Frequently Asked Questions about PixCellent Camera Systems
Common features of PixCellent cameras

PixCellent manufactures a range of high performance camera systems. These systems combine extremely high sensitivity, precision and dynamic range. The electronics have been designed specifically to obtain high quality, low noise images and to digitise the light levels within those images to a high degree of accuracy.


For the highest sensitivity, don't I need an intensifier?

In order to obtain a sensitive camera it is important to capture the maximum amount of light and to add the minimum of unwanted noise. The CCDs used in PixCellent cameras are extremely efficient at capturing light and converting this into useful signal. The performance criterion for this is the Detective Quantum Efficiency (DQE). Typically CCDs have a peak DQE of around 50%, with some special CCDs offering up to 85% DQE. The equivalent for most intensifiers is around 10-20%. PixCellent captures up to eight times more light than an intensifier.
Also, intensified cameras are inherently noisy - images are marred by spurious flashes of light caused by thermally generated signals. PixCellent has designed all its cameras to minimise noise sources and to deliver the highest quality images.
Probably the most serious problem with intensifiers is that of signal induced background. It seriously limits the dynamic range in an image to not much more than 100:1 at best. CCD cameras from PixCellent do not suffer from this effect and offer dynamic range of up to 100,000:1.


But I can frame average my intensifier to produce noise-free images.

Many frames can be averaged together to average out the light levels, but each frame contains this noise, so the effect is to average out the noise, not to remove it. PixCellent cameras are integrating cameras - the image is held on the CCD for a period of time whilst the signal builds up and read-out just once. With PixCellent a very small amount of noise is added on read-out, but the signal is building up on the CCD for the full integration time. This process is much more efficient at producing a high signal to noise ratio imaging than adding noise and then averaging.


What types of cameras are available from PixCellent?

PixCellent cameras can be split into two main generic types, Antares and Capella.
The Antares cameras are the highest performance imaging systems available. Images are digitised to 16 bits (65536 grey level) accuracy. Slow scan read-out electronics ensure the lowest noise possible. The Antares cameras are used where light levels are the lowest (down to 10-11 lux with some cameras) or where the highest precision of light measurement is required. For the ultimate in sensitivity long exposures are required, necessitating cooling of the CCD by Peltier or liquid nitrogen. The Antares cameras have programmable read-out regions, binning, gain and read-out speed.
The Capella cameras also use very low noise electronics and cooling to minimise dark current, but are faster than the Antares. Images are digitised to 12 bits (4096 grey level) accuracy. Optionally the Capella beta digitises to 14 bits (16384 grey level) accuracy. The Capella cameras are extremely flexible, with programmable read-out regions, binning, gain and read-out speed. Read out speeds offered are 500kHz, 1MHz, 2MHz, 4MHz, 5.5MHz and for some CCDs 8MHz. Fully flexible binning is available at 500kHz, 1MHz and 2MHz.



I can only afford a low resolution CCD at the moment, can I upgrade in the future?


Most definitely yes. PixCellent camera electronics are all designed to drive any suitable CCD. The clock and bias voltages are all programmed by using a software configuration file. If a different CCD is to be driven all that is required is for a new camera head to be supplied fitted with the other CCD and a new configuration file. There is no need to replace the electronics, interface card or control software. "Future proof your investment".


Won't ageing components and temperature variations affect my measurements?

No, PixCellent's electronics are uniquely self calibrating. As described above, the CCD bias and clock voltages are held in a configuration file. When the camera is initialised (normally when first powered up) this file is read and all the internal camera voltages and bias points are set according to the configuration file. The levels are measured and fine tuned as required. This ensures that the bias points, and hence the light measurements you make today will be the same as tomorrow, as next week, next month or even next year irrespective of components ageing or variations in ambient temperature.


How often will I need to re-pump my Peltier cooled camera head.

Never - PixCellent's Peltier cooled camera heads are filled with dry air. Careful insulation techniques are used to ensure extremely low CCD temperatures (and hence extremely low dark current) is achieved without evacuation. Many years of experience with this head design has shown that the heads are essentially totally maintenance-free.
If the peak well capacity of a pixel is 200000 electrons, and shot noise on this is 450 electrons, the dynamic range is 450:1 (or 9 bits), how can you claim 16 bits?
This argument assumes that you are only interested in the one pixel which has 200000 electrons. Most images consist of rather more than one pixel! Another pixel within the image may have only 10 electrons in it. The shot noise on this is 3 electrons. To digitise both pixels to the shot noise limit each grey level (or ADC count) must be 3 electrons, and the ADC saturation must correspond to 200000 - a dynamic range of 67000:1, or 16 bits.


Why is it easier to obtain good images with a 12 or 16 bit camera than an 8 bit one?

Most 8 bit cameras are used at less than optimum performance. The camera will only produce 8 bits if the peak light level corresponds to camera saturation - i.e. filling the 256 grey levels available. This means that the light levels, gain and integration times must be carefully adjusted for this. With a 12 or more bit camera excellent images may be obtained over a very large range of light levels without changing gains or integration times. For instance, a 16 bit camera with 256 times as many grey levels as an 8 bit one can produce 8 bit quality images over a range of light levels spanning 256:1 range. It is very easy to guess the required exposure time within these limits! Clearly, to achieve images with better than 8 bit quality requires a little more care, but excellent images are easy to produce.


Why do I need a 12 (or more) bit camera when my highest light level corresponds only to 400 grey levels.

One of the important features of the PixCellent cameras is the ability to resolve very low contrast features. In order to achieve this the camera's inherent noise must be very low and digitisation accuracy must be very high. This is generally not the case with 8 bit cameras. Many customers have commented that when using PixCellent cameras they have seen features which are visible to the eye but cannot be seen with other cameras. This is due to the eye/brain combination being extremely good at resolving low contrast features within even noisy images. If grey levels are relatively far apart these low contrast features are represented as uniform grey. With a 12 bit camera the grey levels are 16 times closer together and able to resolve much smaller intensity differences.


How do I switch off the AGC?

There is no AGC on the PixCellent cameras. High bit depth cameras such as those from PixCellent produce images with much greater intensity detail than can be displayed on a standard monitor and much greater than can be visualised by the eye. It is therefore essential to have the means to select which intensity levels from the image are displayed (i.e. are we interested in visualising bright objects, dark objects, or low contrast features?). PixCellent's PixCel software has an intensity mapping control panel built in to permit this selection. When these controls are used only the display mapping changes, not the underlying data. Hence, any intensity measurements will be accurate even if the feature appears deep black or saturated white on the screen. In order to visualise an image immediately after its capture the software has an autoscale function to attempt to determine the best brightness and contrast mapping functions so that the software optimises the on-screen display for this particular image. This is often mistaken for AGC (Automatic Gain Control), but is only a screen mapping function and does not affect the accuracy of the underlying intensity measurements. The data in the recorded image in memory are unchanged. With the PixCel software, if the displayed image is not ideal the intensity mapping control panel may be used to adjust the brightness and contrast in any of three ways - using brightness and contrast slider bars, moving bright and dark cursors over the displayed intensity histogram, or typing in values for the brightest and darkest feature of interest. Once set press "Apply" to fix these brightness and contrast settings for all future images. Pressing "Autoscale" again will return the system to its autoscaling mode.


What is the sensitivity in lux of the PixCellent cameras?

The sensitivity in lux depends upon a number of factors: The CCD used, the cooling method and the integration time. Normally we quote the limiting sensitivity of our cameras as that light level which accumulates charge at the same rate as the dark current. This gives values of between 10-11 to 10-5 for various cameras in our range.
Why is dark current not the same as dark noise?
Dark current is the signal produced by thermal activity in the CCD. Dark current will depend upon CCD temperature (roughly decreasing by a factor of 10 for every 20C drop in temperature). The average dark current will increase at a constant rate and therefore produces a grey background in long exposures. After a given exposure each pixel will hold a wanted light produced signal and an unwanted dark signal (or dark charge). Associated with any signal is a noise known as "shot noise" which represents the uncertainty of any particular pixel accumulating that signal. Numerically this noise is equal to the square root of the dark signal. (e.g. if dark current is 1 el/pix/sec, the dark signal after a 100 sec exposure is 100 el/pix, but the dark noise is only 10 {square root of 100} electrons).

 

Home:


The pages on this Web Site are Copyright of PixCellent Imaging Ltd., Cambridge, England. Reproduction in part is permitted with acknowledgement to PixCellent Imaging Ltd., and of this Web Site Address (http://www.pixcellent.com/).
Comments, please, to webmaster@pixcellent.com
Site last updated: 12 April, 1999.