Software Readout Of Optical Powers

It is important to have measured optical powers in the firmware (readout by software) for the following reasons:

• they can't easily be measured everywhere (e.g. with multi-fiber breakout cables, and behavior of transceivers)
• measuring with meters is currently a time-consuming manual process which one wants to avoid at 2am
• manual measurement means plugging/unplugging fibers in often dense-fiber conditions (e.g. at a patch panel) which can create further problems
• our fiber installation is become more complex with LCLS2 (more fibers, more delicate single-mode, longer runs, more patch panels)
• would allow for early detection of marginal signals
• would potentially allow for automated checks in future

Single-mode Fiber Cleaning Study

Done by Riccardo Melchiorri and cpo on Dec. 16, 2022 with the FEE alcove teststand

Setup:

• plugged 9 xpm outputs into each of 4 tdet nodes
• all the tdet nodes had new fibers and new transceivers
• the xpm transceivers were old and left uncovered for a long time
• found two transceivers that were create LinkRxErr's in xpmpva at a low rate
• one case was fixed by cleaning the receive diode, the other case was fixed by cleaning the transmit laser output
• A Neoclean-E LC/MU was used to clean the connections (https://www.fs.com/products/39721.html)
• we used a fluke fi 500 fiber inspector to look for dirt in the transceivers

Conclusions: 

• the fiber inspector camera worked well for fiber ends
• the fiber inspector camera worked well (showed dirt) for the transmitting side of a transceiver, whether or not the transceiver was plugged in. needed to be manually "pointed" at the transmitter
• the fiber inspector camera did not show dirt on the receiving side (diode?) of the transceiver (and needed to be manually focused).  but there is strong evidence dirt was present because cleaning the receiving side fixed the LinkRxErr's
• we empirically observed that LinkRxErr's can be produced by both transmit and received dirt (by cleaning only one side at a time)

in future, if dirt is found to be on the transmit side (e.g. using the fiber inspector camera) then we should measure the light received on the far side to see if LinkRxErr's can happen even with high light levels

A picture of the receiving side (diode?) of a transceiver after manual focus (no dirt evident, although dirt was empirically present since cleaning fixed the issue):

Picture of the transmitting side with dirt visible.  The fiber core is visible on the lower right (looks similar to a non-transceiver fiber to our eyes):

This shows which side of the transceiver is transmitting:

Philip Hart Suggested Approaches:

• Obtain and disseminate more complete test kits (this would address the measuring MTP problem)
• we keep spare octopus etc. cables on hand for swapping out the various fiber segments
  
• we keep tested transceivers on hand
  • we intend? would like? to qualify all transceivers when receiving cameras
  • maybe we need a regular program to replace or retest old transceivers
    
• sometimes we have dry compressed air on hand to clean interfaces
  
• write down the various debugging steps somewhere easily locatable
  • I forget where we wrote down our previous flowchart, Kaz
• It would be great if we planned for dying fibers and had system redundancy allowing one to hot reconfigure different lanes
  • of course we would want to have tested spares strung for this
  • having by-hand-only-swappable tested fiber bundles might be resource intense at the patch level etc. but I think it would be worthwhile in many cases