Revisiting an old topic - Adding DC Restoration to a 1950s TV
 

Per very old posts, I added the attached circuit to my CMC TV, a rather standard 17" B&W table top set from 1958 or 59.

It evens out the blacks, but when brightness is high, there is now sometimes a bend in the middle of the picture. It appears more with dynamic images than with static ones. I wonder if this would be improved by reducing the value of the 0.1uF capacitor, or if maybe some sets are unsuitable for this mod.

Two problems about adding DC restoration to a TV.

1: EHT regulation matters more with true black level. Change in mean brightness of the pcture may give unacceptable change in size. This sort of effect might explain your problem.

2: If the set has mean level AGC (common with +ve modulation, probably less so with -ve) then the DC restoration won't be so effective. Partly because contrast will vary with average picture level, partly because sync amplitude will vary with APL. Since a simple DC restorer stabilises sync tip, brightness will still vary with APL.

To clarify how it behaves: almost like losing horizontal lock a bit when the contrast or brightness are turned up high. The picture bends and pulls toward the right, and then straightens out when either the scene changes, or the contrast or brightness are manually reduced. The black level is better with this modification.

That R32 150k you crossed out, what happens if you put that 150k in series with the top leg of the diode? Without that there you're really loading down the signal when the brightness control gets to the extreme ends of it's travel.

I used it but didn't show it, but I did this with a 100k resistor. I tried 150k, but it resulted in inadequate brightness range. With no resistor at all, it ends up being way too bright.

You are not loading down the video when the brightness control is at the end, because the diode only conducts on sync.

However, in the original circuit, the average cathode video voltage is between 0 and 120, but with the diode, the sync peaks are between 0 and 120, and the average is lower (making a brighter picture). You may need to look for a higher voltage for the positive end of the brightness control and put a resistor between the other end of the contol and ground, to get the syncs to be cutoff and normal video to be normal.

Have you experimented with smaller cap values? Could be the RC constant with 150k isn't allowing the .1uf to fully charge. I'm just kind of guessing at what's going on there. I'd like to see the rest of the schematic. Is that in Sams?

Yes but the sync could be loading down, no?

It is not in Sams, but I uploaded it to dropbox just now, the link is below:

https://www.dropbox.com/sh/bm7e4on8q...tAU050hPa?dl=0

I am not sure if it is loading down sync or not, but it does look like there is some effect. I agree, I am thinking maybe to try say a 0,022 capacitor and see how it looks. I also need to change the resistors to give it a bit more brightness range.

Have you observed the signal with a scope? I'm sure old tv nut gave you the correct answer. It's one of those circuits that's almost too simple to not have trade-offs.

I just noticed this - will review the schematic and waveform to make sure I understand

That noise immunity control in series with the sync is interesting. What effect does that have?

Fully counterclockwise, it has a switch. When switched off, the picture is grainy and seems weak. When turned on, the picture becomes progressively overloaded until when it is fully clockwise, it has the same effect as high brightness.

I really don't know what is is for though.

Ok so after looking at it closely, here is my understanding of what is going on. I could well be wrong in some detail:

Looking at the service manual, the AC signal at the plate of the video output tube is shown as being 16V peak to peak, with the big peaks being on the sync pulse. This makes sense from the perspective that during sync, the picture should be blanked. So assuming that there is no DC restore circuit, it looks like at maximum brightness, the signal at the cathode of the CRT ranges between the 0V and +120V, plus the 16V of the video output tube. So at maximum brightness, the AC value of the signal at the CRT cathode is between 0V and 16V, and at minimum value of brightness, the signal ranges from 120V to 136V.

The DC restore function appears to work like this: During the sync pulse, the highest voltage is present, which charges the capacitor. As soon as the sync pulse ends, the voltage at the capacitor is higher than the voltage on the cathode of the CRT, so the diode switches off.

Now, why would elevating the positive end of the brightness control help things? It is usually operated closer to the middle position, or higher. Even if this end were elevated more, how would it affect the DC voltage at the junction of the capacitor and the volume control wiper?

I seriously doubt that the black to white swing of the video signal on the cathode is only 16 volts peak-to-peak on a normal picture. Could you post the waveform from the manual? I suspect it's for an odd signal that's mostly black. Another clue is that the range of the brightness control is 120 volts. If the usual video was only 16 v p-p, this would mean the brightness control has terrible excess range and would be very touchy to adjust.

Suppose that the video amplitude is a more reasonable 60 V p-p (just a wild stab) and the picture is average (not all white or all black). Then without the DC restorer, the cathode voltage swing at min brightness is from 150 to 90 (120 +/- 30). At max brightness it swings +30 to -30.

Add the DC restorer, and the swing at min brightness is from 120 to 60; at max brightness it's zero to -60.

The offset of the effective bias will be on average half the peak-to-peak (depending on scene content), so to darken the picture back to normal, you need to add about half the normal peak-to-peak voltage to the range of the brightness control.

Edit - for my guessed case, this means the brightness pot needs a range of 150 v to 30 v.

Here's the image from the manual. It struck me as odd too, it seems quite low, especially compared with range of brightness control as you say. Also, I would expect a pentode to have more gain. I will have to measure it to be sure. So you think 16V is a mistake in the manual?

Note to self: stop putting the chassis back in the cabinet until all is settled!

I think I've reached the limit of my knowledge, especially without seeing the whole circuit.

Meanwhile,
1) How touchy is the brightness control? Does it have alot of excess range? This will give you an idea of how much the video signal swings compared to the brightness control voltage.
2) Can you cut off the blacks with the brightness control? If so, no adjustment to its voltage range is required, and you may just have to back off contrast (video drive p-p) to make sure bright pictures don't overload things.

0.5) Thanks for paying attention to this thread, I appreciate the advice and interest. If you are interested, I did upload the circuit to dropbox, here https://www.dropbox.com/sh/bm7e4on8q...tAU050hPa?dl=0

1) Not really all that touchy, but only the top of the range shows a good picture. The picture tube is not super strong though, so this is not a big surprise.

2) With the brightness control full counterclockwise, there is no image visible on the screen at all.

12BY7A data sheet says transconductance is 11000 micromhos, or 11 ma per volt. Load is 5.6 kohms, so 5.6 volts per ma. Service data says input is 3v p-p.

3x11x5.6 = 185 v p-p. This could be different in this TV because minimum cathode resistance is 33 ohms (I may bother with actually calculating everything tomorrow). Anyway, 160 v p-p at max contrast does not seem unreasonable. 16 v p-p seems indeed to be a misprint.

That does make a whole lot more sense than 16V. I did check the 12BY7A just for fun and it checks good.

Will the b+ in this series string set even allow those kind of voltage swings? I looked at a similar US made set and they show 30vpp from the video out. :scratch2:

The 33 ohm cathode resistor reduces the gain such that the max possible plate swing is 136 volts. Knock off a bit for the plate resistance, and it's maybe 128 volts.

BUT all this was quick small-signal analysis based on spec sheet data at a plate voltage of 250 volts. You are right - The B+ of 130 v in this set means that the analysis should be done graphically and the gain will be considerably less. I'll attempt this later for fun, but it may be that the answer is 30 or even 16!

That's what I get for going back-of-the-envelope as I'm falling asleep.

Ok, here are the plate characteristics with the load line. You can see that most of the tube's capability at higher plate voltage is wasted.

Transconductance varies around 4000 to 7000 micromhos; call it 5000. Then the output for 3 volt input can vary from 30 to 72 volts p-p depending on the contrast setting.

Linear approximation:
Cathode current swings 5 ma per volt Vg-k.

For 1 v Vg-k swing, 33 ohm cathode resistor swings 5x.033 = .165 V; for 3v input, V g-k swing is 3x1/1.165 = 2.575; 2.575x5x5.6=72.

For 1 v Vg-k swing, 363 ohm cathode resistor swings 5x.363 = 1.815 v; for 3v input, Vg-k swings 3x1/2.815 = 1.07; 1.07x5x5.6 = 29.8

Could it be as low as 16 vpp at minimum? Probably.

Edit: fixed a couple of typos above.

Note: In circuits lab I PO'd the TA when I asked how much the tube characteristics could vary and he obviously had no more idea than I did. (We later became good friends and colleagues when I went to work at Zenith.)

Thanks for doing all this. I suppose it is not a big surprise that the tube is not used to its full potential in a transformerless TV with such low B+. Out of curiosity, during what time period did you work at Zenith?

Motorola '66-'75, Zenith '75-2015.

This sounds like a great career - almost 50 years in the US TV industry! I can't imagine how much change occurred during that time frame. What was the last US designed series of television which Zenith produced?

No doubt you gained some electronics skills prior to '66. Did you do a lot of homebrewing, ham stuff and the like?

Actually, I can't answer off the top of my head. I'm sure someone else here could be more exact. I was in Advanced Development, so stuff I worked on was 2 years or so ahead of production, and sometimes didn't pan out in the end. To paraphrase Edison, the time wasn't wasted, we just found out what not to do.

Actually, not a lot. I was fascinated by color TV technology from the first time I saw it at age ten, and I lucked out in getting to work on related stuff, so I felt I was being paid to play.

So getting back to your sync disturbance issue - your sync is being pulled off the bottom of L7 with very little isolation from your DC restoration diode. So it's no surprise that there is some push-back loading effect on the sync when the diode conducts heavily. I'd be curious to read the thread that this method your using originated from.

https://antiqueradios.com/forums/vie...p?f=3&t=194179

http://www.videokarma.org/showthread...=248278&page=2

Max,

I did some more poking around and ran across this old thread that you started and possibly you'd forgotten about it. Andy pretty much identified the problem with this post I've quoted below. Andy confirms the sync stability remark with another post at the very end. This Marconi set you're working with definitely fits the category of having a cheap sync design where the sync is taken from after the video amp. Another way of doing this would be grid-leak DC restoration which takes place in the grid circuit of the video amp. That may give you the needed isolation. I'll see what I can dig up on that method.

http://mail.videokarma.org/showthread.php?t=250452

Also, if you haven't already, try experimenting with making your .1uf cap value smaller or eliminating it. Possibly you can find a workable compromise.

Kevin,

I do remember that now! I was trying to improve my little 8" RCA, it is incredible how fast time passes, I can't believe it was already 8 years ago.

Thank you for looking into this for me, I will pull the chassis today and try some smaller capacitors. It is looking like I am going to.have a lot of time over the next few months. The Marconi was sitting half finished under the desk I'm using to work at home, so when I was preparing to work at home I pulled it out. I'm using it to play the news, and concert DVDs while I'm passing the time. It has a better picture, and a substantially better speaker then my Predicta does, and this one little improvement would be nice.

I also.got a bunch of tubes for it over the weekend from VK member Gregb,

I noticed the resurrection of this thread and would like to add my own comments. The RCA 630TS employed 1/2 a 6AL5 diode as DC restorer in the circuit depicted below). The RCA sets from chassis KCS28 thru and up to I think KCS47 utilized DC coupling and no DC restorer diode. From about 1953 RCA abandoned DC restoration.

In every case, RCA did not employ 100% DC clamping and compromised with only partial clamping. You can examine the degree of how well black level clamping is achieved by watching active video cutting between dark to bright scenes while examining the vertical blanking bar on the screen (by misadjusting the vertical hold control so the bar is central on the screen). The bar should remain the same shade of grey or black when cutting between scenes.

The later sets without any DC clamping the bar changes with the scene cuts. The earlier RCA sets the bar shade remaiens "reasonably" constant in shade. This doesn't means that there is no noticeable shift in the bar's brightness.

I think the RCA engineers compromised a little and did allow some AC component. Perhaps this was because of normal temperature drift of components which would result in having to constantly adjust the brightness control. Nevertheless, look at the 630 TS and compare it with your original modification. It looks as if R148 was added to isolate the diode capacitance from the video path. Also note R149 the 1Meg ohm resistor across the clamping diode which is necessary as a DC return.

Also note that in the earlier sets, the clamping diode is across the video path to the CRT grid input. Later sets applied the signal to the cathode. The grid circuit does not draw current whereas the beam current passes through the brightness control and thru and around the clamping circuit. A fundamental error in your earlier circuit is that there is no DC path to allow beam current to flow!! So you are going to have to place a lower resistance across the diode which will compromise the clamping. In other words, not to sound to pessimistic, I do not think it will be a simple task getting the cathode driven circuit to work.

Also note that C143 0.25ufd is charged thru R150 which effectively further dilutes the clamping effect.

The bottom line is that it is not a perfect world and engineering is all about compromises. Especially when dealing with analog circuits! I would suggest you carefully think thru this as there are a few gotchas that need to be addressed.

Thanks for posting, this helps a lot! I have been using the TV all week, and what I've noticed, is that high contrast scenes tend to lose cause the TV to lose lock horizontally, and there are also rare scenes which affect vertical lock.

This TV actually wouldn't lose horizontal lock through the entire range of the horizontal hold control previous to being modified.

Maybe it is best to just put it back as it was? I suppose if it was good enough in 1958, it's probably still good enough!

I also wondered about where will the beam current flow, and realized that it flows to ground through the diode when the diode is "on", and when the diode is "off" it flows through the video output tube. There is really no other path. This must be what is going on, because the circuit does work, it just doesn't work well.

Did you get a chance to try any smaller cap values? And if so was there any difference on the hold interaction? I looked at grid leak bias restoration at the video output grid, but that requires DC coupling from the video output to the CRT cathode. Could be done, but it's a lot of re-engineering.

Not yet, I haven't had a chance to pull the TV chassis out of the case again yet.

Let me ask a dumb question. Is the DC signal we are attempting to restore present at the output of the video detector?

Yes. The DC component is broadcast. One of the consequences is that the RF power at sync tips or blanking or black level does not change due to variations in the scene content. Therefore (assuming the TV's AGC circuit is working on sync tips) the output of the video detector is similarly stable.

However, carrying this all the way to the CRT by using DC coupling all the way is not practical in tube sets (too much drift over time/temperature, plus changes every time you replace a tube). It is also not practical in solid state sets even if the transistors do not vary, due to component tolerances. Every stage would need the bias circuits replaced with adjustable potentiometers.

The only reasonable way to stabilize the video DC level is to do it close to the CRT, (or measure it at the CRT and use feedback to adjust it at an earlier point).

The ultimate solid state CRT TVs used feedback to control the CRT bias individually for red, green, and blue, thus achieving not only DC restoration but automatic CRT tracking. The video IC restored DC in the low-level luminance, then added a low brightness pulse at the top of the raster just after retrace (where you can't see it) to measure the CRT beam current near black.

Why not measure zero current, exactly at black, to set cutoff? Because a CRT has a power law current vs. voltage characteristic, so at zero current, the incremental current gain is also zero, which means there is no feedback signal.

By the way, if the DC component was not broadcast, it would of course not be present at the detector, and the whole exercise would be somewhat futile (although it could have some effect by clamping the darkest thing in the scene as black).

Iconoscope camera tubes had inherent AC coupling. The DC level had to be made up by clamping the darkest parts of the scene to not go below black, plus there was a "shader" (a live person) adjusting it along with the spurious shading across the image that the iconoscope was prone to. Someone who does camera electrical alignment is still called a shader today, although they are not doing it all real-time during use.