First White Rabbit Tutorial Workshop, Barcelona (Spain), 7 October 2017
With the occasion of the ICALEPCS 2017 conference, we took the opportunity to organize a special type of workshop. Traditionally, WR workshops had been organized around contributions of WR users who come and show their latest developments. They are also a great time to discuss informally and catch up with the latest developments. This time we to catered also for potential WR users, people who have a problem to solve and would like to understand how WR can help them solve that problem. We therefore decided to organize the first WR tutorial workshop.
This format is adapted to typical attendees of the ICALEPCS conference. But we also wanted to see the usual WR workshop crowd, because workshops also play this role of providing a place for people to meet and discuss, which is often perceived as even more useful than the actual technical presentations. We therefore decided to include some advanced topics in the program as well, so that everybody, including seasoned WR users and developers, could learn something new.
WR CERN Team, from left: Grzegorz Daniluk, Javier Serrano, Tomasz Wlostowski, Maciej Lipinski, Adam Wujek, Dimitris Lampridis
Agenda
| Time | Title | Speaker(s) |
|---|---|---|
| 09:30 - 10:00 | Introduction to White Rabbit | Javier Serrano |
| 10:00 - 11:00 | WR in action: distributed synchronization concepts and demo (odp) (pdf) | Dimitris Lampridis, Adam Wujek |
| 11:00 - 11:30 | Coffee break | |
| 11:30 - 12:00 | MIMO systems (with demo) and distributed oscilloscopes using WR | Dimitris Lampridis, Tomasz Włostowski |
| 12:00 - 12:30 | Phase-compensated distribution of RF signals using WR (with demo) | Tomasz Włostowski |
| 12:30 - 13:30 | How to use the WR PTP Core to make your own WR nodes, advanced concepts (odp) (pdf) | Greg Daniluk |
| 13:30 - 14:30 | Lunch break | |
| 14:30 - 15:00 | Advanced diagnostics in a WR network (odp) (pdf) | Greg Daniluk, Adam Wujek |
| 15:00 - 16:00 | Determinism in WR: priority handling and latency bounds (with demo) | Maciej Lipiński |
| 16:00 - 16:30 | Coffee break | |
| 16:30 - 17:00 | Real-Time streaming of information using WR (with demo) | Maciej Lipiński |
| 17:00 - 17:30 | Calibration (with demo) and remote configuration of WR nodes | Greg Daniluk |
| 17:30 - 18:00 | High-precision time and frequency distribution using WR | Javier Serrano |
| 18:00 - 18:30 | Open floor for use cases, requests, ideas |
Abstracts
Introduction to White Rabbit
White Rabbit (WR) is an extension of Ethernet which provides determinism and a common notion of time to all nodes connected in the same network. By determinism we mean a guaranteed upper bound in message transport latency. The common notion of time is delivered using extensions to IEEE 1588 (PTP), currently undergoing a standardization effort. The accuracy of this common notion of time is better than 1 nanosecond. All the basic building blocks in WR are open source software, gateware and hardware. The talk explains the basic concepts behind WR and provides context for the more detailed talks which come later in the day.
WR in action: distributed synchronization concepts and demo
The presentation will start with a general discussion about distributed synchronization concepts using the distribution of a Pulse Per Second (PPS) over a WR network as an example. A short showcase of available WR devices will follow. The demo will show how to setup a WR network from scratch. Starting with a discussion about needed fiber cabling, through the configuration of WR nodes and switches (introducing some newly-developed web-based tools), to an overview of the basic diagnostics in a WR network using standard monitoring and expert tools. The talk will end with a short demo presenting how changes in fiber temperature can affect the synchronisation between two nodes (aka "the hot air gun demo").
Phase-compensated distribution of RF signals using WR (with demo)
Every node in a WR network is provided with a phase-aligned 125 MHz reference. Having same reference clock phase everywhere enables distribution of Radio Frequency clocks and other RF signals through Direct Digital Synthesis (DDS). This talk will focus on a WR-based RF distribution system capable of efficiently handling modulated RF signals. A live demo will follow.
MIMO systems (with demo) and distributed oscilloscopes using WR
MIMO systems
This presentation will focus on the WR network operating at CERN for exchanging trigger pulses between a set of geographically distributed instruments responsible for beam instability diagnostics in the LHC. The system features low&fixed latency message distribution and can be easily tailored to custom requirements. The presentation will be followed by a demo of a small trigger distribution network.
Distributes oscilloscopes using WR
This talk will introduce the idea of utilising a WR network in order to form an interconnected network of distributed oscilloscopes (and other scientific instruments). The main goal is to leverage the sub-nanosecond synchronisation and time distribution provided by WR to “negate” the actual distance between the instruments and provide the user with a unified view and control (e.g. for trigger sequencing and distribution) of all interconnected instruments, as well as automatic discovery and enumeration of newly connected nodes.
How to use the WR PTP Core to make your own WR nodes, advanced concepts
The White Rabbit PTP Core (WRPC) is the main element of every WR node. It is an HDL implementation of a Gigabit Ethernet MAC including the full White Rabbit synchronization stack. This talk will provide a comprehensive description of the WRPC and its interfaces. We will present how user-defined HDL modules can benefit from the timing information provided by the core and how they can use the pipelined Wishbone interface for sending application-specific Ethernet frames. We will also see how the WRPC instantiation in your own WR node can be simplified by using provided HDL wrappers for officially supported boards and FPGA platforms.
Advanced diagnostics in a WR network
White Rabbit is an extension of a regular Ethernet network. The same way you can diagnose off-the-shelf Ethernet switches and routers using the Simple Network Management Protocol (SNMP), you can also diagnose timing and data problems in your WR network. In this talk we will describe the various SNMP objects that are exported by WR switches and WR nodes as well as how they can be analysed using Nagios (an open-source SNMP manager) to ensure correct operation of the network. Finally, we will emulate various failures that may occur in a typical WR installation, show how these can be detected and fixed.
Determinism in WR: priority handling and latency bounds with demo
This talk explains the mechanism to ensure determinism of message transport latency in White Rabbit, the possibilities enabled by this feature and its configuration. A professional IT tester will be used to demonstrate how the deterministic message transport latency works.
Real-Time streaming of information using WR (with demo)
White Rabbit cores provide a streaming facility/layer that can be used in WR Nodes - the WR Streamers VHDL module. The WR Streamers take advantage of WR's features (i.e. synchronization and determinism) to allow data streaming with low and fixed latency. This streaming facility is a building brick of a system that distributes the value of the magnetic field in real-time to clients in different CERN accelerators. This talk explains the WR Streamers, their integration into your design and their applications. The demonstration will illustrate how the WR Streamers are used at CERN.
Calibration (with demo) and remote configuration of WR nodes
To ensure sub-nanosecond synchronization, the White Rabbit software stack needs to estimate the communication link asymmetry. In this talk we will introduce the White Rabbit link delay model to characterize various factors that make the Master-to-Slave and Slave-to-Master link latencies not equal. To model these asymmetries we will demonstrate the calibration procedure that can be performed in the lab before the WR infrastructure is deployed. We will also show how to use a Simple Network Management Protocol (SNMP) to upload calibration parameters to the WR nodes.
High-precision time and frequency distribution using WR
The talk explains the phase noise formalism for studying and optimizing jitter through the use of Phase Locked Loops (PLL). We then move on to apply these concepts in the context of WR, identifying places in a WR link which impose constraints on any jitter optimization exercise, and discussing how to design a very precise time and frequency transfer system using WR.