Experimental Physics and
Industrial Control System

Korhonen Timo <[email protected]> · Fri, 14 Dec 2007 18:05:31 +0100

Andrew Johnson wrote:

ILC-Controls folk: I'm forwarding Ron Rechenmacher's message to the

EPICS developers who are working on an enhanced timing/event API that

will be included in EPICS Base R3.14.10.

EPICS Guys:

The ILC community are discussing timing/event systems, see the

messages forwarded below.

Do any of you have a network (probably UDP-based) driver pair for the

(old or new) Event system? I don't see any particular difficulty with

creating one that would plug in to the Time Governor code, but I don't

remember hearing anyone talking about having created one.

Not at the moment (unless somebody has written such one and forgotten to

tell :-) ). However, writing such a driver would not

be a big deal, for most of the functionality. I would need to take a

close look at some of the issues. I would like to revise

some of the concepts with the drivers anyway (since many years...)

That brings up another point for the Time Governor: should (do) we

have a standard API for event generators? IIRC the Time Governor

stuff is currently just for event receivers, does it make sense to add

an API call for event system drivers that generates a particular event

number? This obviously isn't going to be useful for high precision

timing, but it could be useful, including as a way to implement a

'time distribution over event system' time driver in a generic manner

(although that would almost certainly break compatibility with

existing clients).

The Time Governor has also a part for event generators, at least in our

(Babak's) version. We have a mixture of (SLS) 'old' and 'new' (Diamond

& later) event

cards and need to distribute some global time information over the

network (and also keep the event time in sync with wall clock time with

a servo loop.)

Adding an API call to generate an event number is simple, but it is not

obvious for me at the moment how that could be used in time distribution

(it is late Friday afternoon over here ...)

By the way, I could not resist to comment on this:

The jitter and the drift with this system are several of 10ps

There must be something wrong with the setup (I am not sure whose

comment this was...) we get a jitter well below 10 ps, measured with a

LeCroy SDA6000 (and the 230-series EVG/EVR). Long-term drift can of

course depend on many issues, mainly (ambient) temperature change, but

we have not seen any problems with this at SLS, and when the devices are

operated in steady state.

best regards,

Timo

ps. And yes, the Micro-Research system is indeed the result of evolution

from the Fermilab-APS line. We at SLS took the APS system

as a reference, redesigned it (with Micro-Research) so that we could use

it for all timing including injection. For Diamond we then collaborated

and had a chance to implement many things which were not possible or

feasible at the time we were building SLS. Shanghai Light Source is

indeed using them, as well as the Australian Light Source. All these

places have contributed several good ideas along the way :-)

- Andrew

-------- Original Message --------
Subject: Re: Timing/Event Systems
Date: Fri, 14 Dec 2007 09:58:51 -0600
From: Ron Rechenmacher <[email protected]>
Reply-To: [email protected]
To: Dalesio, Leo `Bob` <[email protected]>

CC: Kazuro FURUKAWA <[email protected]>, Andrew Johnson

<[email protected]>, Frank Lenkszus <[email protected]>, [email protected],

[email protected], [email protected]

Hi,

I suggest to anyone on this list that can influence the development of

the

interface software that the software be constructed in such a way as

to allow

(either as a compile time config or a runtime config) a way to work

without

the hardware (simulation mode -- perhaps allowing for stimulus via

network

messages). CERN has recently added such capability to their front end

software

environment - see 2nd dark bullet on the 3rd page of:
   http://ics-web4.sns.ornl.gov/icalepcs07/WOPA04/WOPA04.PDF
I've been thinking about this for several years and I thing such thinking
ultimately can facilitate collaboration on such system.

--Ron

LCLS is using this timing on RTEMS. Contact Dayle Koturi or Till

Straumann ([email protected] and [email protected]) It

was installed and ported there with much help from Timo and Eric.

At NSLS II we are considering a modification of this approach as we

are looking to develop a standard data and timing distribution to

embed into our FPGA based device controllers. We are working in

concert with Larry Doolittle at LBL while inclulding Timo at SLS,

Jukka at Micro Research, and Rok Ursic at I Tech (libera BPM folks).

Bob

-----Original Message-----
From: [email protected] on behalf of Kazuro FURUKAWA
Sent: Fri 12/14/2007 5:45 AM
To: Andrew Johnson

Cc: Frank Lenkszus; [email protected]; [email protected];

[email protected]; Kazuro FURUKAWA

Subject: Re: Timing/Event Systems

Hello,

Discussion with Ron Rechenmacher and other Fermilab people were very

interesting. Thank you Ron.

We have started some software evaluation based on the discussion. But

is may take time.

For the Event system;

As far as I understand, Timo at SLS (Swiss light source) took the

idea of the event/timing system from APS/ANL and designed his event

system for SLS, EVG/EVR-100 series. Later for Diamond light source

he designed the modules with newer technologies, EVG/EVR-200 series.

It should have much potential, and Jukka at MRF reported on the

possible embedded EPICS on RTEMS on PowerPC on EVG230. (We normally

don't use that PowerPC, yet.)

The event rate is around 100MHz depending on the rf frequency, we can

transfer 2kbyte of information at the rate of ~100MHz/2000, which is

fast and large enough for most applications to provide additional

beam information.

The system is used at Diamond and BEPCII at IHEP-Beijing. We hope to

utilize it at our running machine, KEK Linac, from the next summer.

It is also planned to be used or is evaluated at many places

including LCLS, LANL, Shanghai, SNS(?), BNL(?). The current

developer of the EPICS device support is Eric Bjorklund at LANL.

The jitter and the drift with this system are several of 10ps, and we

will keep some of the old but more precise system for some areas.

However, most of the signals will be replaced with outputs from this

system.

This week, we are evaluating the new device support from Eric on

VxWorks collaborating with people from IHEP, Ms. Lei and Mr. Xu.

Hopefully, the next week we will evaluate its new functionality soon,

and maybe with RTEMS as well. (Although we didn't succeeded in

building it on RTEMS.)

Regards.

>>> On Thu, 13 Dec 2007 11:52:30 JST, Andrew Johnson

<[email protected]> wrote;

> Frank Lenkszus wrote:

> > > > I took a quick look at the Micro-Research web site. I'm not

sure what > > the genesis of their event system is but it looks very

familiar and > > interesting. What I can tell you is Fermi Lab

published an event system > > several years ago. We at APS looked at

what they had done and developed > > what we considered a higher

performance system suitable for general > > timing at the APS in the

very early 90's. Our implementation added > > features such as

synchronized time stamping for record processing. > > Subsequently

SLS took the APS design, updated it and increased the speed > > by

~x10 while maintaining the programming model. I believe the SLS > >

version became commercially produced but I don't recall the firm name.

> > I'm not sure if they were the original, but Micro Research are

the > current suppliers, so this is the great**N grand-son of the APS

Event > System, versions of which have been used at various

accelerators around > the world since we gave away the design. EPICS

support is pretty much > guaranteed with any version.

> > - Andrew

> -- > When a distinguished but elderly scientist states that

something is

> possible, he is almost certainly right.  When he states that something
> is impossible, he is very probably wrong.  -- Arthur C. Clarke

-----
Kazuro FURUKAWA <[email protected]>
 Linac&KEKB,  High Energy Accelerator Research Organization (KEK), Japan
 Telephone: +81-29-864-5200 x4316,  Facsimile: +81-29-864-0321

--

Timo Korhonen PSI (Paul Scherrer Institut, http://www.psi.ch)

 CH-5232 Villigen PSI

 tel + 41- 56 3103262 fax + 41 - 56 310 3383

e-mail: [email protected]

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System