1. Study Report of Neutron Irradiation to FEE @ Kobe Univ.
T.Higuchi, M.Nakao (KEK)
E.Nakano (OCU)
Jan.26,2011
TRG/DAQ Workshop

2. Study Summary @ “Yayoi” Reactor
Single event upset (SEU)
Number of SEUs is counted in a given period.
RAM block based test was not possible in a systematic way because of too many SEUs. Then, we counted # of verification errors during the firmware downloading.
# of SEUs seemed module dependent up to factor 5 — could be position dependence?
0:1164 1:391 2:849 3:304 4:685 5:223 6:482 7:222 0:closer to / 7:far from the reactor, 0,2,4,6 from rear of FPGA.
Linear dependence on the SEU frequency versus time / power.
Rate at Belle II is not very high, once in 41–215 hours
1/35000 scaling factor for full SuperKEKB luminosity.

3. Study Summary @ “Yayoi” Reactor
Optical transceivers
Both AVAGO and FINISAR module stopped light emission after irradiation (even at the low power cycle), but they seem to have recovered by power cycling.
FINISAR modules have no degradation as far as the data transmission test is concerned.
Two links with AVAGO modules became unstable after irradiation and did not recover by power cycling.

4. Purpose of the 3rd Study
KEKB beam line
Yayoi reactor
Effects/damages on the electronics by irradiation of higher energy neutrons than Yayoi reactor
Yayoi … O(100keV) neutrons
This study … Q = 4.36MeV neutrons
More systematic SEU counts than we did in Yayoi
New firmware counted up # of SEUs by itself.
Automatic recovery procedure from the SEU was implemented in the firmware.

5. Study Overview Dates: Dec.3rd-4th, 2010.
Place: Maritime study, Univ. of Kobe.
Setup
Irradiate 8 MGT FINESSEs and optical transceivers by neutrons, and see how they got damaged and how an automatic recovery procedure implemented in the firmware worked.
Neutron source:
⁹Be + d (3MeV) ¹⁰Be + n (Q=4.36MeV)

6. Beam Line(s) of Kobe Univ.
Ion source
Kicker magnet
Beam line we used
Tandem
⁹Be target
Nakano-san
⁹Be + d (3MeV)  ¹⁰Be + n (Q=4.36MeV)
MGF FINESSE irradiated

7. Setup Schematics 8 MGT-FINESSE + MGTF-JTAG (remote).
Radiation area
8 MGT-FINESSE + MGTF-JTAG (remote).
8 MGT-FINESSE on 2 COPPERs (local).
Remote MGTF-JTAG powered by 5V DC power supplies.

8. MGT FINESSE Brick M.Nakao neutron beam
8 MGT FINESSEs are stacked up so that all SFP connectors come to central region (i.e.: irradiation region of n) of the FINESSE brick.

9. MGT FINESSE Dimensions
186mm
(–26mm,16mm)
14mm
22mm
38mm
SFP
(26mm,0mm)
FPGA
76mm
(0mm,0mm)
24x24mm2
66mm
146mm

10. MGT FINESSE Dimensions
[mm]
Type-A: original
Type-B: flipped
SFP: (–26, 16)
FPGA: (26, 0)
SFP: (26, 16)
FPGA: (–26, 0)
Type-C: 180⁰ rotation
Type-D: 180⁰ rotation + flipped
SFP: (26, –16)
FPGA: (–26, 0)
SFP: (–26, –16)
FPGA: (26, 0)

11.  Distance from ⁹Be to FINESSE surface = 27mm
Neutron Beam Spot D C A C D B
neutron beam D C
 8mm
= Neutron beam spot
 Distance from ⁹Be to FINESSE surface = 27mm

12. Geometry of MGT FINESSEs
SFP
MGT#
R [mm]
θ [⁰] 8 38 45 7 37 6 65 39 5 73 22 4
102 15 3
117 2
138 12 1
142 11
FPGA
MGT#
R [mm]
θ [⁰] 8 31 28 7 45 49 6 65 16 5 73 4
102 3
123 19 2
131 1
145
R … distance from the ⁹Be to SFP/FPGA.
θ … angle between the beam axis and ⁹Be  SFP/FPGA vector.

13. Angular-Dependent Beam Intensity
K.Miyazaki et al (Kobe Univ.).
Relative intensity of n+γ
Ratio of n/γ
X(θ)
Y(θ) 4 30
X ≈ 17 – ––– θ
degree [⁰]
degree [⁰] XY
I(R,θ) ∞
R²(1+Y)

14. Absolute Neutron Counts
Total dose from 16:00 to R=27mm, θ=0⁰ was 7.5x10¹¹, which is equivalent to 7.5 year Belle II operation.
counts R=27mm, θ=0⁰
hour of Dec.4th, 2011
MEMO:
Above picture is converted from the area monitor data.
1 area monitor count corresponds to 580 neutrons R=27mm, θ=0⁰.

15. Relative Intensities MGT# SFP [%] FPGA [%] Neutron irradiation 8
31 (2.3yr)
64 (4.8yr)
From head side 7
25 (1.9yr)
13 (1.0yr)
From tail side 6
12 (0.9yr)
24 (1.8yr) 5
9 (0.7yr) 4
6 (0.5yr)
7 (0.5yr) 3
5 (0.4yr)
4 (0.3yr) 2 1
3 (0.2yr)
100% R=27mm, θ=0⁰

16. Transceiver Damage
We observed no permanent damage in the optical transceivers after the irradiation.

17. Dotted lines indicate neutron counts in an arbitrary vertical scale.
FPGA Damage
n from head
n from tail
Normalized # of SEUs
hour
Dotted lines indicate neutron counts in an arbitrary vertical scale.

18. FPGA Damage No permanent damage is observed.
The FPGAs irradiated from the tail side generated more SEUs than ones done from the head side.
Similar tendency was seen in the study at Yayoi reactor.
Suggestion
Arrange the FPGA’s head side to face the beam axis.
We do not know where neutrons come, but it is quite unlikely to have neutrons from the detector outer side.
Electronics
CDC End-plate n
beam

19. Summary
We observed no permanent damage on the FPGAs and transceivers after <5 year equivalent neutron irradiation of Q=4.36MeV.
The head side of the FPGA is systematically tough against the neutron irradiation than the tail side.
FPGAs on the detector are better to face beam axis to avoid frequent SEUs. ~

20. Future Schedules Short term Long term
On Feb.3rd and 4th, we irradiate neutrons to the electronics again at Yayoi, which is the final term of Yayoi operation.
Long term
Radiation study of γ TIT (not scheduled yet).
[?] Investigation of FPGA-inside to see why we have the head/tail asymmetry.

21. FINESSE Brick – Stacked-up FINESEs
↕ 9mm
14mm ↕
↕ 18mm
30mm ↕
↕ 18mm
14mm ↕
↕ 9mm