Slightly different take on the monitor output for the CPC464 on SCART and GT65

[retro space]

I took a slightly different path with the monitor output. I desoldered the 2 resistors that mix the luma signal on one side, and made a bridge with 330Ω to the 5V rail, so the luma output pin carries a signal to toggle RGB in on TV's over SCART. I did this when I got my CPC 464 3 weeks ago.

But then I ran into a too affordable Schneider GT65 monitor, almost new in box. And I was puzzled why I did not get a picture, only a full green image. Then I remembered I disabled the luma line for RGB switching over SCART. So now I needed the luma signal again. So I opened up my CPC once more, and connected a switch to the landing point of the 2 resistors. One input is the signal from the two "on one side loose" resistors, the other input is the 5V via 330Ω. Now I can use the output for both RGB SCART and the green GT65 screen, without the need for a cable with a separate power pass through. Together with the internally mount ULIFAC, I have a very clean CPC 464 now which works on both the green screen and RGB SCART TV's. Maybe this can inspire more to do this little mod.

[Benedikt]

That leaves me wondering whether a similar approach could work without internal modifications.

Would it be possible to DC-decouple the CVBS luma signal, treat it as AC, run it through a transformer and rectify it?
If so, can the output be strong enough to switch a SCART input to RGB mode?

[retro space]

That sounds more complex. I desoldered 2 sides of a resistor, and soldered a third resistor to a solder blob of the DC socket. You can reverse that almost invisible.

[Benedikt]

That sounds more complex. I desoldered 2 sides of a resistor, and soldered a third resistor to a solder blob of the DC socket. You can reverse that almost invisible.

Actually, looking at the schematic (CPC 6128, at least), it should be safely possible to simply short the Luma output to +5V.
Granted, the 150Ω resistor to Ground will swallow roughly 0.17W, but it is probably a 0.25W part that can handle that.
Even your internal mod might then be possible with a pair of test leads, sans soldering.

[Bryce]

A simple capacitor in the SCART cable is enough to switch the RGB signal. It only needs a voltage, it will get by on most TV's with almost no current.

Bryce.

[retro space]

Do you have a schematic? No external power source please.
What I don't get, is that the TV has no problem with the P2000T RGB output, I don't insert 5V there neither.

[Bryce]

Do you have a schematic? No external power source please.
What I don't get, is that the TV has no problem with the P2000T RGB output, I don't insert 5V there neither.

https://www.cpcwiki.eu/index.php/File:Alt_Scart_Connect.PNG

Bryce.

[Benedikt]

A simple capacitor in the SCART cable is enough to switch the RGB signal. It only needs a voltage, it will get by on most TV's with almost no current.

Bryce.

Right. I totally forgot about the /SYNC line.
But is it guaranteed to work? I.e. will it reliably output at least 1V to an input with 75Ω impedance?
I have a feeling that it is either just within spec or narrowly not within spec.

[retro space]

Do you have a schematic? No external power source please.
What I don't get, is that the TV has no problem with the P2000T RGB output, I don't insert 5V there neither.

https://www.cpcwiki.eu/index.php/File:Alt_Scart_Connect.PNG
Bryce.

Thanks, it worked. I returned the main PCB to its original state, except I still have a ribbon cable soldered to the empty holes next to the edge connector, so my Ulifac can reside internally.

[retro space]

Then again.... maybe not. It works on a Philips TV, but not on a Sony. The previous 5V switch mod did work on the Sony.....
The Sony shows the RGB image for 1 second, then it shifts half a centimeter horizontally and the image becomes B&W.
If you enter the settings menu, the part that has the menu overlay becomes coloured.

[Benedikt]

Before you do your old mod all over again, you could try to weaken the 220Ω resistor between Gate Array and /SYNC output by e.g adding another 220Ω resistor in parallel.

If we assume that the Gate Array outputs behave roughly like normal TTL outputs, their High level should be roughly 3.6V.
The 220Ω resistor in series and the 75Ω input impedance of the RGB blanking input mean that the latter only sees at most (75Ω / (75Ω + 220Ω)) * 3.6V = 0.68V.
That is without the capacitor that is supposed to turn the /SYNC signal into a DC voltage.
If we assume in our back-of-the-envelope calculation that the HSYNC pulse takes 16 out of 1024 pixel clocks per line and that the VSYNC pulse is 16 lines long, we loose roughly 6.7% of the above.
We then end up with 0.63V, which is way below the 1V to 3V specified in EN 50049-1 "Peritelevision connector" for a logical "1" and very close to the 0V to 0.4V specified for a logical "0".
Even if the Gate Array's High level is 5V, we still only get a filtered DC voltage of 0.88V.

With the additional 220Ω resistor in parallel, we should get at least (75Ω / (75Ω + 1 / (1 / 220Ω + 1 / 220Ω))) * 3.6V * 93.3% = 1.36V.
The combined resistance of 110Ω instead of 220Ω still limit the base current of the KTC1815 NPN transistor in the GT65 to a bit less than the specified maximum of 50mA.
You could therefore use the CPC with the additional resistor with both, the GT65 and a SCART TV set, respectively.

[retro space]

For now I've solved it a bit different: I have both a Sony KVM 1450D and a 1440D. The 1450 has 2 input modes RGB and Composite you can go through with the input select button, where the 1440 does an auto-select based on the switch pin signal. I moved the CPC from the 1440 to the 1450, as it allows manual RGB selection. Super clear 80 char and colour! The perfect display for a CPC 464.
I'll remember the 110Ω change for the next time I open the CPC.

[Bryce]

Before you do your old mod all over again, you could try to weaken the 220Ω resistor between Gate Array and /SYNC output by e.g adding another 220Ω resistor in parallel.

If we assume that the Gate Array outputs behave roughly like normal TTL outputs, their High level should be roughly 3.6V.
The 220Ω resistor in series and the 75Ω input impedance of the RGB blanking input mean that the latter only sees at most (75Ω / (75Ω + 220Ω)) * 3.6V = 0.68V.
That is without the capacitor that is supposed to turn the /SYNC signal into a DC voltage.
If we assume in our back-of-the-envelope calculation that the HSYNC pulse takes 16 out of 1024 pixel clocks per line and that the VSYNC pulse is 16 lines long, we loose roughly 6.7% of the above.
We then end up with 0.63V, which is way below the 1V to 3V specified in EN 50049-1 "Peritelevision connector" for a logical "1" and very close to the 0V to 0.4V specified for a logical "0".
Even if the Gate Array's High level is 5V, we still only get a filtered DC voltage of 0.88V.

With the additional 220Ω resistor in parallel, we should get at least (75Ω / (75Ω + 1 / (1 / 220Ω + 1 / 220Ω))) * 3.6V * 93.3% = 1.36V.
The combined resistance of 110Ω instead of 220Ω still limit the base current of the KTC1815 NPN transistor in the GT65 to a bit less than the specified maximum of 50mA.
You could therefore use the CPC with the additional resistor with both, the GT65 and a SCART TV set, respectively.

Very nicely calculated. The bit that's missing is the capacitor, which is extremely difficult to calculate, but it's buffering will have an additional positive effect on the RMS value of the PWM signal. You still wouldn't have a 100% accurate answer though, because the GND of both devices won't be exactly equal, so measuring the pin voltage (or just trying it out) is the only real way of knowing if it will work with a particular TV or not. I did mention above that it doesn't work on all TV's, however, but I found it worked on the majority of them, mainly due to the fact that early SCART TV's were built with discrete components and anything even slightly above 0.4V was considered to be a 1. Later TV's had more integration or even a "SCART chip" and were much more fussy about the voltage levels. So the older the TV, the more likely it is to work.

Bryce.

[Benedikt]

Very nicely calculated.

It is flattering that you see it that way, but I did that right before going to bed and the entire calculation relies on my sleepy brain's very generous estimate that 75 + 220 is 395.

If we repeat the calculation with a little bit more precision, it takes us back to my original assumption that the resulting switching voltage is either narrowly within or narrowly out of spec:

Assuming that the Gate Array output is an ideal 5V source, we get up to 75Ω / (75Ω + 220Ω) * 5V = 1.27V at the voltage divider.
Assuming that the Gate Array output is an ideal 3.6V source, we get up to 75Ω / (75Ω + 220Ω) * 3.6V = 0.92V at the voltage divider.

The bit that's missing is the capacitor, which is extremely difficult to calculate, but it's buffering will have an additional positive effect on the RMS value of the PWM signal.

My little calculation idealized the filter capacitor to a linear voltage averaging device.

But, thinking about it with a more awake brain, the average voltage is not actually what we care about.
We only care about the lowest voltage and that happens to be the instantaneous voltage at the end of the 1ms VSYNC pulse.
Calculating the voltage across a capacitor after discharging it from the full voltage divider voltage through a (75Ω || 220Ω) resistor for 1ms is actually quite simple:

U(1ms) = U0 * e^(-1ms / (1 / (1 / 75Ω + 1 / 220Ω) * 100µF)) = U0 * 0.8363

The calculation ignores the impact of HSYNC spikes during VSYNC as negligible.
U0 is the voltage across the 75Ω resistor of the (75Ω, 220Ω) voltage divider without load. See above.

So with the U0 for 5V we get no less than 1.06V, assuming that the capacitor was fully charged, whereas with the U0 for 3.6V we get 0.77V, which is problematically low.
Realistically, even the 1.06V are too low, because the Gate Array output is clearly not an ideal 5V source.

But there is one variable that we have full control over, namely the capacitor.

For C = 220µF we get
U(1ms) = U0 * e^(-1ms / (1 / (1 / 75Ω + 1 / 220Ω) * 220µF)) = U0 * 0.9219

And for C = 470µF we get
U(1ms) = U0 * e^(-1ms / (1 / (1 / 75Ω + 1 / 220Ω) * 470µF)) = U0 * 0,9627

If we now want to know what kind of voltage the Gate Array has to output in order to consistently keep the RGB blanking input above 1V, we can rearrange everything as follows:

Umin(C) = 1V * 1 / (e^(-1ms / (1 / (1 / 75Ω + 1 / 220Ω) * C)) * 75Ω / (75Ω + 220Ω))

And we get
Umin(100µF) = 4.70V
Umin(220µF) = 4.27V
Umin(470µF) = 4.09V
Umin(1000µF) = 4.00V

But keep in mind that we still assume that the Gate Array output is an ideal voltage source.
Nevertheless, it is safe to say that your type of SCART adapter will work more reliably with a bigger capacitor.

Now I just have to find the time to get out the oscilloscope and measure what the Gate Array actually outputs.