I think i did it.
Ive been sitting on the DC restoration tomfoolery and i think i figured it out:
Any AC signal contains an average DC level, just like our video signal - which swings upwards from 0V (sync). Passing this signal through a capacitor makes it AC-coupled, meaning the DC component of the signal, the one containing all the information regarding the black/blank and peak white levels, vanishes into thin air. The amplifier doesn’t know what a black/blank is, nor does it care, this makes brightness dependent on picture content, as the waveform is now free to drift around based on the average voltage of the waveform, like a boat on waves, with the black/blank and sync pulses turning into an average voltage on an AC signal instead of fixed points, which leads to varying brightness levels and a washed out picture.
We solve this issue by clamping the waveform to a periodically repeating signal, like the sync pulses (remember, they're no longer at 0V since they now swing around a Q-point), thus allowing the signal levels containing the picture information to extend unidirectionally from a fixed reference point.
There are various ways to implement DC restoration in a video amplifier, options range from three components to entire circuits. I believe the best one for my use-case is by rectification in the picture tube input circuit.
This topology allows us to implement brightness control alongside the fixed reference point for the waveform. Another reason as to why I opted for this method over a different one is because the amplifier drives the cathode instead of a control grid, which is now wired to ground. In other words, if our CRT's cut-off voltage were to be -50V, we would need the control grid to be 50V more negative than our cathode in order for the screen to turn black.
As previously mentioned, the signal our video amplifier outputs is inverted, meaning the composite sync is now the most positive part of our signal, with the peak white level taking the most negative place – this is crucial for the clamp's functionality:
Assuming a bias and cut-off voltage of 50V, and a signal amplitude of 20Vpp (example values):
a) A bias voltage is established across R3 via the potentiometer. When the bias is set to be several times higher than the average signal DC, the total voltage variation from peak white to sync becomes lower, this serves to keep the clamp stable. Noise and ripple on a 10Vpp signal clamped to ground cause shifts in the black level to appear a lot more noticeable compared to the same signal clamped to, say, 100V.
b) The diode, connected with the anode to the C-R junction and the cathode to the wiper of our potentiometer, does not conduct, as there is no voltage difference between its two terminals, both sitting at 50V.
c) The highest positive peak (sync) occurs and sets the anode at a higher voltage compared to the cathode, the diode is now forward-biased, allowing the capacitor to accumulate a charge proportional to the level of the sync tip (plus the diode's voltage drop), which clamps the sync tip to 50V. Because this sits at exactly the cut-off voltage, the beam is off – there is no picture.
d) The capacitor is sitting at a higher potential compared to the bias voltage, causing it to (slowly) discharge through the resistor based on their time constant, getting topped off at every positive peak (sync). How fast the capacitor discharges is dependent on the average DC level of the waveform – If the picture is white, the signal spends most of the active time period at the negative peak, say, 30V, leading to a low average voltage, and thus a lower voltage difference across the resistor. In the case of a dark scene, the signal spends most of the active time period closer to the positive peak, say, 40V, causing a higher average voltage and, in turn, a higher voltage difference across the resistor.
e) After the sync tip ends, the signal can only extend in one direction from here, which is downwards to the negative peak (maximum white level) – there is now a picture with a stable black level.
The brightness control works by shifting the point where the signal gets clamped around the cut-off voltage. If our bias were to be set at 40V instead of 50V, the darkest parts of the image would be a brighter black (gray, up to white if you're determined enough), pushing the peak white levels further away from the cutoff point, increasing the picture's brightness. Were we to set the bias at 60V instead, our positive peak (sync) would sit above cut-off, dragging the peak white level closer to cut-off, and thus lowering the picture's brightness.
The time constant needs to be somewhere within a handful of frames, like 50ms. As for the diode, it needs nothing fancy, its only requirment being a minimum leakage current – the 1N4148 will do.
Im not 100% certain on the 10Vpp signal riding on 100V part but otherwise it makes perfect sense to me.
Ive also added a gain/contrast control pot of 220 ohms. The value was picked to be as low as humanly possible without loading the source too much (source impedance of probably 75 ohms and the pot have a current draw of 3.3mA). But now theres a chain of RC filters. We have low-pass filter A formed by the source impedance and the thevenin resistance of the contrast control potentiometer with the input capacitance (cutoff at 75MHz), followed by the high-pass filter comprised of the grid leak resistor and input coupling capacitor (cutoff at 3Hz), and finally, low-pass filter B made up by the source impedance and thevenin resistance and grid leak resistor all added together paired with the input capacitance (cutoff at 57MHz). Side note, thevenin resistance was calculated with one of the parallel resistances added together with the source impedance.
Is this my fault or does this kind of just happen with video amplifiers? Am i supposed to try and calculate what this cluster of RC filters does to the amplitude and phase response of the signal or do i just wing it?
Also, i calculated the input capacitance to be the tube's input C (13.8pF), the transfer C multiplied by the miller effect (0.05 x 251 = 12.55pF, i used the theoretical maximum gain given by gm x RL) and stray/wiring (say 2pF) than rounded up to 30pF.
Edit: accidentally left in the 1k value instead of 220 ohms for the contrast pot in the schematic
Home
Help
Search
Login
Register
