Proposal for optical and cable network of NA62 trigger system rivda 1) , Cristina Lazzeroni 1) , Vlado Černý 2) , Tomáš Blažek 2) , Roman L 1) University of Birmingham, UK 2) Comenius University, Bratislava, Slovakia 13/7/2011 1
Proposal for optical and cable network of NA62 trigger system
Marián Krivda1) , Cristina Lazzeroni1) , Vlado Černý2), Tomáš Blažek2), Roman Lietava1)2)
1) University of Birmingham, UK2) Comenius University, Bratislava, Slovakia
13/7/2011 1
Content• General strategy (old)• General strategy (new)• Old layout, response from detector and new layout respecting also the new
general strategy for:– CEDAR– GTK– CHANTI– LAV– STRAW– RICH– CHOD– SAV (IRC + SAC)– LKR– MUV
• Summary table13/7/2011 2
General strategy (old)
• Blow optical fiber (+ 1 spare) per a rack with FEE => use an optical splitter to fan-out optical signal in the rack
• Use 1 spare laser per TTCex board• Use the optical splitter 1:4, 1:8, 1:16 as
necessary
13/7/2011 3
General strategy (new)
• Blow optical fiber – possiblity to blow only 6,12 or 24 fibers
• Use an optical splitter 1:4, 1:8, 1:16 as necessary to fan-out optical signal in the rack where are LTUs or in destination where is FEE of sub-detector
• Use 1 spare laser per TTCex board
13/7/2011 4
CEDAR (Cristina Lazzeroni)
• The CEDAR has 2 unknown at the moment : how many TEL62 we need,and where exactly they will be located.If we assume they are 2 TEL62, then your proposal is fine.Since the rack will be at the CEDAR place, it doesn't have a number yet,but it will.
13/7/2011 5
CEDAR (new) LTU + TTCex(2 lasers) + optical splitter 1:4
2 optical fibers + 1 LVDS cable
13/7/2011 6
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
CHOKE/ERROR
LVDS (7)
TEL62
TEL62
Fan-in
CHOKE/ERROR
CHOKE/ERROR TTCoc 1:4
spare
Rack EB20 Rack CEDAR-01
GTK (Angelo Cotta Ramusino)• First I should point out that the TTC links would be needed only where the
off-detector GTK readout electronics is located and that is in the electronics house, not in the pit.My baseline design proposal was foreseeing one TTC link per GTK ASIC and therefore:- 10 TTC links per GTK STATION; being there 3 stations this means 30 links total
• A different implementation is possible in which each station would receive just one TTC fiber, convert it and distribute the TTC information locally by an electrical network; in this case then only one TTC fiber per station would be needed. This second approach is trickier and so I would prefer that You foresee, at this moment, to run the 30 fibers to the GTK off-detector electronics, unless costs or other reasons would forbid it.
13/7/2011 7
GTK (new)LTU + TTCex(4 lasers) + 3 x optical splitter 1:16
6 optical fibers + 1 LVDS cable
13/7/2011 8
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7)
GTK1
GTK2
Fan-in
CHOKE/ERROR
CHOKE/ERROR
Farm roomrack S14 + rack S15 + rack S16
GTK2
TTCoc 1:166
Rack EB20
CHIANTI (Fabio Ambrosino)Dear Cristina, since we will use just one TEL62 the scheme looks fine for the CHANTI. The rack name as far as I understood is just CHANTI-01 in TTC8.
Just as a remainder, I recall you that due to a misunderstanding (I was not reacting promptly at the time the order was placed since I was taking care of my father in the hospital) I was ordering for the CHANTI much more lasers than actually needed (10 instead of 2 !!), so that 8 of them can now be used as spares for the collaboration or for the individual groups who did not order the spares. Of course as I already wrote to Ferdi few months ago, if this can be done by scaling part of their cost from Napoli quota it would be very welcome, but in case is not possible there are no problems.Best regards,Fabio
13/7/2011 9
CHANTI (new)LTU + TTCex(2 lasers)
2 optical fibers + 1 LVDS cable
13/7/2011 10
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7) TEL62
spare
CHOKE/ERROR
Rack EB20 Rack CHANTI-01
LAVIF we keep the rack 29 configuration:3 fiber + 3 spares
1:8 splitter
TTCex1+1 fiber
1+1 link
1:8 splitter
1+1 link
1:8 splitter
1+1 link
1+1 fiber
1+1 fiber
13/7/2011 11
The LAV positions (Mauro Raggi)In each of this position you will find a rack with 1TEL62 with a TTC receiver
Need in total 12 x 2 connections assuming that each station receive in its rack 1 fiber and 1 spare
13/7/2011 12
LAV - Computing attenuation
I assume that the 1:8 splitters are sitting in the racks of:ANTI-A1 one 1:8ANTI-A5 one 1:8ANTI-A9 one 1:8
If I understood correctly the constraints we should have a ~ 10db attenuation on each line
LAV1 122,0 d A1 Rack db att.LAV2 129,6 7,6 22,8 m 10,7LAV3 137,2 7,6 LAV4 144,8 7,6 LAV5 152,4 7,6 d A5 Rack db attLAV6 165,9 13,5 28,5 m 10,7LAV7 173,4 7,5 LAV8 180,9 7,5 LAV9 193,2 12,3 d A9 Rack db attLAV10 203,6 10,5 45,7 m 10,7LAV11 218,2 14,6 LAV12 238,8 20,6
13/7/2011 13
LAV - Material required(Mauro Raggi)
• One LTU+TTCex with 5 lasers• 3 1:8 optical splitters +1 spare• 6 optical fibers TTCex splitters• 24 connections splitters-TEL62
13/7/2011 14
STRAW (Peter Lichard)Hi Marian,we re-evaluated our system and we would like to check with you this proposal We would like to have 1 TTCex and 1 LTU underground controlling all the detector.We imagine TTCex having 1 laser + 1 spare laser. The distribution for 4 chambersshould be done with 1:4 optical splitter followed by 1:16 optical splitter for on-chamber modules.We would like to have also a second set of TTex+LTU for lab and eventual spare.From back-envelope calculation on attenuation this scheme seems to provide sufficient optical power.Please, could you check this scheme if it really works?TTCex -> 1:4 -> 1:16 If this proposal is OK for you, we need to have2 pcs TTCex with 2 lasers2 pcs LTU1 pcs 1:4 splitter4 pcs 1:16 splitter Thanks for discussion!cheersPeter13/7/2011 15
STRAW (new) LTU + TTCex(2 lasers) + TTCoc 1:4 + 4x optical splitter 1:16 + 6 optical fibers + 1 LVDS cable
13/7/2011 16
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7)
Module1
Module2
Fan-in
CHOKE/ERROR
CHOKE/ERROR
Rack STRAW1-01 + rack STRAW1-02Rack STRAW2-01 + rack STRAW2-02
TTCoc 1:16
Module3
Module4
TTCoc1:4
6
Rack EB20
RICH (Patrizia Cenci)Dear Marian, thank you for your message. Given the RICH needs, your proposal for the TTC cable network for the RICH should have a splitter 1:8 at the place of the splitter 1:4 suggested by you, which is not sufficient for us.Could you, please, update the summary table in your slide n. 15 according to that? Let me remind that we already agreed with Cristina, in november 2010, the purchase of 1 TTCex + 10 lasers for the RICH. Thank you so much. Best regards,Patrizia
13/7/2011 17
RICH (new) LTU + TTCex(2 lasers) + optical splitter 1:8
6 optical fibers + 1 LVDS cable
13/7/2011 18
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7)
CHOKE/ERROR
Rack RICH-01
TEL62CHOKE/ERROR
TEL62
TEL62
TEL62
TTCoc 1:8
Fan-in
Rack EB20
spare
CHOD (Patrizia Cenci)Dear Marian,
as far as I know, your proposal for the NA62 CHOD TTC cable network is correct, however I'd like to remind you that the CHOD in NA62 is not a Perugia INFN responsibility and we are funded by INFN only for the RICH (you'll know very soon about our requests for the RICH), according to the NA62 MOU.
At present, I'm taking care of making the old NA48 CHOD working at the 2012 Techical Run, by borrowing existing HV and TDAQ modules within the collaboration, at no costs (but some manpower). I hope that will also be possible in the case of the TTC cable network for the 2012 run. This possibility will be soon investigated.
As far as Perugia concerns, I cannot say more that that, I'm sorry: of course, Perugia will not be able to buy the TTC cable network proposed for the CHOD.
Thank you.Best regards,
Patrizia
p.s. the CHOD rack will be at the LAV12/IRC place, assuming that the final detector will be kept at the present position.
13/7/2011 19
CHOD (new) LTU + TTCex(2 lasers)
6 optical fibers + 1 LVDS cable
13/7/2011 20
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7) TEL62CHOKE/ERROR
Rack CHOD-01Rack EB20
spare
IRC (Venelin Kozhuharov)Dear Marian, I'm looking at your presentation which you have distributed for the optical and cabling network. Here is what I have understood as far (which might not be 100% correct):- The IRC/SAC system has ordered just 1 LTU and 1 TTCex with 3 lasers. Since the LTU is only one we need only 1 LVDS cable for trigger data.- The TTCex and the LTU for the IRC are located in the common clock crate.- The IRC RO will be located in a rack which is close to the rack for LAV12. Most probably the SAC RO will be located in the same rack (still not sure for that) since we aim to have common TELL62 for both detectors. Even if we use different TELL62s for the two detector (in case of limited network bandwidth for data transfer) 2 optical fibers are sufficient from the 2 lasers of the TTCex(we didn't choose the option to use a splitter).- An additional optical fiber would be of use in case of failure of one of the conventional fibers. It can be from the third laser of the TTCex. In case of two different locations of the RO then 1 additional fiber would be necessary for each of the ROs (for example from the thirds laser). So I would say that 4 fibers are completely OK with us and might be even more than enough.As a conclusion I think that your suggestion is fine for us.Few considerations:I suppose the distance from the TTCex to the IRC rack (i.e. LAV12 rack) or the probable SAC rack (which in any case won't be further than the end of ECN3) is not that much to require intermediate devices to recover the signal. Since we are going to connect the optical cables directly to the TELL62 we would need attenuators at the TTCex. I suppose one for each channel so three in total. Would you think that usage of a TTCoc would be a better idea and would have some advantage? Thanks and greetingsVenelin
13/7/2011 21
IRC/SAC (new) LTU + TTCex(3 lasers)
4 optical fibers + 1 LVDS cable
13/7/2011 22
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7)
IRC
SAC
Fan-in
CHOKE/ERROR
CHOKE/ERROR
Rack IRC-01Rack EB20
spare
13/7/2011 23
LKR readout (new)LTU + TTCex(2+2 lasers), 10 1x4 splitters
6 optical fibers + 1 LVDS cable
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7)
CREAM Crate 1
CHOKE/ERROR
Rack EB01 ….EB08
TTCoc 1:4
Fan-in
CREAM Crate 2
CREAM Crate 3
CREAM Crate 4
…..CREAM Crate
26CREAM Crate
27CREAM Crate
28
TTCoc 1:46
Rack EB20
LKr L0 (Riccardo Fantechi)• For the CREAM readout we will have 28 crates with 28 TTCrx in a trigger
board. Each crate will have an internal distribution of the clock info to the various (16/crate) CREAM boards. And all the boards will send back their choke/error signals to the trigger board, where they will be OR-ed and sent, from each crate, to a fan-in for the LTU.
• We had a preliminary discussion on a possible scheme and the outcome is the following: 2 (+2 spare) fibers go to the CREAM racks (1 at left, 1 at right) and there a 1 to 16 splitter will fan-out the signals to the 28 crates. This is one possible way which will require some care in laying down the) with a 1 to fibers to the individual crates. An alternative will be 8 fibers (1 per rack4 splitter each and a better layout. Add if necessary attenuators if needed.
13/7/2011 24
LKr L0 (new)
13/7/2011 25
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7)
Crate #1
Crate #2
CHOKE/ERROR
Crate #3
TTCoc 1:16
TTCoc 1:16
TTCoc 1:16
6
Rack EB20 Rack EB09
MUV (Rainer Wanke)Dear Cristina and Marian,I apologize for the late answer:The MUV1+2+3 will be located in one single rack in the electronic barrack with 2 to 4 TEL62 boards in total. Therefore, in principle only 2 fibers (1+1 spare) would be needed together with 1 1:4 optical splitter. However, since we would like to have the possibility to run each of MUV1, MUV2, and MUV3 in standalone mode (they are quite different from each other), I propose to have 3+3spare fibers for the MUV1+2+3. (I hereby assume, that it is possible to have two TEL62 boards with two different TTC signals in one crate, is this correct?)We should also still foresee 1 TTCoc 1:4 module, since the MUV3 will most probably use two TEL62 boards.Bottom line: 3+3 fibers + 1 TTCoc 1:4 for the MUVCheers, RainerI forgot:As everything is in one single rack, one spare optical fiber may be enough.
13/7/2011 26
MUV (new) LTU + TTCex(3 lasers)
4 optical fibers + 1 LVDS cable
13/7/2011 27
40 MHz clock
source
LTU TTCex
TTC ch ATTC ch B
Trig. data from L0proc
Burst
clock
clock
CHOKE/ERROR
LVDS (7)
MUV1 +2
MUV3spare
Fan-in
CHOKE/ERROR
CHOKE/ERROR
Rack MUV-01
spare
TTCoc 1:4
Rack EB20
MUV (comments)
• Ordered 1 LTU, 1 TTCex (1 x 10 = 10 lasers) –• Does MUV need 10 lasers ?• If MUV wants to run MUV1,2,3 independently
in stand-alone then it is necessary to have 3 LTUs + 3 TTCex boards
13/7/2011 28
Optical attenuation• TTCex laser has optical power -3dB.• Losses in optical fiber == 0.1 dB/100m• Losses on ST/PC connector == 0.5 dB• Losses in optical splitter 1:4==7dB, 1:8==10.2 dB,
1:16==12.5 dBThe maximum allowed losses are 30 dB. If it is more
than TTC receiver (TRR-1B43) doesn`t work. So, you can make any combination of optical fibers, connectors and optical splitters until you reach 30dB.
13/7/2011 29
Test of optical splitters
• Attenuation spread in all channels– 1:4– 1:8– 1:16
• Attenuation of 2 TTCoc in series– 1:4 + 1:16– 1:4 + 1:4
13/7/2011 30
13/7/2011 31
Electronics barrack
CLKL0TPLTU
TTCex
EB20MUV1+2+3
EB306
Farm room
GTK1
ST14
GTK2
ST15
6
GTK3
ST16
61 for CEDAR1 for CHANTI4 spare
CEDAR-01
CHANTI-01
LAV1
LAV1-01
LAV2
LAV2-01
LAV3
LAV3-01
LAV4
LAV4-01
LAV5
LAV5-01
LAV6
LAV6-01
LAV7
LAV7-01
LAV8
LAV8-01
LAV9
LAV9-01
LAV10
LAV10-01
LAV11
LAV11-01
LAV12
LAV12-01
STRAW
1
STRAW
2
6 1 for LAV (6,7,8,9)
2 for STAW (1,2) 3 spare
1 for LAV (12)1 for CHOD1 for IRC3 spare
6
STRAW
3
STRAW
4
CHANTI
CEDAR
CHOD IRC
1 for LAV(10,11)2 for STRAW (3,4)1 for RICH2 spare
6
61 for LAV(1,2,3,4,5)5 spare
RICH
LKrreadout
LKr readout
LKrL0
EB01EB08EB09
6
RICH-01
CHOD-01
IRC-01
STRAW1-01
STRAW2-01
STRAW3-01
STRAW4-01
Single mode optical fiber ST-ST connector
LVDS cable for CHOKE/ERROR(not jet specified)
Patch cord (single mode optical fiber ST-ST connector)
2CHODIRC
Summary (ECN3 setup)Num. of
LTUboards
Num. of TTCexBoards
Num. of lasers on
TTCex
Num. of TTCoc
1:4
Num. of TTCoc
1:8
Num. of TTCoc
1:16
Confirmedby
detector(email)
CEDAR 1 1 2 1 - - O.K.
GTK 1 1 3 - - 3 O.K.
CHANTI 1 1 10 - - - O.K.
LAV 1 1 10 - 4 - O.K.
STRAW 1 1 4 1 - 4 O.K.
RICH 1 1 10 - 1 - O.K.
CHOD 1 1 2 - - - O.K.
IRC/SAC 1 1 10 - - - O.K.
LKr readout 1 1 2 10 - - O.K.
LKr/L0 1 1 6 - - 3 O.K.
MUV 1 1 10 1 - - O.K.
TDAQ 1 1 4 - - - O.K.
13/7/2011 32
LTU + TTCex order• CEDAR: 1 LTU, 1 TTCex (1 x 2 = 2 lasers)• GTK: 3 LTUs, 3 TTCex (3 x 3 = 9 lasers) -> NO spare laser on TTCex !• CHANTI: 1 LTU, 1 TTCex (1 x 10 = 10 lasers) -> They need only 2 lasers !• LAV: 2 LTUs, 2 TTCex (1x10 + 1x4 = 14 lasers) • STRAW: 2 LTUs, 2 TTCex (2 x 4 = 8 lasers)• RICH: 1 LTU, 1 TTCex (1 x 10 = 10 lasers) -> Do they need 10 lasers ?• CHOD: 1 LTU, 1 TTCex (1 x 2 = 2 lasers)• IRC/SAC: 1 LTU, 1 TTCex (1 x 10 = 10 lasers) -> Do they need 10 lasers ?• LKR: 2 LTUs, 2 TTCex (2 x 2 = 4 lasers)• LKR/L0: 3 LTUs, 4 TTCex (1 TTCex is spare) (4 x 6 = 24 lasers)• MUV: 1 LTU, 1 TTCex (1 x 10 = 10 lasers) -> Do they need 10 lasers ?• TDQ: 2 LTUs, 2 TTCex (2 x 4 = 8 lasers)• SPARE: 1 TTCex (1 x 4 = 4 lasers)GRAND TOTAL: 20 LTUs, 22 TTCex, 115 lasers13/7/2011 33
TTCoc order
13/7/2011 34
• CEDAR: 1 TTCoc 1:4• GTK: 3 TTCoc 1:16• CHANTI: ---• LAV: 4 TTCoc 1:8 • STRAW: 1 TTCoc 1:4, 4 TTCoc 1:16• RICH: 1 TTCoc 1:8• CHOD: ---• IRC/SAC: ---• LKR: 10 TTCoc 1:4• LKR/L0: 3 TTCoc 1:16 • MUV: 1 TTCoc 1:4• TDQ: ---• SPARE: ---GRAND TOTAL: 13 TTCoc 1:4, 5 TTCoc 1:8, 10 TTCoc 1:16
Optical splitter price
• 1x4 SM Splitter 19" 1U Rack Mountable type Connector & Adpator type : ST Optical Spec : S-grade Unit price= 120 €
• 1x8 SM Splitter 19" 1U Rack Mountable type Connector & Adpator type : ST Optical Spec : S-grade Unit price= 145 €
• 1x16 SM Splitter 19" 1U Rack Mountable type Connector & Adpator type : ST Optical Spec : S-grade Unit price= 215 €
13/7/2011 35
Summary (ordered)Num. of
LTUboards
Num. of TTCexboards
Num. of lasers on
TTCex
Num. of TTCoc
1:4
Num. of TTCoc
1:8
Num. of TTCoc
1:16
Confirmedby
CEDAR 1 1 2 1 - - O.K.
GTK 3 3 3 + 3 + 3 - - 3 O.K.
CHANTI 1 1 10 - - - O.K.
LAV 2 2 10 + 4 - 4 - O.K.
STRAW 2 2 4 + 4 1 - 4 O.K.
RICH 1 1 10 - 1 - O.K.
CHOD 1 1 2 - - - O.K.
IRC/SAC 1 1 10 - - - O.K.
LKr readout 2 2 2 + 2 10 - - O.K.
LKr/L0 3 4 6+6+6+6 - - 3 O.K.
MUV 1 1 10 1 - - O.K.
TDAQ 2 2 4 + 4 - - - O.K.
SPARE - 1 4 1 1 1
13/7/2011 36
Electronic Rack Infrastructure1/ Rack Layout in TCC8
NA62 TC Meeting, Gilles MAIRE PH/DT-DI5 April 2011 37
Electronic Rack Infrastructure1/ Rack Layout in ECN3
NA62 TC Meeting, Gilles MAIRE PH/DT-DI5 April 2011 38
Electronic Rack Infrastructure1/ Rack Layout in Electronics Barrack
NA62 TC Meeting, Gilles MAIRE PH/DT-DI5 April 2011 39
Drawn by R. Fantechi
Electronic Rack Infrastructure1/ Rack Layout in PC Farm Room
NA62 TC Meeting, Gilles MAIRE PH/DT-DI5 April 2011 40
Drawn by R. Fantechi
Electronic Rack Infrastructure2/ First Proposal of Rack
NA62 TC Meeting, Gilles MAIRE PH/DT-DI5 April 2011 41
Rack with Ventilator+ Cost (~1230chf) per rack.- Power dissipated in the cavern.- No monitoring of rack.- No power cut in case of fire or overheat.- No alarm in the fire brigade.
- The cost include the purchase of 1 LHC rack,1 ventilator and 1 cable trays.- Additional ventilator cost 300chf.
Electronic Rack Infrastructure2/ Second Proposal of Rack (similar LHC)
NA62 TC Meeting, Gilles MAIRE PH/DT-DI5 April 2011 42
Rack with LHC Rack monitoring- Cost (~1960chf) per rack.- Power dissipated by air water exchanger. Need to install water circuit.+ Monitoring of rack (temp, humidity, turbine…) with PVSS supervision by CAN bus.+ Possibility of power cut in case of fire or overheat.+ Possibility of alarm in the fire brigade.
- The cost include the purchase of 1 LHC rack, 1 rack monitoring,1 heat exchanger, 1 deflector and 1 cable trays.- Additional heat exchanger cost 90chf.
GTK racks
13/7/2011 43