[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Suggestion for routing the Bcal and Fcal cables
Hall D Electronics:
Hi Elton,
We have a convention for labeling crates and racks. Labeling in the hall
has been established to be Upstream (U), Downstream (D), North (N) and
South (S). The rack locations are labeled in GlueX-doc-747 or my recent
talk for the system review. Presently, we have 8 racks reserved for the
BCAL and 4 racks for the FCAL, instrumentation only. There are
additional racks for HV, LV etc. The BCAL has racks 1, 2, 5 and 6 on the
N1 platform (North) and 1, 2, 5 and 6 on the S1 platform (South); the
FCAL has racks 2, 3, 8 and 9 on the D2 platform (Downstream), for
instrumentation only.
The labeling convention is Xa-b-c-d where X is the platform location (U,
D, N, S), a is the platform level (1 or 2), b is the rack number (see
doc-747, or left-to-right and clockwise), c is the crate number (3
crates in a rack: 1 is top, 2 is middle, 3 is bottom) and d is slot
number in crate (left-to-right). For example, N1-2-3-4 would locate a
module in the North platform, level 1, rack 2 (second from the left),
crate 3 (bottom) and slot 4.
I had this information on earlier versions of doc-747 but I removed it
because it was too much information for a document like that. I will put
this information on a new document sometime tomorrow.
Regarding the cabling in view of the trigger, I agree it is a good idea
to group sections per their sum function. For the BCAL, I presume we are
dividing the detector in two, North and South with the same detector
elements (upstream and downstream) going to the same fADC. In the rack
location drawings, we can locate the instrumentation racks on locations
2, 3, 4 and 5 (instead of 1, 2, 5 and 6). The cable lengths could be
made the same (the distance from rack 2 to rack 6 is about 8 feet or
less than 3 m).
Your suggestion for the FCAL might be a bit messy, as you mention. A
detailed drawing will be required to check the cable lenghts, supporting
structures and cable trays.
With regards to the trigger formation, the optical links from the CTPs
to the SSPs don't need to have a fixed latency but it would be easier to
have one fixed length for each sub-system. Because each SSP accepts 8
inputs from crates (CTPs) in the same sub-system, performing the
boundary energy sums with those inputs should be easy. As we have 6 CTPs
for the N1 and 6 CTPs for the S1 platforms, according to the scheme you
suggest, we could do the boundary Esum with the two unused inputs on
each SSP of the BCAL. This implies a splitter which can be perhaps be
added to each CTP or external splitter. So, I don't see a major obstacle
to this implementation.
Regards,
Fernando
Elton Smith wrote:
> Dear collaborators,
>
> During preparations for the system review we had some discussions about
> cabling since how the cables are connected to the electronics can affect
> the options for creating energy sums for the trigger. The engineering
> staff is also starting to route cables and establish cable trays, so it is
> not to early to consider their layout. Below are two suggestions for
> discussion about how to map the Bcal and Fcal cables to the crates. These
> cabling schemes are motivated by considerations for the trigger. Generally
> a phi-symmetric trigger is desirable for our detector.
>
> Comments/suggestions are welcome.
>
> Cabling for the Bcal.
> --------------------
>
> General: For the trigger we want to cable opposite (left/right) ends into
> the same fADC module so that sums will approximately compensate for
> attenuation. Adjacent modules should be cabled together to create local
> sums of energy deposition.
>
> The crate sums are delivered to the system processors and therefore are
> created naturally the trigger now, without the use of additional data
> lines. Therefore, it makes sense to cable the crates in a way that might
> be useful for the trigger.
>
> One cal wedge: Outer 4 x 2 (left/rigth) = 8 inputs
> Inner: 24 x 2 (left/right) = 48 inputs
> -> One wedge fills 3.5 fADC (16 channel) modules.
> -> Two wedges fill 7 fADCs
> -> Four wedges fill 1 crate (14 fADCs). This leaves 2 empty slots, which
> might be sensible from the heat dissipation stand-point, but also
> associates a single crate with 1/12 of the Bcal (30 degress in phi).
>
> With this scheme we need a total of 12 crates for the Bcal (6 north, 6
> south). Note that 10.5 crates would be used if all slots were cabled
> completely (16 fADCs per crate). However, the even number of 12 splits
> evenly between north and south which would be required by the present
> concept for the location of racks.
>
> I do not know if there is already a convention for numbering wedges or
> crates, but I assume it makes sense to tie them to the coordinate system,
> were y is up and x is pointing north, and phi clockwise relative to x.
>
> north crates phi (deg)
> 1 0-30
> 2 30-60
> 3 60-90
> 10 270-300
> 11 300-330
> 12 330-360
>
> south crates phi (deg)
> 4 90-120
> 5 120-150
> 6 150-180
> 7 180-210
> 8 210-240
> 9 240-270
>
> In order to ensure coverage across the boundary between crates, the
> trigger would need to create overlapping crate sums. These would include
> 1+2, 2+3, 3+4, 4+5, 5+6, 6+7, 7+8, 8+9, 9+10, 10+11, 11+12, 12+1. This
> granularity corresponds to an azimuthal angle range of 60 degrees. For
> example [I'm guessing here] a cosmic-ray trigger could be formed in the
> SSP using opposing regions such as 2+3 * 8+9 or 3+4 * 9+10 or 4+5 * 10+11.
> This concept should be checked by the trigger experts.
>
> Cabling for the Fcal.
> --------------------
>
> General: For the purpose of making coincidences between the Fcal and the
> FTOF, one might consider cabling the Fcal in either vertical or horizontal
> columns (rows). Background rates, however, are a very strong function of
> the distance from the beamline, and so it makes more sense to combine Fcal
> blocks within concentric rings into the same crate. This also keeps the
> azimuthal symmetry of the detector which I believe is a very desirable
> feature of the trigger.
>
> A rough division of rings (within two radii) is given below, where each
> block is 4x4=16cm2. Each crate is assumed to house 16 fADCs (256 blocks)
> For 11 crates this corresponds to 2816 blocks (nominal number is 2800).
> The last crate corresponds approximately to a single ring of blocks at the
> outer layer of the Fcal.
>
> crate area radius(cm)
> (cm2)
> 1 4096 36.1081486
> 2 8192 51.06463346
> 3 12288 62.54114794
> 4 16384 72.21629719
> 5 20480 80.7402748
> 6 24576 88.44653962
> 7 28672 95.53318149
> 8 32768 102.1292669
> 9 36864 108.3244458
> 10 40960 114.1839917
> 11 45056 119.7571808
>
> At the trigger level, different trigger thresholds or selection criteria
> could be easily applied at the crate level. These thresholds should be
> applied to everlapping sums to ensure that showers that straddle the
> boundaries have the same efficiency in the trigger as showers in the
> middle of a sum. This scheme with rings, requires the least number of
> elements in the sum which is two. [Note that cabling in for example
> quardrants will require four elements into the sum, since a shower could
> in principle hit at the corner of for regions].
>
> Energy sums which can be used for selection with a threshold would be the
> sum of two crates, i.e. 1, 1+2, 2+3, 3+4, 4+5, 5+6, 6+7, 7+8, 8+9, 9+10,
> 10+11, 11+12, 12. This granularity provides a reasonable granularity even
> for the simplest trigger scheme which is presently proposed.
>
> This scheme requires cables around an entire Fcal ring routed to a single
> crate. In particular this means that cabling would not be partitioned
> north and south which might be the most elegant cabling solution.
>
> -----------------
>
> Elton Smith
> Jefferson Lab MS 12H5
> 12000 Jefferson Ave
> Suite # 16
> Newport News, VA 23606
> elton@jlab.org
> (757) 269-7625
> (757) 269-6331 fax
>