Speaker
Description
PERCIVAL, "pixellated energy-resolving CMOS imager versatile and large," is a 2-megapixel soft X-ray imager developed for use at FELs and modern-day synchrotrons by a collaboration of light sources (DESY, Elettra, Diamond, Pohang Accelerator Lab, and Soleil) together with Rutherford Appleton Laboratories. To meet the science needs at these facilities, a combination of capabilities is necessary: a large, uninterrupted imaging area with small pixels, high dynamic range, high frame rate, and soft X-ray suitable entrance window to the sensor. PERCIVAL's stitched sensor offers over 4cm x 4cm uninterrupted imaging area (1408x1484 pixels of 27x27 um2). Three gains combine to deliver Percival's dynamic range: in the highest gain, noise levels below 13e- are achieved - suitable for single-photon discrimination down to ~250eV. In the lowest gain, up to 3.6Me- can be digitized per pixel per image. The imager is designed for 300Hz frame rate, operates essentially in a rolling-shutter mode, and can be run correspondingly faster in ROI mode utilizing a reduced number of its rows. BSI processing with thin entrance window makes the sensor suitable for soft X-ray use.
The first generation of the sensor had some shortcomings - in particular, severe crosstalk in the sensor hampered parallel operation of ADC, streamout, and pixel switches. Circumventing these issues by separating the sensitive ADC operation in time from the aggressors resulted in significantly reduced maximum frame rate - and some non-linearities remained. Moreover, inadequate grounding of the pixel matrix resulted in current bias variation over the matrix ... and ultimately in significantly different baselines at center and edges of the detector. This limited the useable dynamic range in particularly in higher gains ... and rendered pixels at the edges of the sensor useable only in higher-noise, lower-amplification modes. The first-generation readout FPGA card and firmware added restrictions on data streamout speed, limiting the overall sensor operation to below 100Hz frame rate.
Today, we are commissioning a 2nd-generation sensor and readout:
The 2nd-generation "respin" sensor's design was modified to eliminate the crippling crosstalk to the ADC, and grounding of the pixel matrix was improved to enable use of the full sensor area also in highest gain modes. In parallel, completely new DAQ hard- and firmware now have the capability to handle data streams at the originally envisioned rates.
To date, we have verified elimination of the crippling crosstalk and established our capability to use the full sensor - i.e. validated the chip design improvements. We are in the process of bringing the system to the original design speed of 300Hz frame rate - as this entails operating both the digital streamout and the ADCs at almost twice the previously-used clock speeds, debugging and commissioning fully will take some time. We will report on the state of the system, and the performances demonstrated with the upgraded sensor and system.
| Workshop topics | Detector systems |
|---|
