[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Anybody with an oscilloscope as an Apple II display?
Charlie Springer wrote:
On Wed, 26 Mar 2008 16:58:09 -0700, Michael J. Mahon wrote
(in article
<bf9c4e5d-9c84-447d-aa73-f97f9db0769d@8g2000hse.googlegroups.com>):
On Mar 25, 9:51 am, heuser.mar...@freenet.de wrote:
Something on the line of this
here:http://www.youtube.com/watch?v=s1eNjUgaB-g
Maybe not as detailed but is it feasible at all?
You need dual D/A's to generate such a display. The speed with which
the outputs
can be changed--which soon becomes the speed with which the outputs
can be
computed, is what limits the complexity of the display.
Although it is possible (with appropriate analog post-processing) to
generate
a straight line segment by sending just the new end point (a vector
display),
in practice one would send several intermediate points so that simple
low-
pass filtering would produce the desired trajectory.
It's actually an older form of computer graphical output, since it
preceded
raster graphics by a decade or more.
-michael
Some scopes have TV inputs and use a built in vertical and horizontal ramp
generator with the video driving the Z axis, usually a 20 volt signal.
But the video showed a display that was clearly an X-Y display, not
a raster display. And the Tektronix 465 does not have built-in
raster generation--but it does have X-Y and intensity.
In the late 70's I tried a couple dozen circuits to make a good analog vector
display using D/A from an AIM-65 (same speed and processor as A II) and an
X-Y scope. I couldn't handle very many vectors in real-time and would need 8
an MHz Zip or the like to make it worthwhile. memory isn't a problem for a
vector display since any line is defined by two (x,y) pairs. If you use 8 bit
converters (256x256 display) you can go pretty fast. I used 12 bits and had
to do twice as many LDA STA pairs per vector.
The problem is that when the D/A's are (re-)loaded, the voltage outputs
will swing rapidly and exponentially damped toward the new values. This
results in a trajectory that is not usually a straight line, nor is it
of constant velocity, so the "vector" is curved and of non-uniform
intensity. In fact, if the values of both D/A's cannot be latched
simultaneously, there will be a "hook" at the start of each vector.
Of course, short vectors look better.
Either linear ramp generators must be controlled by the D/A's or they
must be low-pass filtered and fed redundant points along the desired
vector to get a straight line (and always to get a curved line).
If the vector itself is blanked, and only a "dot" at the end is
intensified after both D/A's settle, then a respectable point scope
can be achieved (many early DEC CRT displays took this approach).
The images on the scopes could be simpler than raster. The words were made
wavy by riding a sine wave and if you use (X,Y,Z) lists where Z is an on/off
bit, and sort them -- well, the scope is probably fast enough that sorting
won't matter -- just cycle though the list turning the dots on and off as
fast as you can. You obviously don't need enough memory for the full raster.
Many might argue that a raster display is "simpler" than a vector
display. ;-)
The continuously regenerated "display list" is the standard way to
feed a vector display. And "local" order is important because the
easiest and most efficient drawing is done with connected vectors.
Intensity gating is not strictly required, since relatively slow
trace velocity obtained by "plotting" 10 or so points per half-inch
produces a bright trace, and fast vectors (more than an inch long)
can easily be below the level of visible intensity, since they are
only 5% as bright as 0.05" vectors.
In the video, four "resting spots" are clearly visible at the
four corners of the display. Of course, they could be rendered
invisible by overdriving the sweep a little.
Oh, the old analog stuff used integrators to get nice straight lines and an
analog multiplier to calculate the brightness, depending on the speed each
vector was drawn across the screen.
Bingo! The electronics for electrostatic deflection wasn't too
bad, but for a larger, magnetic deflection CRT, there was lots of
complexity and power dissipation to get nice straight vectors
without too much dwell time at vertices.
I was looking for an integrator circuit
with what I called automatic endpoint matching to make a better display for
the Battlezone arcade machine, which approximated the vectors with the first
part of a cap charging IIRC.
This approach works pretty well as long as the vectors are short (only
the first few percent of capacitor charging is used).
Raster displays overtook the effort. I still
prefer the look of a vector display for some applications. See the navigation
displays in 2001 A Space Odessey.
Raster displays combined with a frame buffer are so much simpler
electronically that they essentially drove out vector displays as
soon as the price of frame buffer memory dropped sufficiently.
And, of course, they make it impossible to burn the CRT with
faulty software. ;-)
But it was lots of fun to do light pen tracking with a vector scope...
;-) And if feedback from the light pen was used to alter the display
list regeneration, a very responsive display could be achieved.
-michael
NadaPong: Network game demo for Apple II computers!
Home page: http://members.aol.com/MJMahon/
"The wastebasket is our most important design
tool--and it's seriously underused."