I have juste bought again my first computer on ebay... a Speccy 48k. Just had to replace TR4 and it worked.
I have made the composite video modification, simple (wire), with capacitor (100uF), with transistor (BC547), but my TV (Samsung UE40H6400) has difficulties to lock on the video signal:
- about once per second, it loses sync for about half a second
- after some minutes, everything becomes OK.
It does not depend on the type of the modification.
I have no other device with a composite input, so I can't tell if the TV is especially demanding. I guess that the fact that it works after a few minutes is due to the temperature of the components - but I don't know if in the TV or in the Seccy (I speculate it must be in Speccy, because some parts like the ULA get rather hot). I will have to try with my cold spray.
My oscilloscope shows me the following informations when displaying the start screen (c) Sinclair..., i.e. black characters on white (not bright) paper:
- the video signal looks globally good. Not much high frequency parasites on it
- the negative horizontal sync pulse (sync to black) is very deep compared with the white level (black to white): with the transistor mod, 0.5V (sync to black) vs. 0.4V (black to white)
- this is proportional to the inverted Y outpout of the ULA (sync 3.3V, black 2.2V, white 1.3V)
- adding a 47 ohm resistor in parallel to the TV input (which has 75 ohm) reduces the amplitde: sync to black 0.3V (the "official" value for composite video), but doesn't solve the sync problem
- the timing of the line sync pulse seems OK (including the color burst, the this one should be irrelevant for sync issues - by the way, the colors are OK), but the field (frame) blanking pulse lasts for about 250us, i.e. about the same as 4 lines and is only one negative pulse (the signal stays low the whole time). I thought the "official" duration would be 19 lines?
- there are color bursts (the same as at the beginning of each line) on the signal during field blanking (i.e. around sync level).
Another reason I see is the field sync pulse, which is much shorter than I thought it should be and include no pulses. But I don't think that a defective ULA would generate a good signal, only with this part of the signal wrong!
The last reason would be the color bursts during the field blanking sync pulse, could they be the reason? Are they normal?
- cooling the components on the Spectrum PCB has no effect
- the settings of the two variable resistors (VR1 and VR2) has an inflluence on sync. When the color is bad (i.e. big chroma signal on the composite signal with black/white display), the TV syncs BETTER!
Thanks and regards