From steve@hep.man.ac.uk Wed May 26 13:24:12 2004
Date: Mon, 17 May 2004 11:28:44 +0100 (BST)
From: steve <steve@hep.man.ac.uk>
To: Richard Bates <r.bates@physics.gla.ac.uk>
Cc: Steve Snow <steve@mail.hep.man.ac.uk>,
     Didier Ferrere <Didier.Ferrere@cern.ch>, Bettina.Mikulec@cern.ch
Subject: Re: Spine QA at manchester and geneva


Hi Richard,
           For each spine we do a mechanical measurement on the
smartscope, a visual inspection and 3 resistance measurements.
Details are on our web page at 
http://www.hep.man.ac.uk/atlas/module/production.html
look at the 5th bullet under 'Documents' and the spreadsheet 
under 'Spines QA'


The smartscope part takes 5 minutes per spine. It just gives
the length, the longitudinal bow and the width of the slot. We used to 
measure several other features like tilt, twist, transverse bow, etc but 
there turned out not to be correlated with the quality of the final
module so we have dropped them.

The visual inspection takes 10-20 minutes, depending on whether
we find any loose TPG, dust,  stray drops of glue etc, that need to be
cleaned off. The electrical measurements are done at the same time 
and take about 1 minute.

There is a handling time of about 2 minutes for getting the spine in and 
out of it's box and this applies to each of the two steps above.
So the total is 19-29 minutes per spine. This is quite acceptable to us.
The best way to save time is probably to make sure that very few spines
fail at a late stage in their QA.


To deal with each point in turn:
* length is a duplicate of measurement done at cern but it costs almost
no time so I suggest we carry on doing it. 
* bow is measured differently from the check done at cern so I plan to
keep it
* slot width is really a check for damage to the washer and it is hard to
quantify any damage without using the smartscope so I plan to keep it.
* we do not measure spine or spacer thickness. We rely entirely on the
cern measurement wo it is important that you carry on doing that.
* the resistance measurements are quick and simple to do. They do catch a
few faults. It would help if these could be done at cern.
* the visual inspection is the time consuming part. I would not want to
drop it, but it could become quicker if the quality of spines is higher.

So suggestions are

* reduce length measurements at cern to a sampling, eg 20% of spines,
  if you think this is safe.
* add 3 resistance measurements to your cern QA and we then drop them.
* make sure the spine boxes are really clean. This would save time
  cleaning dust that came from the box.
* make sure the sticky tape used to close the boxes is easy to
  open and re-seal. This would save some handling time.
* make sure that obvious faults like cracked ceramic or missing glue
  coverage at the cooling contact area are caught by inspection at cern.


cheers, Steve.
 
On Wed, 5 May 2004, Richard Bates wrote:

> 
> Hello Steve and Didier,
> 
> I am aware that you spend some time inspecting the spines before using
> them. Can you tell me what meaurements you do and how much of your time
> this takes up? We do make measurements of the spines both at IHEP
> and at CERN. It would be nice to see if any of the tests are unnessary
> duplications of effort, with associated time. I understand that some of
> the measurements that you perform may well have been introduced when they
> were not preformed by the spine producer, and it would be good to know if
> you see significant differences in your measurements and the measurements
> that are on the spine check sheet.
> 
> Another question to didier: how easy would it be to add the spine check
> sheet fields to the database? at present we have all the numebrs in an
> excel spreedsheet and entering them into the database should not be too
> much hard work, if there is a sensible place to put them.
> 
> 
> thanks
> 
> yours
> 
> Richard
> 
> 
>