Message-Id: <199601312156.NAA02714@desiree.teleport.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Wed, 31 Jan 1996 13:57:36 -0900
To: sramilin@@msuvx1.memphis.edu
From: semisys@teleport.com (Terrance Hodgins, aka Semi-Intelligent Systems)
Subject: Re: URGENT HELP REQUIRED (fwd)
Cc: rsc@europa.com (Roger Cathey)

Hello.  My name is Terry, aka semisys@teleport.com
My friend Roger forwarded your message to me, as we had just been
talking about an old project I did years ago for the Air Force Weapons
Lab in Albuquerque, New Mexico.

Here is the text I sent him to answer his questions...

====================================================

Um, I think I would put it in other terms:
I had to build some circuitry, and modify the frame-grabber boards, to allow
either software or hardware trigger of frame acquisition.  The frame-grabber
boards inherently digitized the incoming signal into an 8-bit (256-level) gray
scale.  So one frame was like 256 KBytes.
The act of snapping the picture is what freezes the picture in time.  The
TV cameras were special $7000 CID cameras.  They use a Charge Injection
Device imaging array, rather than the commoner CCD, making them better
for use in very low light levels.  Which is, ironically, what the situation is.
These guys were firing high-power gigawatt lasers in a darkened room, which
you would think would really light up the place, but no.  Because a laser beam
doesn't diverge, almost none of the light is visible from any angle other
than dead
center in the beam, which is where you don't want your eye or your camera to be,
or it gets a hole punched in it....
The ideal time to fire the laser is during the vertical retrace time of the
TV camera.
That way, you essentially have a blank slate that will absorb the image, and
then
you read it out as you scan in one TV image.  Each time you scan a CID array,
you erase it, which is handy, because that leaves it clear for getting the
next image.
So you just let the cameras run, and throw away all the frames except the
one that
comes in immediately after firing the laser.
I still don't know what those guys were learning from all of that.

=======================================================
Now, to address your questions:

>----Forwarded----
>Path: 
>europa.com!rainrgnews0!psgrain!news.uoregon.edu!news.corpcomm.net!newstan
>d.syr.edu!newsstand.cit.cornell.edu!news.kei.com!uhog.mit.edu!news.mathwo
>rks.com!uunet!in2.uu.net!newsfeed.internetmci.com!swrinde!cssun.mathcs.em
>ory.edu!nntp.msstate.edu!news.memphis.edu!msuvx1.memphis.edu!sramalin
>From: sramalin@cc.memphis.edu (SaRaDa)
>Newsgroups: sci.electronics.basics
>Subject: URGENT HELP REQUIRED
>Date: 19 Jan 96 11:43:05 -0500
>Organization: The University of Memphis
>Lines: 56
>Message-ID: <1996Jan19.114305@msuvx1.memphis.edu>
>NNTP-Posting-Host: msuvx1.memphis.edu
>
>Hi electronic guroos
>I need some urgent (what else :)) ) help. Any kind of help
>is greatly appreciated and will be acknowledged individually.
>Thanks a lot. Here we go with the problem.
>
>I have a signal from the CCD camera which on CRO shows as
>60 Hz (or 16.66 millisec). We need to control a shutter with
>this signal. Yes, the shutter should close and open with a
>frequency of 30 Hz. But it should operate with this 60 Hz 
>signal of the camera. What I need is, to convert the CCD camera's
>signal (60 Hz - I am told this is the frame rate, two frames make
>a picture and therefore camera is taking pictures at 30 Hz) to 
>30 Hz. The On time of the 60 Hz signal is 0.4 millisecs. But I
>need the 30 Hz signal with an on time of 4 millisec or more.
>(This time must be sufficient to drive the shutter open)
>
Yes, the frame rate is 30/second.  Your are talking about
interlaced scanning, where you first scan all the odd-numbered
lines of the picture, and then go back up and scan all the even-
numbered lines.  The idea behind that is to eliminate visible
flicker on the TV screen.

>So the first step is 
>
>1.  To convert the 60 Hz (0.4ms Ontime) to 30 Hz (with 4ms Ontime).
>
You can use a 74LS74 IC, or any of several other types, just
"flip-flops" to divide any frequency by two.  That is very easy.
Since you are at a university, you should be able to find plenty
of budding EE hardware engineers to help you with the
soldering-iron part of the project.
What is trickier is to get synchronized with the start of the frame.
Since you are getting two vertical retrace pulses from your
vertical sync during each frame, you have the problem of
figuring out which pulse is coming at the start of a frame,
and which pulse is coming from the middle of a frame.
You don't want to be synchronized "out of phase", where
you are beginning to photograph at the second half of a
frame, and then continue into the next frame to get the
first half.
The dead give-away of where you are is the relationship
between the horizontal synchronization pulses and the
vertical sync pulses.  There are 525 lines in a standard
NTSC TV frame.  When you divide that by two, you
get a fraction, like 262.5.  Which means that when you
are starting the first vertical retrace in a frame, you are
starting in the middle of a scan line.

So you need some way to detect whether you are in
the middle of a scan line, or at the start of a scan line,
when you see the vertical retrace starting.  Simple edge-
detection and an AND gate will get you that, so that
you have just one pulse that starts at the start of a
frame.  Basically, you will be wanting to yank the
SET pin of the flip-flop when you detect the concidence
of the starting edges of the vertical

>Second, I need a shutter to be made or purchased which opens
>and closes with the above 30 Hz signal. It should open with
>the ON TIME of the signal and close after a time t, that is
>adjustable. I mean I can open the shutter with the signal
>of 30 Hz but then it remains open for a time t that can be
>either manually or otherwise adjustable.
>
Are you absolutely certain that you need a shutter?
The above systems (4 of them, each with 4 cameras, all synced
together) had no shutters.  We just took advantage of the
fact that the CID (or CCD) array is self-erasing.  Discard
all frames but the one you want.

>So the second problem is
>
>A Shutter that is programmable with this 30 Hz. signal but
>more importantly, the shutter should have adjustable ON time. 
>The on times are minimum in 1 Millisecond range to some few 
>seconds range.
>
>This is how the application goes. I have a laser beam that
>is to be exposed EXACTLY WHEN the camera takes a signal or
>vice versa. I have a CCD (COHU 4815) camera that is connected
>to a frame grabber installed in a Pentium with software ready
>for data analysis etc. But the timing is not available right 
>now. I need to control the timing as said above.
>
Whose frame grabber?  I was using Imaging Technology Inc.
(ITI) frame grabbers, one FG-151, and four PC grabbers.
Five boards altogether.  The PC grabbers actually acquired
the frames, and then the FG-151 did the processing.

We were not able to synchronize the frame grabbers to the
laser, so we synchronized the firing of the laser to the vertical
retrace of the frame grabbers, and just let them free-run.
It worked great.

>Thank you for your time and send me replies at the following
>address, I am not a regular reader of these news groups.
>
>P.S :- Some time ago I designed a rotating spin wheel for
>this mechanism but the DC motor sucked me hard, and then
>the step motor did the same, and when they worked, the timing
>became a problem. I do not think I will go back to them and
>waste some more time breaking my head. Thanks for your help
>once again. Regards.
>
>PPS :- If this is all confusing to you, sorry, English remains
>       my second language and I can explain you personally more
>       if you send me mail.
>----
>sramalin@msuvx1.memphis.edu
>

Not a problem.

Nice to talk with you.

*               Terrance Hodgins                *
*   Chief Boohoo of Semi-Intelligent Systems    *
*           semisys@teleport.com                *
*      http://www.teleport.com/~semisys/        *