To: E614 Collaboration
Frome: Robert Henderson (TRIUMF)
Re: Results of last weeks meeting to discuss Vladimir cradle concerns
Date: 27 June 00


Hello Folks,

  I will try to briefly summarize the results of last weeks meeting to
discuss Vladimir cradle concerns. Nate suggested I address each of the
items
on his list in order.

> Hi Robert,
>
>   Maher tells me that you will be posting the minutes of the meetings
> last week regarding the cradle design.
>   I'm hoping that your minutes will deal with each of the items on the
my
> version of the list of questions - and correct my answers to some of
the
> questions where necessary.  Perhaps you could go through in numerical
> order to help people sort the topics.
>   I'd prefer that we not take the time to go through the discussion
over
> again, and if these responses are posted we won't need to go through
> them at the meeting this week.
>
> regards
> nate

  Just before I get into this, let me just mention non-magnetic
cylinders. As
I emtnioned in last weeks meeting, designing and making our own
cylinders
out of strictly non-magnetic materials is NOT a big deal. Now Roy Moore
has
found a local (Vancouver) company that will probably make them for us,
which
is even better. Roy has already forwarded my general specs to them (in
text).
I have Hans Wallitzer making a generic drawing indicating our
requirments,
then Roy will give them that and we'll see what they say. Looks good so
far.

  Back to Cradle questions. Here goes, my comments will appear after
"RoH"

Regards Robert Henderson

          --------------------------------------------------

1) It is not clear how the modules will be aligned with respect to the
horizontal and vertical directions.

The modules will not be aligned with respect to one another in the
transverse
directions. We will rely upon calibrations to determine the relative
wire
positions in (x,y).

RoH: We have discussed this several times in meetings. The Z position of
each
===  wire-plane is supposedly determined by the citals, that is why we
have
     them. The wires will be mapped relative to each other before
assembly.
     This leaves a possible overall offset and angle error for each
     wire-plane to be determined by tracking residuals. So that each
plane is
     `corrected' to a `true line', we plan to have at least 4 `surveyed'
(UV)
     modules.

     During the meeting Vladimir suggested we try to do corrections for
each
     wire using cosmics. The general concensus was that this would be
very
     difficult.

2) It is not clear how the stack can be aligned with respect to the
magnetic
field.

The chambers will be aligned with respect to the magnet by adjustment of
the
track. (Robert: how much range of motion is available for adjustment of
the
track? Left/right? Up/down?) It seems to us that we need adjustability
of the
order of 1 mR. Thi s is mm scale, but can we get +/- 10mm adjustments?
How
fine is the pitch of the adjusting screws? If we are to adjust to 1mR,
then
we need to be able to make adjustments of roughly 100 microns. Can we
make
adjustments of 100 microns laterally on the cradle? Note that we will
see
misalignments to tens of microns, so adjustability to 100 microns would
leave
a visible effect.

We need to know the range of the track adjustability, and the pitch of
the
associated threads.

RoH: See response next item
===

3) Vladimir points out that we will be sensitive to the alignment of the

stack with respect to the field at the level of 0.1 mR. The study of
this
alignment requires a stability of the alignment over 24 hours. Do we
have
this stability? Can we get sufficient alignment data in 24 hours?

As pointed out at the end of comment 2, we may not align the chambers to
the
level of their resolution. Nonetheless, we do require stability of the
alignment at the level of 0.1 mR. Indeed, we do require this stability
over a
running period of a month or of months. Do we have stability of the
stack
with respect to magnet at the level of 30 microns? If not, we dither the

resolution of the data obtained.

(note: while we are sensitive to the alignment of the track with respect
to
the field at the level of 10's of microns => ~ 0.1 mr, we are sensitive
to
the orientation of the yoke to the beam at the level of 10 mr since
cos(14mr) = 0.9999).

What is the expected stability of the track? How much is the data
"dithered"
by movement of the cradle?

RoH: Mechanically we should get the small movements of the track okay.
Also
===  our CCD cameras should be able to determine how much we are moving
     the cradle. Stability should not be a issue and again our cameras
can
     watch for movements of the cradle during data taking.

     During the meeting we spent considerable time discussing how we
plan to
     align the stack with the magnetic field, even though this doesn't
really
     effect cradle design. There are basically two ways of aligning the
stack to the field:

(1) Look at the positrons spiraling out from the target and plot the
centers
    of the `loops' as a function of Z, they shouldn't change. The
concern I
    have with this method is that the positrons are multiple scattering
asnd
    losing energy. Also we must correct for our field map, since the
spirals
    cover considerable volume. I wonder if we can get the centroids
    accurately enough?

(2) We could use the incoming muons. No matter what their incoming
direction,
    they spiral very tightly around the magnetic fiels lines, hense they

    `paint' the field directly and in the small central region where we
have
    the greatest concern. Of course, they multiple scatter even worse
than
    the positrons, so individual muon tracks are of no use. But I
suspect
    that if we look at the centroid distribution of many muon tracks for

    different planes (different Zs) covering say 25 cm, we might get a
very
    high accuracy determination of the field in this region.

  We discussed these approaches for awhile. No unanimous opinion was
forthcoming, though technique (2) had more support. We should study both
ways
in montecarlo to determine which is best.

4) Vladimir asks about limits to acceptable g-shocks to the cradle, and
consequent permanent changes in the alignment of the modules.

We cannot tolerate shocks to the cradle. We do not have the coefficient
of
friction for the citals, but we know that the coefficient was bigger
than
what Pierre could successfully measure. Certainly procedures must be
developed that ensure that shocks be limited to a fraction of g.

When we are moving the cradle to the cart, and moving the cart, what is
the
magnitude of shocks which we should expect?

RoH: Chambers can take a lot more G-forces than people suspect. Peopl
must
===  understand that teh chamber stack is a lot more compliant than the
     cradle. It will bent easily is the cradle flexes a little. Also,
during
     all movements of the cradle, the axial compresssion forces will be
     dropped to about 20 kg at each cital column. This should enable
each
     module to slide independently of its neighbor.

5) Vladimir reports coefficents of thermal expansion, and postulates
temperature variations of 10 degrees celsius. Will the modules slide
laterally on the ledge over a scale of 250 microns? Will they remain
against
the reference surface?

First, what precision of thermal control can we obtain, and how? This
requires discussion and planning. I suggest that we are talking about
something like 23 - 28 Celsius. We should be able to control the
temperature
better than this, but we have not d one the planning to my knowledge. In
this
case, a variation of 10 degrees is an upper limit.

As concerns the axial direction, can we be certain that the chambers
will
slide freely against the lateral stop?

As concerns the vertical direction, we rely upon an applied normal force
to
provide friction to lock the modules together. Variations in the normal
force
would enable the modules to slip relative to one another.

Vladimir concludes that the expansion/contraction of the cradle will
reduce
the compression of the stack by as much as 70 microns. Maher tells me
that
the spring force of the citals is 372 microns / 20 Mpa = 160 microns /
ton of
force. Therefore, if th e length of the cradle relative to the stack
increases by 70 microns, we lose 0.5 tons of force. Do the proposed
hyrdraulics take up this variation in the required applied force?

So... how do we control the temperature at the stack? What temperature
variation can we expect? Who will look after the stabilization of the
temperature?

How can we be certain that the stack will slide laterally without
binding?

RoH: This item took most of the meeting. There are two directions of
===  differential thermal expansion to consider. Z-direction and
     XY-directions:

Z-direction: The weight of each module rests on its two feet which rest
-----------  on the rails inside the cradle. The frictional forces are a

             fraction of this weight, about 0.15 for delrin on polished
brass
             and less if its lubricated. Each G10 gasbox will feel these

             residual frictonal forces. They try to distort the box, NOT
the
             wire-planes. The axial forces on the cital columns limit
the
             distortion of the box, but of course they will also distort

             minutely. I have no data, but I suspect this will NOT be a
             problem. This was alos the concesus of the meeting.


XY-directions: This is a bigger worry, since we're now dealing with the
-------------  considerable axial compression forces. Specifically, when
the
               axial forces are applied, the thick US G10 plate is
               friction-locked to the bumpers. With each column having
an
               axial force of 200 kg, each bumper might have a
frictional
               force of 50 kg. So, if the temperature should drop, the
cradle
               will contract more than the G10 (or glass) and try to
`pull
               the thick G10 (and the connected module) down' against
the
               internal rails with a LARGE force (say 200 kg).

               This sounds bad and of course a similar situation occurs
at
               the DS end of the cradle. Nearer the middle of the cradle
the
               compliance of the stack soon solves this problem.

               The situation outline above is NOT really true,
especially at
               the DS end. The bumpers are NOT fixed at the cradle
endcap,
               instead they continue a considerable distance to the
pneumatic
               cylinders. In the present design, the bumper is `fixed'
at the
               cylinder piston, about 11 inches from the bumper/G10
comtact
               point. If we ensure the pumber is not touching the metal
hole
               in cradle endcape, it is free to flex in the XY
directions. I
               asked Jan to calculate how flexible a 1" dia. Al rod
would be,
               here is his response:

> From: SMTP%"jsou@Phys.UAlberta.CA" 23-JUN-2000 16:09:20.41
> To: RHEND
> CC:
> Subj: RE: Details.
>
>   Hi Robert,
> I checked for the force required to bend a 1"dia. Aluminum rod,
implanted
> fixed at one end, by 170microns at the free end. It turned out to be
only
> 5.87 lbs. So you should be O.K. no matter what.
> ....
>                                      Cheers, Jan

                As indicated, a 1" bumper rod deflects quite easily in
the
                XY directions. Vladimir mentions movement of 70 um,
Jan's
                email suggests this would require only about 1.1 kg,
very
                small compared with the 50 kg friction forces available.
This
                made it clear to us that the same sitaution should be
setup
                at the US end of the cradle. This is easily done by
extending
                the US bumpers, making the fixture points 11" US of the
                bumper/G10 contact point. Hans is incorporating this
change
                into the cradle drawings.


6) Vladimir poses a number of questions which are related to the problem
of
the distribution of forces upon four citals given that three points
define a
plane.

It is not clear how we can be certain that each module is coplanar with
its
neighbors, given four-point contact. Demanding good optical contact at
all
four citals may not be the "right" answer.

Regarding vertical support of the cital columns, the argument that has
been
made in the past is that the stacks are going to be flexible in the
transverse direction, so they will deform until they are supported (at
several points) along their length. Not e: (a) This does NOT guarantee
that
every module will be supported. (b) I think this implies that the
transverse
alignment of the modules globally will be determined by the shape (and
changes thereof) of the cradle, so Vladimir is quite correct that it
needs to
be very stable.

By "globally", I mean that a single module won't move substantially with

respect to its nearest neighbors, but the overall (smoothed) shape could

change if the center of the cradle changes height with respect to the
ends.

How much variation in the coplanarity of the modules can we expect,
working
in the real world of three point support? Suppose we fix the (x,y) wire
positions using beam. Can we subsequently use cosmics at an angle to fix
the
z-positions of the planes (assuming that wires within a plane actually
lie
in a plane)?

RoH   The glass plates and gas boxes are quite flexible for small
deflections
===   in the XY plane. So when the axial forces are applied ALL of them
will
      try to assume the shape of the thick US G10 piece! This G10 piece
is
      supported at FOUR points and I was the one that initially pointed
out
      my concern at this. The best we can do is bring all four bumpers
in
      contact `by feel', then lock them. Also making the G10 as thick as

      possible (4 inches in my design). Note that smaller axial forces
      minimize deflections. We could think of trying to do an optical
setup
      to check that this G10 plate is not bending too much when the
axial
      forces are applied. This would be difficult, expensive and time
      consuming. Note also, if we make the axial forces too large, the
US
      cradle endcap will eventually start to bend, another reason for
not
      using too large and axial compression force.

      Inspite of our best efforts, the thick G10 will undoubtedly flex
to
      some degree. What is tolerable? Hard to say. Certainly the
chambers
      will still operate with a `saddle' shaped distorion of the frames
of up
      to say 200 um. For the pysics? Who knows, this is a montecarlo
      question. I'd guess a flex of 50 um or less would be okay. With
four
      points, there are only a limited number of saddle modes and the
      amplitude can of coruse vary. I think is we search in chisq for
these
      modes and amplitudes we can probably identify and correct if it is

      large enough to be significant.

Created for the The Center for Subatomic Research E614 Project Projects Page.

Created by The CoCoBoard.