Re: Brain teaser: Would 300 Hz TV solve the European - US compatibility problem?

On Tue, 22 Sep 2009 09:10:16 -0700, Jim Thompson
<To-Email-Use-The-Envelope-Icon@xxxxxxxxxxxxxxx> wrote:

On Tue, 22 Sep 2009 08:34:01 -0700, John Larkin
<jjlarkin@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:

On Tue, 22 Sep 2009 11:23:50 -0400, Phil Hobbs
<pcdhSpamMeSenseless@xxxxxxxxxxxxxxxxxx> wrote:

John Larkin wrote:
On Mon, 21 Sep 2009 19:33:14 -0400, Phil Hobbs
<pcdhSpamMeSenseless@xxxxxxxxxxxxxxxxxx> wrote:

John Larkin wrote:
On Mon, 21 Sep 2009 22:28:58 +0000 (UTC), don@xxxxxxxxxxxxxx (Don
Klipstein) wrote:

In <sqmfb59cb4entecqf4qd43acb2aal84gpm@xxxxxxx>, Paul Keinanen wrote:
On Mon, 21 Sep 2009 11:33:51 GMT, Jan Panteltje
<pNaonStpealmtje@xxxxxxxxx> wrote:

Working on defeating that Mantis...
The question: how to display a 50Hz camera on a 60 Hz monitor,
without any frame skipping, and without large frame stores (say memory).
So 50 Hz versus 60 Hz,
The first common product is 300.
Refresh rates are going up all the time on LCDs...

Solution:
A LCD with 300 Hz V sync.

How about it?
In some past olympics, 300 fps has been used in the slow motion
cameras e.g. for 100 m track and field or ski jumping. In both cases,
the actual performance is below 15 s, so buffering the signal at the
camera is sufficient (RAM). The readout can be at standard 25 or 30
fps frame rates.

As long as you have sufficient illumination, 300 fps is not a huge
rate.

Doing it at low illumination levels, you really have to think how many
photons will hit a pixel in a frame and in a specific wavelength
band.
How about at 10 lux, a "brightish streetlighting" level?

In my experience with a few datasheets, most CCD sensors do not make
much use of apertures larger than f/2.8, which is 20.25 degrees.

10 lux over that size of an aperture will result in the sensor receiving
.623 lux. (Going by fraction of output from a lambertian radiator being
within 10.125 degree of axis - I hope I got this right.)

.623 lux is .623 lumen per square meter. A lumen of "typical white
artificial light", counting only portion between 420 and 670 nm, is about
3 milliwatts, give-or-take depending on what the light source is (a little
more for incandescent due to heavy deep red content). So now we have
about .0019 or so watt per square meter.

A fairly typical CCD pixel is a 2 micrometer square. Now we have
7.6E-15 watt per pixel.

I would guesstimate 25% of this gets past the dichroic filter for red,
green or blue. That leaves 1.9E-15 watt per pixel, give or take a bit and
a bit more plus some and-then-some.

Now, for 1/300 second: That means 6.3 E-18 joule.

How many photons is that? A 540 nm photon (between "mid-green" and
"middle of visible spectrum") has about 2.3 electron-volts of energy.
Multiply that by 1.602E-19 (reciprocal of number of electrons in a coulomb
of charge), and this very-slightly-yellowish green photon has 3.685E-19
joules.

This means 17 photons entering a "green pixel" per frame at 300 fps from
a scene illuminated with 10 lux of illumination - multiplied by
reflectivity of the scene and transmission of the camera's lens assembly.
I would say 75% of that for white walls (13 photons on average per frame),
a bit less than half for anything else (7-8 photons per frame). A bit
more for red pixels, a bit less for blue pixels, because energy per photon
varies inversely with wavelength.

As for percentage of photons entering a CCD pixel being successfully
sensed - I don't know that one. But already I see signal/noise being low
enough to cause pictures to be very noticeably grainy. Even if all
photons getting past the camera optics and color filters of the CCD are
sensed, I see S/N something like 9 dB for 300 fps at 10 lux.

If the sensor is a 10 or 12 megapixel one (2.5-3 million blue pixels,
same number of red ones, and 5-6 million green ones) used to generate
video content of VGA resolution (.31 megapixel), then the snow may be
"down to a dull roar".

- Don Klipstein (don@xxxxxxxxx)

The ultimate night-vision TV camera is an intensified CCD...

http://www.andor.com/scientific_cameras/istar/features_and_benefits/Default.aspx

which puts a gated microchannel plate amplifier in front of a CCD.
Shutter speeds go down to a couple ns. B+W only, usually. This one has
a nice digital delay generator built in. Frame rate is still limited
by CCD readout.
_Penultimate_. Electron-multiplying CCDs have bragging rights--by
turning one pot you can go from bright sunlight to starlight, and the
worst-case SNR penalty is 3 dB.

Amazingly cute.

Cheers

Phil Hobbs

The ICCDs probably beat them on shutter speed. 1.2 ns isn't bad,
considering that it takes a 300 volt swing to gate the photocathode.

John


That's certainly true. EM CCDs shutter just like other CCDs. Of course
you can set the multiplication even after the light has been detected,
which is a bit of a parlour trick with an image intensifier.

Cheers

Phil Hobbs

One nice mode takes advantage of the best properties of the mcp and of
the ccd: stroboscopic imaging. The mcp can be gated thousands of
times, with high rate and very fast shutter speed, and the ccd can
accumulate the results beween readouts. Sort of an optical 11801.

Oh, Hamamatsu used to make an optical-input sampling scope that used a
photocathode, a streak-tube sort of sampler deflection plate thing, a
slit, and a PMT, all in one jug.

John


Eons ago, when Tektronix was still a real company, I consulted TEK on
an amplifier that used such a "one jug". "One long jug" would be a
more appropriate description... think the old scopes on their own cart
size of box ;-)

...Jim Thompson

The older they are, the longer the jugs?

BTW, anyone have an opinon on whether the Tek 7000 mainframes are
still a good buy?

.

Quantcast