WIRCam Cross-talks

Table of contents:

Description of the Crosstalks

Until March 2008, WIRCam was affected by three types of cross-talks. We liberally named them, the "Negative", "Positive" and "Edge" cross-talks. To better grasp the explanations, familiarity with the hardware involved in reading out the arrays is recommended. First, figure 1 gives an example of the Negative and Edge cross-talks.

Figure 1. The Edge and Negative Cross-talks on a Fully Detrended Image.

Caption:The edge cross-talk is seen at right. It gives a 3D impression and affects only a ring of pixels where the source star has fastest rate of flux changes. Note how the center of each doughnut remains unaffected. On the left is an example of negative cross-talk. This is a more serious effect because it creates a complete doughnut with pixels having negative values and an amplitude of about 50 adu, regardless of the source brightness. On June 29 2007, the negative crosstalk was understood and fixed in hardware. In March 2008, the edge cross-talk cause was understood and fixed in hardware (for more detailed, see the WIRCam Changes History web page).

Figure 2. The Positive Cross-talk on a Fully Detrended Image, After Edge and Negative Cross-talks Removal.

Caption:The Positive cross-talk is a more classic case where a doughnut of positive signal appears and is caused by a bright source on a nearby amplifier. This is limited to a video board and affects only certain video boards. On the right is a summary of the three cross-talks caracteristics. The most affected board (video board #3 of extension 4 -- third from the bottom - pixels y=1025-1536) was modified on June 19 2007. (for more detailed, see the WIRCam Changes History web page)

An important note, there is no cross-talk between arrays.

Removal Using the Medamp Technique

The first effort at characterizing and removing the cross-talks made use of the "Medamp" technique. By this we mean isolating then subtracting what is common to all 32 amplifiers. This effectively seems to remove the edge and negative cross-talks which both affect all 32 amplifiers. But it does not remove the positive crosstalk. Note that the assumption is that the amplitude of the edge and negative cross-talks is the same ona ll 32 channels. We tried inconclusively to prove/disprove that assumption. If amplifier-dependant, the amplitude variations must be less than 10%.

We experimented doing the medamp at various stages of the processing and found the best results when removing the crosstalk as the very last step, after sky subtraction. Rigorously, it should actually be the very first step since crosstalk effects are produced in the very last stages of image generation.

Figure 3. Slicing a Detrended Image into its 32 Amplifiers.

Figure 4. Constructing the "Medamp" Image, 2048x64 Pixels Big.

Figure 5. Subtracting the "Medamp" to the 32 Amplifiers

Figure 6. A Closer Look at a "Medamp" Image

Caption: One can clearly see two types of cross-talks: the negative and edge cross-talks. The red circles represent on which amplifier can the source of the cross-talk be found. The Negative cross-talk only occurs when a bright source is located on the first of the four video boards of a detector. Edge cross-talk happens for sources located anywhere on the arrays.

A Before/After Example

The before/after results of applying the medamp technique is examplified in figures 7 and 8 on a detrended image. The technique leaves the positive cross-talk. One could apply a similar technique (but using only the 8 amplifiers of a videao board) to subtract that positive cross-talk. But this was not implemented in `I`iwi version 1.

Figure 7. Before Figure 8. After