Hi-Res CCD Camera Survey

   4 Drawings:
1) Overall Block
2) Single channel using Variable Gain Amplifier
3) Single channel using Multiplying DAC as Variable Gain module
4) Offset & Gain Control

All CCD sensors have a variation in gain and offset between photosites; this is known as PRNU, or Photosite Non-uniformity.
Its effect is to limit the apparent S/N ratio. However, since it is a stationary phenomenon--unlike dark noise--it can be characterize,
mapped and ultimately canceled out. This can be accomplished either as a analog subtraction at the ADC input, or as a
digital subtraction--as in the figure--a LUT stored map of the gain and offset coefficients.

CCD Array Size	# Pixels	Frame Rate	Pixel Rate	Frame Rate	Pixel Rate
1k x 1k	1,048,576	x 30	31,457,280/sec	x 60	62,914,560/sec
Coverage:	x = 120deg.	y = 120deg.	3 cameras	8.53pixel/deg.
Coverage:	x = 60deg.	y = 60deg.	12 cameras	17.06 pixel/deg.
CCD Array Size	# Pixels	Frame Rate	Pixel Rate	Frame Rate	Pixel Rate
2k x 2k	304,194,304	x30	125,829,120/sec	x 60	251,658,240/sec
Coverage:	x = 120deg.	y = 120deg.	3 cameras	17.06 pixel/deg.
Coverage:	x =  60deg.	y =   60deg.	12 cameras	34.13 pixel/deg.

 Exposure Control 

Until RAP is Available:
Two cameras stacked & registered(looking at the same scene): one camera's exposure is predicated on the brightest part of the scene--peak light; the second camera's exposure is based on some deterministic lower light level, thereby capturing previously noisy parts of the scene. Of course, the over exposed image data from that sensor is useless and discarded--as is the noisy data from the other camera. 

By combining the best of both sensors the overall dynamic range of such a system is enhanced in one of two ways:. 

The temporal resolution of the extreme areas (noisy & over exposed) is equal to one camera, while the non-extreme areas--gooddata--are equal to the sum of both cameras. 

If the two cameras are setup for maximum spatial resolution--in the staggered pixel mode--then the spatial resolution of the two extreme areas (noisy & over exposed) is equal to one camera, and the non-extreme areas--when combined--are equal to the sum of both cameras ( i.e., resolution = X2). 

Stacked cameras:

The photosites of each camera are registered on the same scene elements.

Each camera is transfering its image sequentially in concert with the other cameras, having the effect of increased frame rate (1 camera FR = 7.5 FPS; 4 cameras, effective FR = 30 FPS).

Each camera's sensor has the property of shifting its registration a few microns in either X and/or Y directions.

If two cameras were shifted the distance of 1/2 photosite spacing in the X axis relative to the two remaining cameras: the horizontal resolution would be doubled; the vertical resolution is unchanged; the horizontal frame rate has been reduced by 1/2; the vertical image resolvable picture elements is unchanged

Max Frame Rate, all cameras registered on same image, outputting twice as fast as one camera 

Max Resolution: Each camera offset, looking at different parts of the same image     (between photo sites).. 

Best of Both: In a multiple camera--more than two--there can be a "Mix & Match' arrangement: some cameras     grouped for max frame rate , while the rest are setup for max resolution. 

Some areas of a scene are over exposed (blooming) while other areas are under exposed --noisy. 

Sensor Survey Table 1

Stacked CCDs Dwg

Shift Sel Dwg

ADC Dwg

Stacked CCDs TxT

Imaging Metrics

Kinks Dwg

Options Table Dwg

Effic v Light Plots

Solar Spctm Plots

Other Pages

Top Bottom

Scenario:
Two cameras stacked & registered (looking at the same scene): one camera's exposure is predicated on the brightest part of the scene--peak light; the second camera's exposure is based on some deterministic lower light level, thereby capturing previously noisy parts of the scene. Of course, the over exposed image data from that sensor is useless and discarded--as is the noisy data from the other camera. 

By combining the best of both sensors the overall dynamic range of such a system is enhanced in one of two ways:. 

The temporal resolution of the extreme areas (noisy & over exposed) is equal to one camera, while the non-extreme areas--gooddata--are equal to the sum of both cameras. 

If the two cameras are setup for maximum spatial resolution--in the staggered pixel mode--then the spatial resolution of the two extreme areas (noisy & over exposed) is equal to one camera, and the non-extreme areas--when combined--are equal to the sum of both cameras ( i.e., resolution = X2).

This is done by periodically dumping excess photo-electrons from the "wells" before they fill and overflow, and cause blooming. 

This dumping continues until transfer time occurs: This is where all of the accumulated photoelectrons remaining, are "dropped thru" to the waiting analog bins below, and are then shifted out--pixel by pixel--as raw video. 

The above can be thought of as normal exposure to over exposure. In the case of under exposure, the integration time is extended beyond the normal transfer interval (e.g., 1/60 sec) i.e., dumping is held off beyond the normal readout time, or frame transfer time (one frame time). 

Of course, for extended exposure time: frame time can be extended to some multiple of normal frame time (e.g, 1/60 sec) by buffering previous images and allowing longer integration time

Short Explanation:
The relay lens(s) captures the focusing/converging image light (cone of confusion) from the taking lens before it converges or focuses on the focal plane, and collimates it (stops the converging process). The beamsplitters (cemented prisms) steal some percentage of the image light and allows the remaining light to pass undistorted, to the next beamsplitter, and so on. 

By the proper selection of the beamsplitter's transmission/reflection ratio, equal amounts of image light go to each of the finite number of CCD array sensors.

Scratch Pad Area
____________________________________________

T h r e a t s :

Intentional:

Sun Light/Reflections/Glare

Not enough light

Are all WLs needed, can some bands of light be eliminated (notched out)

Are there Optical Communications WLs that the system needs to be Immune from??
Environmental:
The mechanisms that can cause the most grief:
1) Alignment of all of the optical elements! (goes without saying--but I said it!)

2) Focus of the relay lens, i.e., where the front relay lens intercepts the primary image cone of light. This will determine whether any residual divergence or convergence exist over the distance to the secondary relay lenses (CCD end), and ultimately the focal planes of all of the CCD arrays.

That is to say, image registration and the magnification factor (mag = 1.00) will be the same for all arrays if this parameter is correct.
Fixes for any errors:
Some of this can be compensated for--within reason--by the computer, registration being the main one that comes to mind.

more later...