ZX Spectrum+ video ? red on green problem

Boleslaw Jamroz · January 2015

This is my first post on the forum, so let me introduce myself. I am retired for several years. For 30 years I have been a fan of the ZX Spectrum. I learned programming on it, which allowed me to change profession. Since then I have worked as a computer programmer. Now I have a lot of time. My sentiment to an old friend ZX Spectrum ordered me to improve it.

My ZX Spectrum+ (issue 4a) from the beginning does not work as it should.
There are no sharp boundaries between the colored stripes.
Black letters on a white background are surrounded (the left and right sides) by lighter spots (see text “scroll”).

Very bad letters are displayed when written in red ink on green paper. On the sample photos are shown printed uppercase "A" and some graphics.

Recently I decided to tackle this problem.
I replaced all the electrolytic capacitors. Also I replaced the transistors TR8 and TR9. Unnecessarily, because the original transistors were OK. I used a trimmer to adjust the quartz 14.000 MHz frequency. I checked all the resistors in the analog part of the system. Unfortunately, I got no improvement.
I measured the power supply voltages at the relevant circuit points. All voltages are in line with service manual.
Quartz:
X1: 14.000 MHz (I could not get 14.000000)
X2: 4.433619 MHz (PAL)

Are the photos, that I showed, evidence of damage to my ZX Spectrum+ ? Whether it may be a feature of this Spectrum release?

fog · January 2015

Are you using the composite mod also ?

Boleslaw Jamroz · January 2015

I skipped the RF module. I use Video In on the TV.

balford · January 2015

I think what you're seeing here is standard 48k 'dot crawl'. It's caused by the system oscillator (14MHz) not being in sync with the PAL clock used by the colour encoder chip, as they're generated separately.

They fixed this on the 128 machines by having a single oscillator and deriving both CPU and colour clocks from that single source.

B

Boleslaw Jamroz · January 2015

balford wrote: »

I think what you're seeing here is standard 48k 'dot crawl'. It's caused by the system oscillator (14MHz) not being in sync with the PAL clock used by the colour encoder chip, as they're generated separately.
They fixed this on the 128 machines by having a single oscillator and deriving both CPU and colour clocks from that single source.
B

Thank you for the explanation of the causes of the phenomenon.
It follows from this that it is a disadvantage of all ZX Spectrum with two quartzes. This is a serious drawback of the Spectrum 48K electronic system.

rga24 explains this in more detail:
?S-video eliminates dot crawl. Dot crawl is caused by the compromises inherent in composite video - the combining of black and white (luminance) information and colour (chrominance) information into a single (composite) signal. It exists in every composite colour system, to one extent or another. It's impossible to perfectly separate the two signals once they have been combined. Dot crawl is particularly bad on the 48K Spectrum because there are two crystals - 14 MHz and 4.43 MHz. The relationship between the pixels and the colour subcarrier changes all the time. Other systems, such as the 128K Spectrum, have a single crystal and a fixed relationship between the pixels and the colour sub carrier.?

With this in mind, I separated chrominance and luminance. Then I increased the level of the chrominance signal in an extra transistor. I brought these two signals to S-Video input of my TV. Now the picture is perfect.

Thank you for your help.

Falco-F8L · January 2015

Boleslaw Jamroz wrote: »

With this in mind, I separated chrominance and luminance. Then I increased the level of the chrominance signal in an extra transistor. I brought these two signals to S-Video input of my TV. Now the picture is perfect.

May I ask you to detail and document your mod so others can take advantage of it? Thanks!

Boleslaw Jamroz · January 2015

Today I put the plate with the elements inside the RF module housing. In a few days I will give details and photos.

Ox. · January 2015

https://www.youtube.com/watch?v=KKFOxLqHl1s

Everything is in the above video, the example shown uses 2 output from the spectrum for chroma and luma but this could be combined into a single standard s-video out connector if required so that a ready made standard s-video cable can be used.

Boleslaw Jamroz · January 2015

Modification “S-Video" removed the phenomenon of "dot clawl". However, the picture is not crystal clear. You can see a phase shift between colors. This is clear seen in the details of the width of one pixel. For example, the letters do not have a color (red or blue on a green background). Also, you can see the color shifts on the edges of the larger elements of the image.
There is also a phase shift between the luminance and chrominance. I tried to correct this shift. I put a small inductance in the path of the luminance (black element in the shape of roller near the S-Video socket on the picture number 7). Shifting got, but also suppressed the high frequencies, which manifests itself in blur.
But keep in mind that the structure was built over 30 years ago and at that time all was content. As lovers of antiques, we need to come to terms with the imperfection of the device. This is its undeniable charm.

1024MAK · January 2015

To match the luminance and the chrominance, in theory, there should be a delay line. But in practice, when the Spectrum was made, the cheap imperfect use of a resistor and a capacitor was considered good enough.

Keep in mind that display technology has moved on a lot from the curved screen 14 inch CRT televisions that only had UHF inputs that the ZX Spectrum was designed to be used with.

@ Boleslaw Jamroz - thanks for documenting your modification :smile:

Mark

ikci · January 2015

Hi Boleslaw :smile: Nice to see you here. We are from the same city.

Don't forget to visit www.speccy.pl - this is Polish forum about Zx Spectrum.

balford · January 2015

That's a nice modification, very neatly executed :) Thank you!

B

Boleslaw Jamroz · February 2015

I decided to check the quality of the video signal from the YUV-RGB converter for Spectrum48K by Pera Putnik 1995.
As a base I used the serial printer interface. It has a lot of space inside, so I placed there the converter plate.
I used BC550 and BC556 transistors (300MHz). Resistor R23 (15k) has been replaced by an adjustable resistor 20k. To increase the stability of the converter, I put the 4k7 resistor between the "CLK" CD4027 and ground.

And here is the result (better than S-Video):

To connect the converter to the TV, I used a flat 10-wire cable. Every second wire of the cable is connected to ground. This is a good way of shielding signal wires from interference.

To free bus YUV of unnecessary traffic and load, I cut off (only for test) the following elements of ZX Spectrum 48K: C68, C69, C70, C71, R48 and quartz X2.
I did not notice the impact of this change on the quality of the video, as well as on the noise level.

1024MAK · February 2015

Looking good there :smile:

If there was a smilie for hand clapping on this forum, I would would have added that to my post :grin:

Mark

Boleslaw Jamroz · February 2015

Diagram of the YUV-RGB converter for Spectrum 48K can be found on the website of Mr. Pera Putnik.

Boleslaw Jamroz · February 2015

Black horizontal lines on the upper and lower edges of the graphics and letters I removed reversing the lines Q and NOT-Q of the CD4027 (YUV-RGB converter).

Before the change:

After the change:

Boleslaw Jamroz · February 2015

ikci wrote: »

Hi Boleslaw :smile: Nice to see you here. We are from the same city.
Don't forget to visit www.speccy.pl - this is Polish forum about Zx Spectrum.

Hi ikci
Thank you for the information.

Boleslaw Jamroz · February 2015

The quality of the video, obtained from the YUV-RGB converter by Pera Putnik 1995, should be compared with the image quality, which give other professionally done devices of this type for Spectrum 16K, 48K, 48K +. I mean two converters: "Adaptation Peritel ZX Spectrum" (France) and "Ferguson MA 20 RGB Interface" (UK). Unfortunately, I do not have access to them.
Therefore, I appeal to colleagues, having such converters, for help. It's about taking a few pictures with close-ups of screen printed letters "AAAAAAAAAAA" in red and blue, on a green and white, as I did, to demonstrate results.
I will be grateful.

kpuchatek · February 2015

Hello from Poland !

For RGB crystal clear solution ;) you may check:
http://www.fruitcake.plus.com/Sinclair/Spectrum/Spectra/SpectraInterface.htm

But this requires quite advanced piece of technology.
Then you get quality similar to 128K picture or maybe better. Should buy it one day, but somehow I prefer CVBS as TV connection.

Boleslaw Jamroz · February 2015

kpuchatek wrote: »

Hello from Poland !

For RGB crystal clear solution ;) you may check:
http://www.fruitcake.plus.com/Sinclair/Spectrum/Spectra/SpectraInterface.htm

But this requires quite advanced piece of technology.
Then you get quality similar to 128K picture or maybe better. Should buy it one day, but somehow I prefer CVBS as TV connection.

Hello!
Thank you for the information. However, I can not open the web page.
But I was looking here:
http://www.sellmyretro.com/offer/details/RGB-SCART-Interface-for-Sinclair-ZX-Spectrum-16K~~48K-2740?lang=en_EN

The picture is really very good. Almost like digital.

Boleslaw Jamroz · March 2015

It turns out that not only the Sinclair Cambridge Scientific had trouble with colors. Big IBM also had a problem with it. :roll:

http://en.wikipedia.org/wiki/Color_Graphics_Adapter#mediaviewer/File:CGA_CompVsRGB_Text.png

Renegade · November 2015

Boleslaw Jamroz wrote: »

Modification “S-Video" removed the phenomenon of "dot clawl". However, the picture is not crystal clear. You can see a phase shift between colors. This is clear seen in the details of the width of one pixel. For example, the letters do not have a color (red or blue on a green background). Also, you can see the color shifts on the edges of the larger elements of the image.
There is also a phase shift between the luminance and chrominance. I tried to correct this shift. I put a small inductance in the path of the luminance (black element in the shape of roller near the S-Video socket on the picture number 7). Shifting got, but also suppressed the high frequencies, which manifests itself in blur.
But keep in mind that the structure was built over 30 years ago and at that time all was content. As lovers of antiques, we need to come to terms with the imperfection of the device. This is its undeniable charm.

Very nice indeed.
It is always good to watch one same modification made a different way by someone else.

However, I didn't quite understand if the result of using an inducter was positve or not. Did the shifting got any better? Or worse? By the way, which value or range of values did you use?

Just one last thing that's intriguing me, why the choice of a 1uF capacitor on the path of chrominance?

Julio · December 2015

Boleslaw wrote: ...Resistor R23 (15k) has been replaced by an adjustable resistor 20k. To increase the stability of the converter, I put the 4k7 resistor between the "CLK" CD4027 and ground.

Can you put your schematic?
You add another resistor to scheme?
regards
Julio.

ZX Spectrum+ video ? red on green problem

Comments