Minutes of the 15^th Timing Working Group Meeting (TimWG)
Held on Friday the 1^st November 2000

Present:

Agenda

• Approval of previous minutes
• Review of timing distribution, Gary BEETHAM
• TTC costs and availability, Bruce TAYLOR
• Fast and slow timing over TTC, Jean-Jacques SAVIOZ
• No timing network, Julian LEWIS
• Implication of timing distribution over common purpose network, Michel JONKER
• AOB

Approval of previous minutes

There were no corrections on the previous minutes.

Review of timing distribution

Gary BEETHAM raised up the question of the 400Mhz distribution. It has been agreed by SL/HRF to take out the distribution of this signal from the general timing distribution system. He reviewed also the different types of signals (bunch frequency, revolution frequency, slow timing, beam synchronous timing) to be distributed around the machine (see slide attached) and outlined the possibility to distribute them over a single optical fiber to all the clients by using the TTC system.

TTC costs and availability

Bruce Taylor reviewed the cost and the availability of the different components used in the TTC system (TTCmi, TTCtx, TTCrx...). A first rough cost estimation for one output (providing all the different types of timing) is actually in the order of 250CHF/destination. More details on costs can be obtained either from Gary BEETHAM or from Bruce TAYLOR. Major components of the TTC are already available or in final development phase. Major activity is actually on the packaging of the TTCrx circuit. Regarding the question of long term support, Bruce TAYLOR explained that it is actually foreseen that give Electronic Pool (EP) and On line Systems (OS) sections of the EP/ESS group provide the support for the experiment side. Nothing is actually foreseen in case the accelerator uses the same solution. It appeared rapidly that the major part of the cost is inside the installation of the fiber optics infrastructure around the machine in comparison with the cost of the hardware components. 

Fast and slow timing over TTC

Jean-Jacques SAVIOZ presented the  different possibility to transmit BST messages across the TTC system (see slides attached). The goal is to transmit to all the SL/BI acquisition crates distributed around the machines timing signals (bunch frequency and revolution frequency) and beam synchronous commands. The idea is to use the TTC system, which already transmit bunch frequency and revolution frequency, by inserting BST message inside the TTC frame. Simultaneously it appeared also possible to extend this approach for the transmission of a limited number of slow timing events. During his presentation, he outlined the question of compensation of the fiber length and radiation hardness of the TTCrx ASIC and FPGA. Finally he reminded that two independent systems will be implemented (one for each ring) in order to be able to make simultaneously different measure. He confirmed that the proposition made by SL/BI can transmit slow timing event which sufficient precision and latency but raised up the question of support and maintenance if this solution will be used.

No timing network

Julian LEWIS presented a proposition for the realization of a slow timing system without a continuous distribution of timing event across a dedicated timing network (see slides attached).  The basic idea of this new approach is that each module becomes a independent timing generator and that modules, distributed around the machine, are synchronized together by the PS 10Mhz frequency modulated by the GPS one pulse-per-second signal. Each module has a resident relative time tables describing the "partition" to be played. These tables have been previously downloaded through a standard common purpose network with a reliable network protocol. The current idea is to use this approach for the replacement of the TG8 card inside the PS complex.

Slow timing over 'common purpose' network media

Michel JONKER evaluated the possibility and the implication of  using a common purpose network for the distribution of slow timing (see slides attached). He explained that if slow timing over 'common purpose' network appears to be feasible, the weakest point appears to be the guarantee of synchronization between the different equipment i.e. the guarantee of reception of the event message by the equipment. This problem can only be solved if a confirmation message is send back from the receiver to the sender. In order to solve this problem a complex software synchronization protocol, including registering mechanism and recovery procedure, has to be implement. This approach will in any case reduce significantly the availability and reliability of the system.

Julian LEWIS explained that his proposition relies on highly reliable protocol, that deterministic protocol can also be used and that the number of transaction between the sender and the receiver will not affect the availability of the system because the 'partition' to be played has been previously loaded inside the module and network communication is only used to load new 'partition' or modify the current running 'partition'. Gary BEETHAM concluded that this subject has to be discussed more in details in a further meeting.

Next meeting

Next Meeting will be held on Friday 1st December 2000 at 14:00 in 864 1C01.

Agenda

1) Progress on RF timing requirements, Philippe BAUDRENGHIEN
2) SLTC presentation, Gary BEETHAM
3) Lab. demonstration
5) AOB