Experimental Physics and
Industrial Control System

"Shen, Guobao via Tech-talk" <tech-talk at aps.anl.gov> · Mon, 10 Aug 2020 15:47:13 +0000

Hi Kiman,
It seems that our DAQ system for APS-U is a good solution for your needs.
It uses EPICS7, and is based on AreaDetector.
You can either save your data into a HDF5 file using the AD HDF plugin, or stream it to your second level server for further process.
Here we plan to use it with our data management system.
Performance wise, it can handle a few 100 MB data rate with streaming mode.
Since you use it for PM purpose, the performance is not a concern.

Thanks,
Guobao

On 8/9/20, 10:14 PM, "Tech-talk on behalf of Ha, Kiman via Tech-talk" <tech-talk-bounces at aps.anl.gov on behalf of tech-talk at aps.anl.gov> wrote:

    Hi Mark,

    This application we called Post-mortem(PM) epics client and needs to run once a beam dump condition.
    As you know, the beam dump is not open it depend on machine status, but I say test mode once per ~5 miniature.
    We have a C based PM epics client application for beam operation support. 
    The actual waveform size is up to 1.7 M points and a few thousand waveform PVs, for the RF, PS, BPM, AI, FOFB, BxB .

    We were trying to build a new PM epics client for the new LLRF system; that's why I am collecting information from epics experts on which concept is a more generic way for future support (maintainability, flexibility, technical support..)

    My colleague Yong Hu was working with python-based epics client applications and already got some results. 
    And other experience and comments are welcome.

    Thank you for your sugession!
    Kiman

    -----Original Message-----
    From: Mark Rivers <rivers at cars.uchicago.edu> 
    Sent: Sunday, August 9, 2020 10:16 PM
    To: Ha, Kiman <kha at bnl.gov>
    Cc: tech-talk at aps.anl.gov
    Subject: Re: HDF5 client for EPICS 3.15

    To give good advice on this we need to know more about the performance requirements. If we assume the waveforms are doubles (8 bytes per sample) then each capture is 1.8 M * 40 * 8 = 576 MB. How often do you need to do this, e.g. once per second, once per hour, or once per day? That could determine if a Channel Access client is practical or if you need something within the IOC.

    Mark

    Sent from my iPhone

    On Aug 8, 2020, at 3:41 PM, Ha, Kiman via Tech-talk <tech-talk at aps.anl.gov> wrote:

    ?
    Hi All,

    NSLS-II LLRF system has a new circular buffer for the LLRF system oscilloscope function.
    A single waveform size is 1.8 M point and a total of ~40 PVs.

    Does anyone have and any EPICS HDF client program for waveform PVs to HDF5 files?
    Our application is simple steps if the beam dump caused? Read waveforms and write HDF5 files to a specific directory (DD/MM/YYYY).

    Thanks,
    Kiman

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System