Details
-
Type:
Improvement
-
Status: To Do
-
Resolution: Unresolved
-
Fix Version/s: None
-
Component/s: ts_main_telescope
-
Labels:
-
Story Points:2
-
Sprint:TSSW Sprint - Sep 13 - Sep 27, TSSW Sprint - Sep 27 - Oct 11
-
Team:Telescope and Site
Description
Add FA pressure reading from primary and secondary cylinders. The readings are returned in ILC function 119. The current code reads response, but don't store it nor broadcast it in telemetry (ILC/ILCResponseParser.cpp::_parseReadDCAPressureValuesResponse). XML schema needs to be extended with pressure reading (which was removed in commit https://github.com/lsst-ts/ts_xml/commit/56e3034bd836b47f3afdf76fa6824723dc66d5ab#diff-e3f4298d14aa57bcbecec0bd668322a5155cf1004cef634f29d77bf453105a8f).
Pressure values shall be broadcasted in 5Hz. A strategy to read out 1/10th of FA every ~53Hz outer loop shall be developed.
Attachments
Issue Links
- is blocked by
-
DM-32228 Resolve problems with pressure readout
- To Do
Activity
The problem is most likely in ILC firmware. I will create a ticket, which will describe the problem and will block this DM issue.
Petr Kubanek
added a comment - The problem is most likely in ILC firmware. I will create a ticket, which will describe the problem and will block this DM issue.
We need to decide whether or not the pressure measurement of each individual force ILC is necessary for any global control function or just a nice to have (AKA telemetry).
As I remember, the reporting of the pressure in the cylinders was only intended for valve tuning on the bench, not for normal operations, not was never tested for that with the whole array of ILCs. Pressure reporting was only used with a NI development kit to communicate with Matlab.
Pressure measurement was a very late addition to the pneumatic ILCs that required developing a mezzanine board to read the digital pressure sensors to operate the flow control valves. This was done at a time when we didn't have the conditions to perform any thorough testing of the entire system.
Because of the sensitivity of the optics to heat, the microcontrollers used in the ILCs are extremely low power, thus their processing power is very limited yet enough to accomplish the original mission with room to spare. Indeed, hardware and software functionality were increased over and over, so maybe we finally hit a physical limit.
Pressure monitoring in the system is accomplished by the hardpoint ILCs, which is all we need for pressure health assessment, so if this works properly then we have accomplished our goal.
As shown in our own SPIE paper, cylinder pressure is not a representative parameter because of slip and stick of the internal seal, so I don't see the point in broadcasting it. This decision was made at the early stages of development of the control system (all of them) where instantaneous outputs of the different control algorithms wouldn't be stored anywhere.
Bottom line: force ILC cylinder pressure measurement was only intended for internal use of each ILC and broadcast in normal operation was never a requirement nor was tested for that.
Oliver Wiecha
added a comment - We need to decide whether or not the pressure measurement of each individual force ILC is necessary for any global control function or just a nice to have (AKA telemetry).
As I remember, the reporting of the pressure in the cylinders was only intended for valve tuning on the bench, not for normal operations, not was never tested for that with the whole array of ILCs. Pressure reporting was only used with a NI development kit to communicate with Matlab.
Pressure measurement was a very late addition to the pneumatic ILCs that required developing a mezzanine board to read the digital pressure sensors to operate the flow control valves. This was done at a time when we didn't have the conditions to perform any thorough testing of the entire system.
Because of the sensitivity of the optics to heat, the microcontrollers used in the ILCs are extremely low power, thus their processing power is very limited yet enough to accomplish the original mission with room to spare. Indeed, hardware and software functionality were increased over and over, so maybe we finally hit a physical limit.
Pressure monitoring in the system is accomplished by the hardpoint ILCs, which is all we need for pressure health assessment, so if this works properly then we have accomplished our goal.
As shown in our own SPIE paper, cylinder pressure is not a representative parameter because of slip and stick of the internal seal, so I don't see the point in broadcasting it. This decision was made at the early stages of development of the control system (all of them) where instantaneous outputs of the different control algorithms wouldn't be stored anywhere.
Bottom line: force ILC cylinder pressure measurement was only intended for internal use of each ILC and broadcast in normal operation was never a requirement nor was tested for that. The timeout problem might be related to wrong cabling, similar to the timeouts we saw on HP #6 (https://jira.lsstcorp.org/browse/FRACAS-89).
Petr Kubanek
added a comment - The timeout problem might be related to wrong cabling, similar to the timeouts we saw on HP #6 ( https://jira.lsstcorp.org/browse/FRACAS-89 ). 
See screencast. Pressure readouts is timeouting on random FAs. Looking for cause..