Confluence will be unusable 23-July-2024 at 06:00 due to a Crowd upgrade.
Press 8/8
Closed for editing, 8/4
We had a few fixes and feature updates for the PMPS UI diagnostic tool in July.
The following environments were released during May, June, and July:
pcds-5.4.2 is planned to be released shortly, at the end of July or at the start of August. This is planned to be the first update with "gentler" dependency updates to minimize the potential for picking up unexpected behavior on update.
Note that any applications using pcds-5.3.1 should update if they plan to use hutch-python in an experiment setting. There is a bug in this version that can lead to dangerous results where the history of separate ipython sessions will get mixed during execution, so someone else's "move" command can end up in your history and it is very very easy to accidentally run their command instead of re-running yours.
We've been working on tracking down the reasons why various Python apps are slow to load and trying to minimize the parts we can control. To this end, we've already found a bunch of speed ups and work is ongoing. Some of these are tricky because, while startup speed is important, having a design tradeoff that adds slowness later can be just as bad.
A project page is opened here and work is ongoing: Hutch Python/Lightpath Device Loading Slowdown Findings/Fixes
Margaret Ghaly Vincent Esposito
We've been working to rebuild the lightpath application, a tool that aims to give a high-level summary of the beam, where it's pointing, and which devices are blocking. In order to properly represent the facility, significant changes were made both how lightpath organizes devices and how those devices are represented. As of the writing of this newsletter, we have completed the major infrastructural changes to lightpath, implemented a new device interface, and begun to spot check the app's performance for select end stations.
Design details and FAQ's are being gathered at this page, which (like the lightpath app) is a work in progress.
Mirror: Waters, Nick Vacuum: Jing Yin Tong Ju Motion: Maarten Thomas-Bosum Zachary L Lentz PMPS: Margaret Ghaly Tong Ju Image: Tong Ju Govednik, Janez
Scientists will run dream mirror check out experiment and test all PMPS and veto groups at the same time
TMO had a strange problem back in April: whenever they would move IM4K4 in or out, it would vibrate the stand violently to the point that it would cause a nearby turbo pump to crash, which was extremely disruptive to endstation operations.
Stepper motors are known for having some problems with resonance and high vibrations. Typically in industry you would pick servo motors if extremely smooth motion was a requirement. This isn't a requirement here, but "do not shake the stand when moving" is absolutely a reasonable ask.
All of the PPM imager/power meter combination units were tuned based on a single instance before it was installed onto the beamline. As such, the parameters were optimized for a lab bench in the temporary clean room in B750 rather than for the beamline as-installed. With the rush to close out the installation of the L2SI components, it looks like this had never been revisited.
In general, we expect every instance of a motor to have slightly different tuning needs based on the precise specifications by which the assembly was installed. This won't always be worked on to completion because often there won't be any problems, and because the most important specification for the majority of our motors is position accuracy and precision of the end position, along with motion reliability, which is not covered by this vibration case.
It was observed that the PPM units have higher vibrations than one would expect in the FEE, RIX, and TMO, culminating in this turbo pump incident. So, we took some time to do the following for every installed/active PPM:
After these adjustments, the PPMs are silky smooth, quiet, and reach their destinations reasonably quickly.
It should be noted a huge quantity of our work is done on Github.com, all development is tracked there. Jira issues capture a significant body of work as well, but at least as much work is also captured in the closure of Github tickets (issues) associated with our various codebases. Unlike Jira, getting a consolidated metric of work done in a past period is not possible without a paid subscription to Github. Roughly speaking over 80 projects were touched since April 8th, with multiple changes of various sizes.
Getting issues...