Maker Pro
Maker Pro

Quasi-interesting CRT puzzler

A

Ancient_Hacker

Jan 1, 1970
0
Okay, this is kind of a retarded question, but here goes.

Let's say you want to convert you basic PC monitor from raster scan to
vector.

You pull off the yoke connector from the main board and find out the it
still runs okay, well except that you've burnt a little spot of
phosphor in the center of the screen unless you were prescient enough
to back off the brightness, contrast, and the three screen controls
first.

Then you hook up a DC supply to the V and H yoke coils and find out it
takes about 3 AMPS to do a full screen deflection in the vertical
direction, about TEN amps in the horizontal direction.

(I think that's because the horizontal deflection has to happen faster,
therfore lower inductance in the h coil, therefore fewer gausses per
amp).

So we need some pretty hefty drivers, many amps plus many volts if we
want fast deflection.

Not a huge problem so far.

But now we realize we don't want to burn the screen if we're drawing
short or slow vectors, and we don't want the lines to be invisible if
we're drawing them long or quickly.

So we need some sort of Z-axis brightness modulation, somehow
proportional to the "writing rate", like in inches per millisecond or
thereabouts.

Anybody have any idea how to compute this on the fly as it were?

Ideally I'd like to be able to draw 6000 full length vectors per
second.

A quick web search didnt find anything promising.


Regards,

George
 
G

Graham Holloway

Jan 1, 1970
0
Ancient_Hacker said:
Okay, this is kind of a retarded question, but here goes.

Let's say you want to convert you basic PC monitor from raster scan to
vector.

You pull off the yoke connector from the main board and find out the it
still runs okay, well except that you've burnt a little spot of
phosphor in the center of the screen unless you were prescient enough
to back off the brightness, contrast, and the three screen controls
first.

Then you hook up a DC supply to the V and H yoke coils and find out it
takes about 3 AMPS to do a full screen deflection in the vertical
direction, about TEN amps in the horizontal direction.

(I think that's because the horizontal deflection has to happen faster,
therfore lower inductance in the h coil, therefore fewer gausses per
amp).

So we need some pretty hefty drivers, many amps plus many volts if we
want fast deflection.

Not a huge problem so far.

But now we realize we don't want to burn the screen if we're drawing
short or slow vectors, and we don't want the lines to be invisible if
we're drawing them long or quickly.

So we need some sort of Z-axis brightness modulation, somehow
proportional to the "writing rate", like in inches per millisecond or
thereabouts.


Write at a constant speed. Variable speed is trouble and will require manual
adjustments to compensate for CRT variables.
Anybody have any idea how to compute this on the fly as it were?

Ideally I'd like to be able to draw 6000 full length vectors per
second.

A quick web search didnt find anything promising.


Regards,

George


Graham H
 
A

Ancient_Hacker

Jan 1, 1970
0
To paraphrase Gene Kranz "Constant speed is not an option". I need to
draw some straight lines, arcs, curves, all coming from different
analog sources, each with its own deliberate and preset and in several
cases, unchangeable sizes, lengths, and drawing rates.

I know it's doable as I just got a new $400 crown fitted to my third
molar. Ouch.

Explanation: ( My dentist is a "somewhat less than perfect kids
dentist" )

Further explanation: (He has an ancient "Asteroids" arcade video game
juggernaut in his waiting room to keep the kids busy))

Further Explanation: (Being a bit off-, I had to go back for several
refittings to get the crown to sit right so I could actually eat
anything other than chocolate milkshakes (not that I was complaining),
I spent lots of time in the waiting room.)

Much Further explanation: ( The game draws many different length but
seemingly constant intensity vectors and arcs to draw the asteroids )
 
J

John Woodgate

Jan 1, 1970
0
dated said:
So we need some sort of Z-axis brightness modulation, somehow
proportional to the "writing rate", like in inches per millisecond or
thereabouts.

Anybody have any idea how to compute this on the fly as it were?

Ideally I'd like to be able to draw 6000 full length vectors per
second.

You can overcome the problems by retaining the raster scan at zero
brightness and using the vector information to brighten the spot. I
don't know exactly how it's done (comparators, I suppose), but I have a
Rohde & Schwartz sweeper that does it that way, and the servicing
information on it.

6000 vectors/s doesn't represent a very high video frequency.
 
T

Tony

Jan 1, 1970
0
This is sort of an obvious answer, but since the write speed in each
direction has to do with the rate of change of current, which is
proportional to the voltage on the coils, couldn't you just use op amps
to sum the absolutes of the suitably scaled rates of change of voltage
and use that to modulate spot intensity? I recognize spot velocity
isn't proportional to the two linear sum of the two coil voltages, but
it may work well enough as an approximation. You could, I guess, get
really fancy and via analog methods modulate the spot brightness in
proportion to the square root of the sums of the squares of the squares
of the voltages. That would be an entertaining and not too difficult a
circuit to design.
 
A

Ancient_Hacker

Jan 1, 1970
0
John said:
You can overcome the problems by retaining the raster scan at zero
brightness and using the vector information to brighten the spot. I
don't know exactly how it's done (comparators, I suppose), but I have a
Rohde & Schwartz sweeper that does it that way, and the servicing
information on it.

Thanks, I found a tek patent on exactly such a circuit. Not that I'd
copy it exactly. But the general method is obvious once you hear it.

You just differentiate the x and y signals and that gives you a measure
of the velocity. Ideally you'd square and square root the sum to get
the actual distance per time info, but just adding them should be good
enough for starters.
6000 vectors/s doesn't represent a very high video frequency.

It does if you're trying to do it with a deflection yoke! To delta the
current form -10 to +10 amps across a typical 2millihenry deflection
coil 6000 times per second takes how many volts? Waay out of the SOA
of most transistors. I wish there were a shortcut.
 
A

Al

Jan 1, 1970
0
I know it's doable as I just got a new $400 crown fitted to my third
molar. Ouch.

Explanation: ( My dentist is a "somewhat less than perfect kids
dentist" )

Further explanation: (He has an ancient "Asteroids" arcade video game
juggernaut in his waiting room to keep the kids busy))

You got a crown for only $400???? Where is your dentist. Here, in
Taxachuesetts it's at least $1100.

Al
 
J

John Woodgate

Jan 1, 1970
0
In message <[email protected]>,
dated Thu said:
It does if you're trying to do it with a deflection yoke! To delta the
current form -10 to +10 amps across a typical 2millihenry deflection
coil 6000 times per second takes how many volts? Waay out of the SOA
of most transistors. I wish there were a shortcut.

Indeed; I meant as a video frequency for the brightness modulation of
the raster.

True vector displays date from the valve/tube era, with higher impedance
coils and BIG voltages. The horizontal output valve/tube would typically
work with 3 kV pulses on the anode.
 
E

Eeyore

Jan 1, 1970
0
Ancient_Hacker said:
Okay, this is kind of a retarded question, but here goes.

Let's say you want to convert you basic PC monitor from raster scan to
vector.

You pull off the yoke connector from the main board and find out the it
still runs okay, well except that you've burnt a little spot of
phosphor in the center of the screen unless you were prescient enough
to back off the brightness, contrast, and the three screen controls
first.

Then you hook up a DC supply to the V and H yoke coils and find out it
takes about 3 AMPS to do a full screen deflection in the vertical
direction, about TEN amps in the horizontal direction.

(I think that's because the horizontal deflection has to happen faster,
therfore lower inductance in the h coil, therefore fewer gausses per
amp).

So we need some pretty hefty drivers, many amps plus many volts if we
want fast deflection.

Not a huge problem so far.

But now we realize we don't want to burn the screen if we're drawing
short or slow vectors, and we don't want the lines to be invisible if
we're drawing them long or quickly.

So we need some sort of Z-axis brightness modulation, somehow
proportional to the "writing rate", like in inches per millisecond or
thereabouts.

Anybody have any idea how to compute this on the fly as it were?

Ideally I'd like to be able to draw 6000 full length vectors per
second.

A quick web search didnt find anything promising.

I one worked on a spiral scan CRT system for radar. That was fun !

Graham
 
G

Graham Holloway

Jan 1, 1970
0
Eeyore said:
I one worked on a spiral scan CRT system for radar. That was fun !

Graham

Tuned deflection coils, more efficient use of power?

I worked on a radar display in the early 70's that used deflection coils of
around 10uH total. We used a class A push-pull output stage with
centre-tapped coils and a series inductor. We got around 5MHz small signal
bandwidth. The energy stored in the inductor provided the high voltage
needed for fast transitions of the beam. The system could display a 625 line
TV picture and write all over it with vectors and characters in the vertical
flyback period. Problem - 800 watts per axis!

Graham H
 
H

Helmut Sennewald

Jan 1, 1970
0
Ancient_Hacker said:
Thanks, I found a tek patent on exactly such a circuit. Not that I'd
copy it exactly. But the general method is obvious once you hear it.

You just differentiate the x and y signals and that gives you a measure
of the velocity. Ideally you'd square and square root the sum to get
the actual distance per time info, but just adding them should be good
enough for starters.


It does if you're trying to do it with a deflection yoke! To delta the
current form -10 to +10 amps across a typical 2millihenry deflection
coil 6000 times per second takes how many volts? Waay out of the SOA
of most transistors. I wish there were a shortcut.


Hello,

So let's do the math.

V = L * di/dt

Now I assume the worst case. The factor two is for forward and
backward movement so that we are back at the starting point.

V = +/-(2e-3 * 10*2*6000) = +/-240V

This requires a +/-300V supply with 10A.

I would say possible, but mission useless. Ok, it's a hobby project.

Best regards,
Helmut
 
Ancient_Hacker said:
[...]

You just differentiate the x and y signals and that gives you a measure
of the velocity. Ideally you'd square and square root the sum to get
the actual distance per time info, but just adding them should be good
enough for starters.

Adding the absolute values, I assume.

I could see a problem with differentiating input that is switched
from one analog source to another. Unless the transition is smooth
you'll see big derivatives you may have to control and/or ignore.

Anno
 
R

Robert Latest

Jan 1, 1970
0
On Thu, 17 Aug 2006 12:37:56 GMT,
in Msg. said:
You got a crown for only $400????

Not a very good one it seems.

robert
 
J

James Beck

Jan 1, 1970
0
grg2 said:
Okay, this is kind of a retarded question, but here goes.

Let's say you want to convert you basic PC monitor from raster scan to
vector.

You pull off the yoke connector from the main board and find out the it
still runs okay, well except that you've burnt a little spot of
phosphor in the center of the screen unless you were prescient enough
to back off the brightness, contrast, and the three screen controls
first.

Then you hook up a DC supply to the V and H yoke coils and find out it
takes about 3 AMPS to do a full screen deflection in the vertical
direction, about TEN amps in the horizontal direction.

(I think that's because the horizontal deflection has to happen faster,
therfore lower inductance in the h coil, therefore fewer gausses per
amp).

So we need some pretty hefty drivers, many amps plus many volts if we
want fast deflection.

Not a huge problem so far.

But now we realize we don't want to burn the screen if we're drawing
short or slow vectors, and we don't want the lines to be invisible if
we're drawing them long or quickly.

So we need some sort of Z-axis brightness modulation, somehow
proportional to the "writing rate", like in inches per millisecond or
thereabouts.

Anybody have any idea how to compute this on the fly as it were?

Ideally I'd like to be able to draw 6000 full length vectors per
second.

A quick web search didnt find anything promising.


Regards,

George
Get a few schematics for the old WellsGardner Color XY monitors used in
the likes of Atari Tempest and so on.
They basically had a deflection coil that was symetrical for the X and
Y, used identical X and Y deflection circuits and used the amplitude of
the drive circuit to handle the bigger X than Y (yes it could drive the
guns off the edge of the phosphor). The guns were driven just like they
are in a raster scan system just with the addition of a "Spot Killer"
circuit that would kill the guns if there were no X or Y deflection,
keeping you from burning a hole or line in the phosphor.
The X and Y amps look a lot like a beefed up audio amp.

Jim
 
R

Rich Grise

Jan 1, 1970
0
You can overcome the problems by retaining the raster scan at zero
brightness and using the vector information to brighten the spot. I
don't know exactly how it's done (comparators, I suppose), but I have a
Rohde & Schwartz sweeper that does it that way, and the servicing
information on it.

6000 vectors/s doesn't represent a very high video frequency.

This is really the only practical way to do it, other than to buy
an actual vector monitor. Converting one would be a nightmare - for
one thing, the yoke is dramatically different from a raster scan
monitor's yoke - both sets of coils are the same (in a vector
monitor) with about two dozen turns of about #18 wire. And you'd
have to take out the deflection amps, while somehow keeping the
HV supply, and replace them with some hefty power amps.

Having worked on both kinds of monitors back in my video game
repairman days, I'd say it's a non-starter.

Thanks,
Rich
 
R

Rich Grise

Jan 1, 1970
0
To paraphrase Gene Kranz "Constant speed is not an option". I need to
draw some straight lines, arcs, curves, all coming from different
analog sources, each with its own deliberate and preset and in several
cases, unchangeable sizes, lengths, and drawing rates.

I know it's doable as I just got a new $400 crown fitted to my third
molar. Ouch.

Explanation: ( My dentist is a "somewhat less than perfect kids
dentist" )

Further explanation: (He has an ancient "Asteroids" arcade video game
juggernaut in his waiting room to keep the kids busy))

Further Explanation: (Being a bit off-, I had to go back for several
refittings to get the crown to sit right so I could actually eat
anything other than chocolate milkshakes (not that I was complaining),
I spent lots of time in the waiting room.)

Much Further explanation: ( The game draws many different length but
seemingly constant intensity vectors and arcs to draw the asteroids )

Next time you're in there, ask him if he'll open up the game and let
youi look at the monitor yoke. That should give you some idea of what
you'd be up against.

Sorry,
RIch
 
A

Ancient_Hacker

Jan 1, 1970
0
Helmut said:
Hello,

So let's do the math.

V = L * di/dt

Now I assume the worst case. The factor two is for forward and
backward movement so that we are back at the starting point.

V = +/-(2e-3 * 10*2*6000) = +/-240V

This requires a +/-300V supply with 10A.

Yipes! Luckily I overstated the actual situation-- we only need to
draw about 20 full-screen lines 60 times a second, so we only need 10
amps at 1/5th of that, or 60 volts I think. In addition we need to
draw up to 100 1/4 inch high characters at 60Hz, so applying the fudge
factors, hmmmm.... 60 volts may be plenty. Well within the capability
of a HV op amp with a couple booster transistors. And a heat sink And
a fan.

Crazy project!
 
Top