Television, 1947 style- 12
More tinkering over the last week- the PSU unit is now complete and up and running, with the two U31 rectifiers doing an excellent job at bringing the set up gradually. Having discovered that the U31 has a 25v heater, and subsequently poking around on Ebay to look for a dedicated 50v transformer to supply the two in series (necessary in case the heater of one goes o/c in service, to prevent the other from being massively overloaded) a faint bell rang, and I realised that my previous bumper box of cheap transformers had a 48v unit that I'd put aside as 'probably not useful'. After a bit of a rummage, I found it and wired it in, and it's doing the job splendidly, although a little close to its full rating. I've had the set running today while I carried out a few tests, and although it's getting warm, it's not getting hot, so I think it'll be OK.
I've also found the PSU was delivering a maximum of 450v at mains input- previously I'd been slightly under-running it. As a result, I've fitted the 1.5K 20W resistor included in the original circuit, to give a lower HT for the timebase valves, the 450v being reserved now for the frame and line output valves, as originally designed. It's now running reliably enough for me to give it full mains voltage immediately, rather than winding it up slowly- and apart from my stab-in-the-dark line output transformer (of which, more anon) there are no signs of any components being under stress. This is good news.
Today's experiments have involved testing the VF (video frequency) input for potentially tricky voltages; attempting frame synchronisation, and testing the outputs against a 9" picture tube.
First was testing the VF input tails; these go across the first frid of the frame synch blocking valve and -HT, with a 1M resistor coupling the two. (I'd forgotten about that resistor until now- I might have to look into altering that.) The potential difference between them is less than 0.1v, which is fine. However, both sides of the feed are at 300v with respect to earth, which is less helpful. This means that if I plugged the media player straight into it, the media player would be at elevated potential, which it might not like very much. It would probably be wise to couple it indirectly via a transformer, if I can do that without appreciable signal loss. It's not urgent yet, but it does more or less rule out testing for synch locking. I did rig up a basic test with a transformer, but I couldn't get any locking- possibly because of the transformer, and possibly because the line timebase isn't fast enough. Job for another day.
What I did do was identify the frame and line scan coils on the CCTV monitor and hook them up to the valve timebase. Vertical (frame) scan is slightly patchy:
Note the variations in intensity- this is the result of that oddly-shaped wave from the last post.
This is the first sight of both timebases driving the tube together- what's known as a 'raster'. This requires a signal to be fed to the cathode of the tube- I don't have a frequency generator, so I hooked up the media player to the back of the monitor, displaying a blank image to give me a bright spot. The image won't come through, as it's not being fed to the timebases (see above), only to the tube. The EHT for the tube is still being supplied by the circuit associated with the monitor, not my creation. One step at a time!
The raster in the above photo is pretty grim. It's deficient in both height and width, retrace lines are visible, nothing's terribly even, and there's nowhere near enough scan lines for 405, let alone my goal of 625 lines. This is a result of the line timebase running far too slowly. (it should be 15,625 c/s.)
After considerable tweaking (both with the timebases and my camera!) I got this, which is an improvement, just. I now have full height, which is good going- in the original design, the frame output fed the scan coils directly, which would probably have blown the coils on this tube up. I fitted an old 8v mains transformer as a wild guess, to step down the output voltage, and it's now giving me full height with a little wiggle room on the height control on either side. I just hope it's not the transformer 'ringing' spoiling the sawtooth form of the trace. There's more width in this photo, and a slightly cleaner scan, but still only about one fourth of the full scan width. This is a result of the line output transformer, another stab in the dark, which simply isn't putting out enough voltage for the full scan. The previous photo was using half the secondary of the transformer; the above photo was achieved by using the full secondary winding, so a doubling of output voltage. I've just purchased a small transformer on Ebay with a selection of ratios that I hope will enable me to fix this problem and fill the screen. Note also, where in the previous photo there was a lack of parallellism in the raster, the above is much better in this respect. I'm not sure why, but I suspect this is also a product of an ill-matched transformer. I've also discovered that what i thought was a ferrite core is, in fact, a laminated one potted in some form of adhesive, which would explain why it was struggling as I raised the output frequency. At 15,625 c/s the core needs to be ferrite, and I'm hoping that a high-frequency core will lead to less noise from the transformer and a cleaner scan. So, there's not a lot more I can do at the moment, but it looks like we're well on the way. I must try and find out whether the laminated core was playing a part in keeping the line timebase speed down- certainly there was a point in the frequency increase when the tone stopped being 'clean', and became rather nasty, accompanied by a breakdown in the scan on the screen. We shall see.
I've also found the PSU was delivering a maximum of 450v at mains input- previously I'd been slightly under-running it. As a result, I've fitted the 1.5K 20W resistor included in the original circuit, to give a lower HT for the timebase valves, the 450v being reserved now for the frame and line output valves, as originally designed. It's now running reliably enough for me to give it full mains voltage immediately, rather than winding it up slowly- and apart from my stab-in-the-dark line output transformer (of which, more anon) there are no signs of any components being under stress. This is good news.
Today's experiments have involved testing the VF (video frequency) input for potentially tricky voltages; attempting frame synchronisation, and testing the outputs against a 9" picture tube.
First was testing the VF input tails; these go across the first frid of the frame synch blocking valve and -HT, with a 1M resistor coupling the two. (I'd forgotten about that resistor until now- I might have to look into altering that.) The potential difference between them is less than 0.1v, which is fine. However, both sides of the feed are at 300v with respect to earth, which is less helpful. This means that if I plugged the media player straight into it, the media player would be at elevated potential, which it might not like very much. It would probably be wise to couple it indirectly via a transformer, if I can do that without appreciable signal loss. It's not urgent yet, but it does more or less rule out testing for synch locking. I did rig up a basic test with a transformer, but I couldn't get any locking- possibly because of the transformer, and possibly because the line timebase isn't fast enough. Job for another day.
What I did do was identify the frame and line scan coils on the CCTV monitor and hook them up to the valve timebase. Vertical (frame) scan is slightly patchy:
Note the variations in intensity- this is the result of that oddly-shaped wave from the last post.
This is the first sight of both timebases driving the tube together- what's known as a 'raster'. This requires a signal to be fed to the cathode of the tube- I don't have a frequency generator, so I hooked up the media player to the back of the monitor, displaying a blank image to give me a bright spot. The image won't come through, as it's not being fed to the timebases (see above), only to the tube. The EHT for the tube is still being supplied by the circuit associated with the monitor, not my creation. One step at a time!
The raster in the above photo is pretty grim. It's deficient in both height and width, retrace lines are visible, nothing's terribly even, and there's nowhere near enough scan lines for 405, let alone my goal of 625 lines. This is a result of the line timebase running far too slowly. (it should be 15,625 c/s.)
After considerable tweaking (both with the timebases and my camera!) I got this, which is an improvement, just. I now have full height, which is good going- in the original design, the frame output fed the scan coils directly, which would probably have blown the coils on this tube up. I fitted an old 8v mains transformer as a wild guess, to step down the output voltage, and it's now giving me full height with a little wiggle room on the height control on either side. I just hope it's not the transformer 'ringing' spoiling the sawtooth form of the trace. There's more width in this photo, and a slightly cleaner scan, but still only about one fourth of the full scan width. This is a result of the line output transformer, another stab in the dark, which simply isn't putting out enough voltage for the full scan. The previous photo was using half the secondary of the transformer; the above photo was achieved by using the full secondary winding, so a doubling of output voltage. I've just purchased a small transformer on Ebay with a selection of ratios that I hope will enable me to fix this problem and fill the screen. Note also, where in the previous photo there was a lack of parallellism in the raster, the above is much better in this respect. I'm not sure why, but I suspect this is also a product of an ill-matched transformer. I've also discovered that what i thought was a ferrite core is, in fact, a laminated one potted in some form of adhesive, which would explain why it was struggling as I raised the output frequency. At 15,625 c/s the core needs to be ferrite, and I'm hoping that a high-frequency core will lead to less noise from the transformer and a cleaner scan. So, there's not a lot more I can do at the moment, but it looks like we're well on the way. I must try and find out whether the laminated core was playing a part in keeping the line timebase speed down- certainly there was a point in the frequency increase when the tone stopped being 'clean', and became rather nasty, accompanied by a breakdown in the scan on the screen. We shall see.