it is 700mV exactly. This way you get a more harmonized output signal which has identical level for RGB and composite video.

About RGB input of AD724, Jose Leandro (who made the prototype of Harlequin for me) and McLeod/Ideafix modify from the rev A. They want to connect both composite and RGB monitor at the same time. You can read detail how McLeod work on schematic from link below (file name "improving_rgb_output_harlequin_superfo.pdf")

https://onedrive.live.com/?cid=3AC85591 ... 90140D!107

Have you try your Harlequin to both composite and RGB at the same time? If yes how it look like?
I don't have RGB monitor so I cannot tell.