Bringing vector graphics displays back to life

"Boudewijn Dijkstra" wrote in news:blcki1$pmi$ snipped-for-privacy@azure.qinip.net:

Perhaps you could find an old video game board, such as Omega Race or Asteroids. These boards used low end CPU's (6800?) to run those games, so if they could manage it, an AVR or a modern 8051 derivative should be able to no sweat.

The biggest issue is one of performance. Unless you scope is a storage scope (rather unlikely), you will need to continually refresh the image. The speed of the controller determines how many vectors you can draw without flicker. That will also be influenced by the persistence of the phosphor.

For any reasonable performance, I think you would want to build a vector generator, either with analog circuits or high-speed digital. In the past they were mostly done with analog generators which, I think, generated a smooth ramp from the starting to ending voltage on both X and Y coordinates.

If you just want to demonstrate the principle, you could do that with simple software, but you would be limited in the number of vectors. You could drive D/A converters with a Bresneham algorithm. Let's say that you can generate a point every 10 us (probably aggressive for 8051) and that your scope resolution is 256 x 256. Say you average vector is 50 pixels long. That would be 500 us. for the vector. If you refresh 30 times/second, you could draw about 66 vectors. That's why vector-generating hardware is used.

Thad

far

basicly

can

most

greatly

-->Neil

far

basicly

can

most

greatly

I did Tempest signal analysis for the USAF during the mid 80's when Tempest was considered a threat. I had an oscilloscope that had X, Y, and Z-axis inputs. The Z-axis input was sometimes labeled the "intensity" input. Anyhow, I was able to display a television signal on an oscilloscope by playing the demodulated television signal into the Z-axis input on the scope. The X input was fed with the horizontal clock signal, and the Y input on the scope was fed with the veritcal clock signal.

If I remember correctly, the resulting picture was upside down, but we were using this test to read monitors from local office buildings, not to watch television, but the same principle applied. Perhaps you could do the same thing? Just display some graphics on a monitor, then put an FM receiver near the monitor, tune it a little and then feed the demodulated output into the scope's Z-axis input. You would need to generate the X and Y signal inputs somehow. I think that either the X or the Y signal was a sawtooth waveform. Sorry I'm not being too helpful, but I think that this approach might be less expensive, and I know that it will work.

The normal vector generation technique does NOT require DACs that fast. Instead, the hardware only writes to the DACs for the line endpoints. The analog circuitry charges (or discharges) a capacitor to ramp the deflection voltage.

Go to the library and find the November 1976 issue of Byte. Read Steve Ciarcia's article "Make Your Next Peripheral a Real Eye Opener" And maybe some other articles in that issue; I think at least one other article covered vector generation.

There's errata to the article in the March 1977 issue. I haven't read that, so I'm not sure exactly how relevant/important it is.

A few years later Byte ran a construction article for a complete standalone intelligent vector graphics system, based on Ciarcia's vector design. I don't recall the title, author, or exactly when it appeared. Early 1980s, I think.

For fun, look up "How to Implement Space Wars (or Using Your Oscilliscope as a Telescope)" by Dave Kruglinski in the October 1977 issue.

I think a 68000 is overkill. I've seen the kind of thing you describe implemented in discrete logic with EPROMs, an address counter, and R-2R ladders as the DACs. The R-2Rs are on the EPROM data buses, the address counter is on their address buses, there is a single timebase (a 555 rigged as astable multivibrator) clocking the address counter. The X-Y inputs of the scope are connected to the R-2R outputs. The intensity control isn't necessary (for a simple circuit); you just trace out the shape you want repeatedly (as long as you can live with everything being joined up).

In fact, the circuit I saw is part of a funny story: It was built by two physics students in first year at the University of Melbourne (Australia) in 1990. They were radio hams, and generally electronics mavens, and first year "introduction to the oscilloscope" physics lab was a waste of time for them. So they built this circuit and concealed it in a hollowed-out pineapple. They ran the X-Y and GND wires up the undersides of the pineapple leaves, and reassembled the fruit. They clipped the scope probes to the leaves (carefully concealing the wires underneath) and called the teacher over to say "Look sir, this pineapple is doing strange things".

I forget exactly what the pineapple was "thinking" but I recall the first screen had a vector image of a pineapple with a thought bubble coming out of it. The next few had text, something like "THEY DO NOT SUSPECT", "WE WILL SOON TAKE OVER", etc.

Note that their circuit was a *TAD* more complex than I've described. Each "frame" was refreshed at the full timebase rate from the LS bits of the counter, then there was a gap in the counter->address line mapping, so that each frame would stay onscreen for [n] clocks.

Asteroids=6502, Omega Race=Z80. Both had a digitial vector generator that generated the lines.

Check out

-->Neil

Is this what you talking about?

scroll down a little bit...

Seems like the guy pulled it off with a microcontroller and a few fast DACs. It's a project that's cool enought that I would try to duplicate it if I had the time.

-Jim

Cool! An another cool thing is that the tube looks very much like the one I recently bought. It was (is?) for sale as a 'decorative paperweight' from Conrad (dutch/german mailorder). Cost only a few (7?) euros and turned out to be a new CRT in box.

Pitty the page does not give more info on how the tube is controlled, any info on that anywhere?

Sounds extremely dangerous to the ignorant. What if that 'paperweight' falls off the table and implodes? It would be quite easy to lose your eyesight.

Never thought about that. And I personally would not leave it out of it's foam-packed box until I actually use it (as CRT), but someone could.

Just checked to be sure: It has not been made useless, so vacuum must still be intact. It is a 3", 7" long, tube. No labels 'danger' or 'fragile'.

Conrad is an electronics mail order company so it is not very likely that 'normal' people will order there. It was advertized in a special with old stock items, not in the normal catalog. Not sure if you can still order it though.

Why it was labled 'paperweight', I don't know. Thinking about it, the words 'paperweight' and 'fragile' do not go together very well.

"Neil Bradley" wrote in news: snipped-for-privacy@corp.supernews.com:

Ah, in that case if you can find one of these boards, you could scrounge that, maybe. Of course, finding the documentation could be interesting...

I've built one. His design does not need fast DACs.

Just get one of these.

At $239, I'll pass...

For small strokes, based on an xy location, one needs two integrators and mixers to combine the x, xdelta, and y, ydelta positions. You also want z axis (blanking) control. It is then easy to create a 7 segment display (which needs eight, one of which is alway blanked) which returns to the point of origin. I have done this to write numbers etc. on CRT displays. More complex patterns can be generated with no more hardware.

By mixing the ydelta into the x display, you can create the degree of slant required. One more resistor.

Operation is a matter of:

Select x,y in the main registers (d to a or any other way) launch a fixed pattern on the xdelta, ydelta lines. The x and y deltas may be plus, zero, or minus a standard voltage. Simultaneously launch a digital pattern on the z line, which describes the segments of the pattern unblanked.

Repeat as desired.

Having the stroke generators return to the point of origin is necessary to avoid accumulating errors in the integrators.

Yeha, that was admittedly a guess. That would be my solution, being a software puke and all.

On closer inspection of his equipment, I noticed an abundance of small IC's (I assume op-amps) and trim pots labeled 'SIN' 'COS' 'XPOS' 'YPOS'. Maybe alot of the shape generation is analog, while postioning of shapes is handled by a small micro?

I'd like to buy a kit, but the price is steep, though I defintily think it's a fair price considering the work put into designing something like that.

-Jim

"Bob F." schreef in bericht news:3f79f720$0$43852$ snipped-for-privacy@news.twtelecom.net...

probably

were

display.

Tempest

were

into

I think you missed the point. My intention was to demonstrate the principles of the operation of a vector graphics display, not to use a scope as a raster display. Besides, your idea could be realized in under 8 hours, which is far too less for the assigment ("come up with your own 80 hour task, which must be approved by a teacher").

"Lewin A.R.W. Edwards" schreef in bericht news: snipped-for-privacy@posting.google.com...

or

could

I think a 68000 is not overkill if it has to dynamically build a vector list from character input which is coming from serial port. This was the next 'step' in my idea.