Background

In the following, we refer to detector segments whenever we talk about the chip, and channels when we talk about the integrating electronics. In practice, they are the same.

To calculate the photons flux (say, in kHz) from the voltages read by the ADC, the following steps are necessary. The names given in fixed-width font correspond to the variables in sm_par (stored in the data file) and the entries in the file sidet_calibration.dat. Note that in each case the vectors are stored with sm_par_n_adc_channels entries, whereas only sidet_n_segments entries are significant.

1. Correct for crosstalk. If $\vec S$ is a vector of signals for all channels (but only one scan pixel), the corrected signal is given by
   

   $$\vec S_{\mbox{corr}} = C_{n \times n} \cdot \vec S$$ (1)

   
   
   where $C_{n \times n}$ (crosstalk_matrix) is an $n \times n$ matrix, $n$ being the number of segments.

2. Apply the calibration formula [1, Eq. 3.9]
   

   $$\Phi = \frac{F}{C} \frac{U_{\mbox{out}} - U_{0} - C \times ... ...\times t_{\mbox{dwell}}}{t_{\mbox{dwell}} - t_{\mbox{dead}}}$$ (2)

   
   
   where
   • $\Phi$ is the photon flux in kHz (if all the units are specified as below)

   • $F$ (photons_per_femtocoulomb) is the number of photons necessary to create 1fC of charge in the chip. This depends on the photon energy, but what is constant is the energy necessary to create 1fC of charge in silicon. This is 22594.25ev/fC, corresponding to 3.6eV per electron-hole pair. In the detector calibration file, we store ev_per_femtocoulomb, whereas in the data file we store photons_per_femtocoulomb (as a scalar number each).

   • $C$ (mvolts_per_femtocoulomb) is the calibration constant of the integrating electronics, giving the voltage output per fC of charge created in the chip. This specified as a vector, with a separate entry for each channel. It is predetermined by an off line measurement in Pavel's lab, and then tuned so that the relative values between channels are correct, as described in Michael's thesis [1].

   • $U_{\mbox{out}}$ (in mVolts) is the output voltage of the detector for each channel as measured by the ADC. This is calculated from the scan data, using the ADC parameters adc_voltoffsets and adc_voltsperbit.

   • $U_0$ (u0_mvolts) is the output voltage extrapolated for a zero dwell time and has no physical significance, but it comes out as a linear-fit parameter as described below. Note that even in theory $U_0$ is not zero! The output voltage should be zero (in theory) when the dwell time is equal to the dead time. Hence, $U_0$ must be negative.

   • $I_{\mbox{dark}}$ is the dark current for each segment. The product $C \times I_{\mbox{dark}}$ (c_idark_mvolts_per_ms) is determined by measuring the dark signal (with no x-rays incident on the detector) for various dwell times and applying a linear fit.

   • $t_{\mbox{dwell}}$ is the pixel dwell time of the scan in ms. For piezo scans it will be constant and will be determined from sm_par.clock_hertz, but for stepper scans it can vary from pixel to pixel (due to motor acceleration) and is recorded for each pixel in the data array.

   • $t_{\mbox{dead}}$ (deadtime_ms) is the detector dead time. It is given by the delay plus the width of the S2 pulse on the Quantum Composers timing module. It is 0.101msec as of March 21, 2005.

To determine these parameters, several calibration steps are necessary. For a full calibration (say, if a new detector is installed), you'll have to do the following:

1. Determine the preliminary $C$ parameters (mvolts_per_femtocoulomb) by an off-line measurement in Pavel's lab.

2. Determine the crosstalk matrix (crosstalk_matrix) from a detector map scan.

3. Determine the product $C \times I_{\mbox{dark}}$ (c_idark_mvolts_per_ms) and the voltage offset $U_0$ (u0_mvolts) from a number of dark scans by applying a linear fit (output voltage vs. dwell time).

4. Fine tune $C$ (mvolts_per_femtocoulomb) by measuring the mean charge per pixel in a detector map and correcting $C$ so that the charge is the same in each segment.

5. Make sure the ev_per_femtocoulomb factor is entered properly in the calibration file (which is 22594.25 as long as we are using silicon chips).

6. Enter the dead time (deadtime_ms) into the calibration file.

The calibration is most easily done with the IDL program detcal_gui. The following sections describe how to do each calibration step.