I do not agree in this point. A capacitor wouldn't help to avoid this discharge. As you probably know how a capacitor works, a potential difference between both ends (one at RCA connector and one at ULA) would result in a current peak which goes through the capacitor in the ULA until capacitor has been loaded with the voltage difference. Under normal circumstances your items you should not have these problems. If you have a metal case, it has to be connected to earth or if you have a plastic case it should have a galvanic isolation due to some industrial standards. So if you have problems with a discharge caused from these components you should verify your items. Something must be wrong.RetroTechie wrote: Another reason why not is electrostatic discharge, especially when (dis)connecting big equipment with the Zeddy in between. Say you have a PC soundcard connected to ear input, and you plug in RCA connector to TV. The tip of the connector is normally the first to touch, so any voltage equalization between TV-ZX81-PC then goes through the ULA pin. By the time the outside of the RCA connector (ground) makes contact, the damage is done.
Well my description was not exactly, you are more exactly. A normal Video signal has 1 Volt pp - so the average current is less with a capacitor in comparison to a directly connection to the ULA. But (!) it is anyway too much. So 1 Volt would give peak currents of about 14 mA. Depending on screen content could be effective less average but with a empty screen (white background) it's more near to 14 mA. A R/C combination of 75R and 100uF has a T of 7.5 milliseconds.RetroTechie wrote: Wrong! The other end of the capacitor will go to voltage level it's loaded to. For example schematic of my old TV shows a plain 75 Ohm resistor to ground at its composite video input (SCART). So if ULA pin output averages 3V, capacitor would charge to 3V, other end would average near 0V (and thus little point in adding a diode if your goal was a 0V average), and the ULA pin would behave as if it were loaded with 75 Ohm to a 3V DC source. Supplying current when output >3V, sinking current when output <3V, but on average certainly less than with 75 Ohm to ground. Agree with the value though - 100 uF or more is good, with smaller values you might see some distortions in screens with big black (horizontal) bars & such.
This means, for a horizontal sync period it does not go to saturation and it's resistance is nearly zero in this calculation. As you know, the output of the ULA is much more (about 2.5 Vpp) - so could calculate up to 35 mA. Even if voltage breakdown with load the chip has more internal power dissipation. So I would be careful connection ULA pin only with a capacitor to a 75 Ohm load.
I didn't exactly try to measure it out. As I developed a video signal reconstruction circuit I found that it is important to have fast transitions form lo/hi and hi/lo for a crisp picture. It does make the transitions new and better than ULA and kills most of internal noise from the ULA. To have fast transitions you need high slew rates which contain a significant amount of 3rd, 5th even 7th harmonic of maximum pixel frequency (could be 3.25 MHz, exactly the cpu frequency).RetroTechie wrote: You'd think so, but when I tried this once with different capacitor types (like tantalum, which usually has much better high-frequency behavior than aluminium electrolytic), I couldn't find any visible differences. And as long as the pixel frequency gets through, damping higher frequencies may actually improve image quality.So I guess for this application cheapo electrolytic is good enough, and other factors (ULA, output circuit, power supply, video cable, TV) are more important for image quality.
So for a good signal/picture we are talking about 22 MHz or more which should pass without significant damping. Picture would get weak and you have a kind of antialiasing effect which does not look very well on modern TVs.
If you look at this picture, you can guess why big electrolytic capacitors in combination with 75 Ohm load are not best choice.
http://de.wikipedia.org/w/index.php?tit ... 0220081224
(either better, or worse)
Pictures help 
Standby alert 