DigitalDownConverter — Overview

Package Metadata

Field	Value
Name	`dbbc`
Version	`1.0.0`
Description	Processing for VLBI modes using the EDD backends
Authors	Niclas Esser nesser@mpifr.de, Tobias Winchen twinchen@mpifr.de

Entry Points

Command	Module
`edd_ddc_processor`	`edd.dbbc.ddc_processor:main_cli`
`edd_vdif_packer`	`edd.dbbc.vdif_packer:main_cli`
`edd_field_system_interface`	`edd.dbbc.field_system.interface:main_cli`
`edd_fs_mock`	`edd.dbbc.field_system.fs_mock:main_cli`

Dependencies

aiokatcp
setuptools_scm
edd-core
redis
numpy
scipy
matplotlib

Description

The DigitalDownConverter (DDC) is an EDD pipeline that receives wideband time-domain data from the MPIFR EDD Packetizer and performs GPU-accelerated digital down-conversion to extract narrow subbands. Each LO frequency produces a lower and upper sideband channel. The pipeline supports one or two polarisations by using one GPU per polarisation on separate NUMA nodes.

Processed subbands are published as SPEAD2 multicast streams in the DigitalDownConverter:1 format for consumption by a downstream VDIFPacker, or written directly to disk.

Responsibilities

Receive time-domain data via SPEAD2 (MPIFR_EDD_Packetizer:2 format)
Perform polyphase filtering and digital down-conversion on GPU
Publish per-channel subband streams via SPEAD2 multicast or write to disk
Expose a KATCP control interface for remote configuration and lifecycle management
Report output data rate via the output-rate sensor

Operational Context

Item	Value
Upstream	MPIFR Packetizer (4 Gsps ADC)
Downstream	VDIFPacker pipeline, or disk
Control interface	KATCP on `--host` / `--port`
GPU required	Yes — one per polarisation
NUMA sensitive	Yes — capture process pinned to input NIC NUMA node
Container image	`edd_dbbc`
Ansible role	`roles/ddc_processor/`