[HN Gopher] How-to guide: Commissioning a Sensor Physics R&D Lab
       ___________________________________________________________________
        
       How-to guide: Commissioning a Sensor Physics R&D Lab
        
       Author : MagneLauritzen
       Score  : 15 points
       Date   : 2026-03-30 11:19 UTC (2 days ago)
        
 (HTM) web link (gist.github.com)
 (TXT) w3m dump (gist.github.com)
        
       | MagneLauritzen wrote:
       | A practitioner's guide based on my own experience setting up and
       | operating sensor development and characterisation labs at CERN,
       | LBNL, DESY, and the University of Bergen.
        
         | fuzzfactor wrote:
         | Good writeup that can be useful to many :)
         | 
         | For decades I've built labs primarily for chemical
         | instrumentation, sometimes with electronic capabilities in
         | excess of other chemical labs. This article emphasizes success
         | in building dedicated electronic labs, sometimes having
         | outstanding chemical abilities as an option as well.
         | 
         | There are a number of things in common either way
         | 
         | With the gas lines, don't trust contractors, do as much of the
         | design, build, inspect, and verify your own self. This is
         | _usually_ an embarrassing failure otherwise.
         | 
         | Use Swagelok (top-shelf, having polished threads) brass
         | fittings with straight hardened copper tubing, available in 20
         | foot lengths, manually cut to size and hand-dressed to
         | outperform the near-perfect output of motorized abrasive cutoff
         | alternatives. Don't make me laugh if you think a plumber's
         | tubing cutter is appropriate, that's a desperate tool for when
         | you can not physically remove particles from an emergency
         | modification of an established system.
         | 
         | https://www.swagelok.com/downloads/webcatalogs/EN/MS-01-107....
         | 
         | Your specialty gases need to have the lengths of tubing
         | manually cleaned with high-purity solvents or steam anyway
         | before installation, and that washes away the particles
         | generated during the cutting process.
         | 
         | Stainless steel tubing & fittings work too and can have an
         | impressive look, but the exact same extreme effort will not
         | result in as leak-free of a system than copper & brass, which
         | makes a big difference with light gases like helium or hydrogen
         | which are escape artists. For utility vacuum (or aggressive
         | gases which might not be best centralized anyway), stainless
         | might be needed if corrosives are being sucked in for
         | centralized trapping, but I never would consider anything less
         | than a dedicated local vacuum pump(s) for high vacuum work on
         | an individual basis.
         | 
         | A vacuum line may benefit from larger diameter tubing, but one
         | of the good ways of doing it for the gases is to use 0.25 inch
         | diameter tubing rated to handle thousands of PSI. Even if you
         | are starting out with a single gas cylinder/regulator in the
         | supply area and delivering the typical 125 psi to the benchtop
         | locations. With a number of further downstream workstations in
         | use simultaneously someday, 0.25 tubing will not be able to
         | deliver enough gas through that small diameter at 125 psi. But
         | if everything is done correctly, conversion can be made later,
         | to run the same gas distribution lines at higher pressures with
         | appropriate new central tank regulators, or even at full tank
         | pressure. Then using an additional local gas regulator at each
         | workstation. Alternatively upsizing the tubing diameter by one
         | notch would then require more wall thickness and then it
         | becomes a real PITA compared to 0.25 inch.
         | 
         | Therefore the completed specialty gas delivery lines should
         | ideally be leak tested over a period of days using full tank
         | pressure of about 2500 psi. And under these conditions you
         | should be able to develop the proper engineering technique for
         | assembling the brass-to-copper so that it almost always shows
         | no detectable leaks every time.
         | 
         | You will also need a flow damper on each main line to limit or
         | shut down in case of major full-pressure uncontrolled escape
         | during an unplanned lab incident. You don't want liters per
         | second of flammable or asphyxiant gases to be able to pour into
         | the lab on top of any other undesirable occurrences. Not even
         | oxygen, which has its own specialized hazards including self-
         | ignition of trace combustibles in the tubing by surprise, like
         | oily residue or even fine metal particles. Oxygen has been
         | known to behave safely for years before an unusual pressure
         | shock occurs and ignites an internal hazard that was lurking
         | overlooked the whole time.
         | 
         | Linux is still useless for most chemical instruments, and
         | Windows has not been suitable for the internet for years, so I
         | need a completely air-gapped LAN of the local lab PCs, each of
         | which can be recovered to its original "immutable" condition
         | from backup, without anything ever having been connected to the
         | web. Mainly for productivity, but safety is also a major
         | consideration for all corporate IT to be barred from labs.
         | They're good at what they do, but their hands are more than
         | full with office machines where it's still a lot uglier that it
         | should be for all the work they are putting in.
         | 
         | It's a lot easier to train natural scientists to do everything
         | with PC's that their particular lab is going to need, than it
         | is to train corporate IT operators to do an outstanding job of
         | anything a unique lab needs.
        
       ___________________________________________________________________
       (page generated 2026-04-01 23:00 UTC)