EVENT NAME Epics Collaboration Meeting 2020 EVENT DATE: TUESDAY, 20 OCTOBER, 2020 - 04:00 PM to 10:00 AM EVENT BY: HEINZ JUNKES Posted Questions [05:56 PM] Steve Hicks asked : Can you comment on the move away from Galil to PMAC? 6 upvotes | 2 answers | 1 reply Heinz Junkes answered - answered by andy Andy Foster answered - Hi Steve, I don't have any direct experience with the Galil, but many synchrotrons and telescopes have used Delta Tau, now Omron, motion controllers. They are extremely flexible with what you are able to do. Bit of a learning curve, but very flexible. Steve Hicks replied - Yes, I used one 20+ years ago. Very powerful but user hostile - maybe expert friendly :) [05:55 PM] Pedro Gigoux asked : What type of PMACs are you using? 1 upvote | 2 answers | 0 reply Heinz Junkes answered - answered by andy Andy Foster answered - Hi Pedro, we are seeing more and more of the PowerPMAC now. We still use Turbo PMAC II (Ethernet connected) as well. [06:21 PM] jerry seitz asked : network based connects? 1 upvote | 1 answer | 0 reply Freddie Akeroyd answered - Sorry, did I say those words? We generally look for devices that can be made network accessible (with minimum additional hardware) as they are easiest to use with a virtual machine. [06:25 PM] jerry seitz asked : IIoT Industrial Internet of Things. How do you see EPICS role in it this ? 1 upvote | 1 answer | 0 reply Heinz Junkes answered - There are some implementations using TheThingsnetwork, filling EPICS PVs via MQ, will provide a link until tomorrow [06:32 PM] Jure Krasna asked : Do you use hardware pass-through with Virtual Machines in production? Did you do any experimentation with that? Did you find it reliable? 1 upvote | 1 answer | 0 reply Freddie Akeroyd answered - I’m guessing you mean passthrough for e.g. USB access? We haven't used this to date. We use Microsoft Hyper-V for virtualisation and have for a long time, I believe such access was not supported in Hyper-V when we started using it for VMs (VMWare has supported this feature for a while I understand) but this may have changed in later HyperV versions. I’ve dropped a note to our system admin in case he has anything to add. [07:25 PM] Roland Mueller asked : @Ralph E.: could you please comment on an envisaged role of control system archivers as a source of metadata self-describing the instrument set up and environment? 1 upvote | 0 answer | 0 reply [07:30 PM] Uwe Hergenhahn asked : Do you deposit those data that form the basis of a published work, or everything you measure? Do you deposit after measurement or after a publication has come out? 1 upvote | 0 answer | 0 reply [04:51 PM] Maren Hauschildt-Purves asked : Did EPICS 4 just disappear? 0 upvote | 4 answers | 3 replies Heinz Junkes answered - No, its part of EPICS 7 (3 +4 = 7) Maren Hauschildt-Purves replied - Understood, Heinz, but as standalone it is gone, I guess? Ralph Lange answered - Not as a standalone tar file anymore, correct. Some parts of PVA can be compiled for use with an EPICS Base < 3.16, but not all. You will always need a compiler that does C++11. Ralph Lange answered - Probably best to ask on Core-Talk with a description of what exactly you need in which environment. Maren Hauschildt-Purves replied - Was really just curiosity because when it started it sounded like the next version to come. Ralph Lange answered - It is the next version that has come. When we dropped the idea of creating a new IOC from scratch, V4 was really just PVA, and the IOC (aka V3) was just developed further. Maren Hauschildt-Purves replied - Thanks Ralph. [06:30 PM] Roland Mueller asked : Can you please comment on a use case of the sample environment command protocol SECoP? Slide before Summary 0 upvote | 0 answer | 0 reply [06:30 PM] Kevin Peterson asked : Is the code for emulated devices available on github? 0 upvote | 1 answer | 1 reply Freddie Akeroyd answered - Hi Kevin, they are available at https://github.com/ISISComputingGroup/EPICS-DeviceEmulator Kevin Peterson replied - Thanks! [06:37 PM] Gang Li asked : Has Virtual Machines in production two NICs, one is for RS232/422/485，the other is for CA/PVA? 0 upvote | 1 answer | 1 reply Freddie Akeroyd answered - We find our moxa nport boxes for serial are quite stable being on the same NIC we use for CA/PVA, but we do have a second NIC on the VM as we have other network based devices that do not like random network broadcasts for example. Gang Li replied - Thank you for positive response [06:47 PM] Dirk Zimoch asked : Can https://github.com/FreeOpcUa be used with your device support? 0 upvote | 2 answers | 0 reply Ralph Lange answered - The SDK by Unified Automation is currently the only client library that works with the support. The structure is modular, though, to allow other client libraries to be integrated. With the existing client being <4k€ (source code license), no one started such an effort yet. Ralph Lange answered - I would go for open62541 - they are the most complete and best supported OSS client. [06:47 PM] Joerg Penning asked : How about Profinet? 0 upvote | 0 answer | 0 reply [06:49 PM] Hitesh Dhola asked : MODBUS, OPC UA, StreamDevice Out of these three which one is easy to implement in EPICS and which one gives highest performance/reliability ? 0 upvote | 3 answers | 2 replies Mark Rivers answered - All are easy and reliable! Hitesh Dhola replied - :) Thanks. I was reading a document on IDM about performance/speed of OPC UA and s7 drivers for plc. There were some comparative analysis, where there was limitation of how fast one can access or stream data. I don't remember which protocol was having speed bottleneck. But I thought something like that also might happen here. Ralph Lange answered - TCS has run numbers comparing the Zimoch driver (TCP block transfer) with OPC UA. Result: For small numbers of variables, TCP transfer was faster. For large numbers (>20k), both were working fine at 200ms. Ralph Lange answered - Going with a standard, you get some reliability by the tests that devices have to undergo for being able to claim to implement a standard. With strings over TCP, a bad implementation in the device can easily break your neck. But that's a device quality issue, not a way-of-integration thing. Hitesh Dhola replied - Thanks [07:27 PM] Henrik Carling asked : Thank you for the interesting talk! Are you also involved in the legal aspects of data sharing (licenses, sensitivity to GDPR etc)? 0 upvote | 0 answer | 0 reply [07:29 PM] Ernest Williams asked : Any such efforts you know from the US? 0 upvote | 0 answer | 0 reply [07:53 PM] Chitra Venkataraman asked : On what basis would you select the XPS C8 as compared to the 8-axis Geobrick 8-axis for motion control? 0 upvote | 0 answer | 0 reply [07:59 PM] Saeed Haghtalab asked : Thanks Mark for the nice talk. For non-clean ASCII communication devices, what is your recommendation? Implementing a rather complicated streamdevice protocol file or developing an (quite simple) asyn driver? 0 upvote | 0 answer | 0 reply [08:00 PM] Gang Li asked : what kind of motor controller would you recommand, if want to control beamline step motor(two phase or five phase)? 0 upvote | 0 answer | 0 reply [08:02 PM] Dirk Zimoch asked : Gang Li, 5 phase motors are difficult. Try to avoid them 0 upvote | 0 answer | 0 reply