Re: Suggestion for routing the Bcal and Fcal cables

[Tim Whitlatch <whitey@jlab.org>]

Good morning Elton,

We can start to look at this from the mechanical layout point of view.

Cheers,

       Tim

Fernando J. Barbosa wrote:
> 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
>>   
>
begin:vcard
fn:Timothy Whitlatch, PE
n:Whitlatch;Timothy
tel;work:757-269-5087
tel;fax:757-269-5279
version:2.1
end:vcard