Well perhaps this is a little more exciting than the above...............................
Since I started working on this analog computer project I've been pondering over what kind of X-Y oscilloscope would make the best display device. For example, something like this:
http://www.analogmuseum.org/english/examples/bouncing_ball/ just isn't as impressive as it should be when viewed on a tiny CRO tube screen with a 10 by 8 cm grid.
So this afternoon I started investigating the idea of converting an old TV set with magnetic deflection into an oscilloscope. An old little TV set which has been sitting in my cupboard for the last 11 years unused was sacrificed for the purpose. The pics below show how far I got. To my delight, my basic theoretical understanding of how to undertake the conversion proved entirely adequate and the conversion is really going to be a piece of cake.
I started by reverse engineering the RGB driver board (plugged into the rear of the CRT) and disabling the video. I reconfigured the gun driver amplifiers so that I could adjust the intensity of each color beam independently/manually. In the photo below I just have the green gun operating; the others are biased to cathode current cut-off. I then identified the connections to the separate horizontal and vertical deflection coils and disconnected them from the sets circuitry. So, from that point on, I am only utilizing the TVs motherboard to provided the operating voltages for CRT.
The vertical deflection coil is actually quite sensitive. My AWA G231 oscillator was easily able to drive the coil for full scale vertical deflection on the CRT at frequencies of 1kHz or so and less. To approximate a current drive in this initial test, I am simply driving the coil through a low value series resistor. So long as the value of the resistor is low compared to the inductive reactance of the deflection coil (at the operating frequency) current drive is adequately approximated and a linear(ish) deflection appears on the screen.
For driving the horizontal coil, however, I had to knock something up. I initially assessed the deflection sensitivity of the horizontal deflection coil by connecting it to my bench power supply in the CC mode. By dialing the dc current through the coil up and down I could therefore manually move the dot across the screen. A current of approximately 1A was required to shift the dot from the center of the screen to the edge (with no current through a deflection coil the dot returns to the center of the screen for the respective axis).
I then made a really simple and crappy (but adequate for "proof of concept") single-ended class A, current-output power amplifier stage capable of driving +/-1A peak into the horizontal deflection coil. This amplifier stage was build up around an MJL21193/MJL21194 power transistor pair screwed down onto an adequate heat sink. This amplifier is shown in the photographs. Being single-ended class A a bit more than 1A is constantly drawn from the supply (note the current reading on the bench supply providing the +/-15V supply rails). By driving this horizontal deflection coil amplifier with my crappy little Jaycar "pocket" signal generator I has able to get some nicely linear Lissajous figures up on the screen. yahoo!
The next step was to knock up a crappy triggered "timebase". For this I used a 555 wired as a one-shot monostable. It is triggered (via a transistor switch configured as a crossing detector) by the AWA G251 oscillator output signal, providing the vertical deflection. For a linear ramp, a current source is used for the 555 one-shot and the linear ramp voltage across the timing capacitor is buffered with a BC550C (c-grade for high gain) emitter follower and then applied to the H-coil driving I-out power output stage input via a pot to vary the horizontal sweep width on the screen.
This allowed me to get a conventional, stable oscilloscope-like display of a ~500Hz sine wave on the screen, as shown in one of the pictures below. Note the feint retrace line though as I wasn't bothered to implement retrace blanking at this early experimental stage.
Now that I know exactly what needs to be done to turn this old TV into a scope with 10-20 kHz bandwidth or so, I can begin with the electronics design proper. I am going to have to build class B I-out deflection coil driver amplifiers with several hundred volts of voltage compliance. I also intend to do away with the TV's original circuitry entirely, building my own PSU to deliver all of the necessary voltages for the CRT. For the gun drivers, I currently intend to make them digital - each gun either on or off in accordance with logic control signals with a common, overriding cut-off control input (for retrace blanking) and with a common brightness control that varies the "on" threshold of the thee guns in unison and thus the beam current and the resultant display intensity. A 7-position rotary switch will select the trace colour (3 bits actually gives 8 "colours" but one of them isn't very useful as it is black
).
What I also intend to implement is alternative logic RGB inputs. This will allow, via multiplexing, the display of multiple traces off of different colours. So the 7-position rotary switch will rather be an 8-position one with the eighth input selecting external logic signals for the trace colour control. For example, I will eventually be able to run Bernd Ulmann's "ball in a box" simulation (as linked to above) on my machine, but instead displayed on my large screen TV scope, with multiple bouncing balls with different coefficients on screen at once displayed in different colours!
Also, as a bonus, my parents (on the big LCD-screen craze) pensioned off today a perfectly fine CRT TV with a screen twice as big as the one shown, which is due for delivery here tomorrow morning; another to be sacrificed for TV-to-CRO conversion