Main Electronics Unit (MEU) and Data Processing Units (N-DPU)

Each Main Electronics Unit (MEU) gathers in the same box:

• 6 N-DPU boards: each N-DPU board is responsible for handling two normal cameras.
• 2 SpaceWire routers: one main and one redundant.
• A redundant Power Supply Unit that converts the primary voltage received from the SVM into the secondary voltages needed for powering the N-DPU boards and the routers.
• A Motherboard for internal connections.

There are two modes of data acquisition from the N-FEEs and data handling by the N-DPU.

1. A nominal science mode acquiring imagettes (or CCD windows) from the N-FEE which have already been extracted by the FEE from the full image. The data of all 4 CCDs from each of the two cameras are transferred inside 25 s to one N-DPU. The data products after data processing are sent as TM packets to the ICU.
2. A test and calibration mode acquiring full CCD images from the N-FEE. These data are not processed and sent as TM packets to the ICU. The data of one N-camera/FEE are sent to the N-DPU within 50 s. Only one camera is read-out at the same time due to the high amount of data and the limited SpW interface bandwidth from the FEE to the DPU and also from the DPU to the ICU. That means two cameras are read-out and the data transferred to the ICU within 100 s.

Each N-DPU board is connected to 2 N-FEE via only 2 SpaceWire links configured to run at 10 Mbps for window mode or 100 Mbps for the full image mode. Each N-DPU board is connected to the nominal router and to the redundant router. Nominally, both MEU routers are working in cold redundancy. However, to handle certain failure cases, both MEU routers can be switched on simultaneously and can work in hot redundancy.

In full image mode the instantaneous data rate between 2 N-FEEs and one N-DPU is about160 Mbps, i.e. the data rate over one link is 80 Mbps, compatible with a SpaceWire link configured to run at a bit rate of 100 Mbps.

In FEE window mode, where the FEE extracts and sends imagettes to the DPU, the data rate is significantly lower.

The data rate between one N-DPU and the MEU router is in average about 700 kbps. The full data rate  between one  MEU  and the  active  ICU  is 4.9  Mbps from all 6 N-DPUs.

In the nominal science observation mode, the main tasks of the N-DPUs are:

• To acquire and store CCD windows as a stream of pixels extracted and sent by the FEE from one 451034510 px image within 6.25s and all CCDs per normal camera within 25 At the FEE level windowing is done to reduce the data volume transferred via SpaceWire.
• To reconstruct the star windows/imagettes to be able to process them (N-DPUs: 636-px windows, every 6.25s, nominally for P1, P4 samples).
• To reconstruct and compute offset, background and smearing data (6.25s)
• To correct pixel data with offset, background and smearing data (6.25s)
• To detect and remove outliers (nominally for P5 samples) (600s)
• To compute light curves and centroids and send them to the ICU (for P5 samples) (600s)
• To estimate the attitude (every about 5400s)
• To update the photometric masks positions to correct for differential kinematic aberration (every about 5400 s)
• To transmit imagettes, light curves, centroids, offset, background and smearing data to ICU every 25s or 600s
• To collect and transmit HK data and send them to the ICU (every 25s)

 

Data products and number of extracted windows for 4 CCDs

Data product (cadence)	Number
Imagettes (25s)	25750 ([P1: 2 cameras35970 + P4: 231675 + P5: 234730 + Guest observer: 23500] star windows)
Light curves (50s)	54736 ([P5: 2 cameras37868 + Guest observer: 2 cameras319500] star windows)
Light curves (600s)	141630 (P5: 2 cameras370815 star windows)
Centroids (50s)	7868 (P5: 2 cameras33934 star windows)
Background values (25s)	800 (2 cameras3400 Background windows)
Offset values (25s)	16 (2 cameras323 4 CCDs á 2255 pixels) offset windows
Smearing values (600s)	36080 (2 cameras34 CCDs á 4510 overscan rows) smearing rows
Housekeeping	120 values

The application software running on each DPU performs the complete data reduction and photometrical extraction process. It is triggered as soon as a set of windows extracted from a full-frame image is available.

The needed processing power has been estimated by prototyping LEON2@100 MHz processor simulator. The measured CPU occupation rate is about 37% which implies sufficient margin for the development process up to the Flight Model. The conclusion is that the normal N-DPU board can be implemented with at least one LEON AT697F processor working at 100 MHz. The CPU load margin could be used to improve the algorithms, to implement new algorithms, to process more targets, to update with a higher frequency the photometric masks or to reduce the processor frequency. The total memory required per N-DPU board will be 256 Mbytes, assuming +50% margin in the number of stars to be monitored per CCD.

 

References

PLATO – Revealing habitable worlds around solar-like stars
Definition Study Report, ESA-SCI(2017)1, April 2017