1. dBLM Hardware and Signal
My current understanding
Your input appreciated!

2. System layout
16L2
Amplifier
40dB
(factor 100)
Splitter
Attenuates by 6dB (factor 2)
ROSY box
One waveform input and one histogram input
pCVD 500um thick 1 cm^2
Long cable
??m
Attenuation?
Digital signal
IR7
Splitter
Attenuates by
?? dB
Amplifier
20 OR 40dB
(factor 10 OR 100)
pCVD
500um thick
1 cm^2
Attenuator 20dB (factor 10)
Long cable
??m
Attenuation?
Digital signal AC AC
ROSY box
one waveform input and one histogram input
Splitter
Attenuates by
6dB (factor 2) DC
ROSY box output given in Volt. 1V corresponds to ? pCVD signal?
16L2: aproximately amplification by 50

3. dBLM Loss Signal The bunch loss signal is the integrated charge.
Pile-up (signal duration longer than 25ns) , baseline droop, saturation effects make it difficult to integrate.
A good approximation is Arek’s algorithm, when the two point are well defined (do not suffer from noise fluctuations) – good for IR7.
Assume rise time and fall time do nto depend on signal height
Integrated charge and voltage delta connected by scaling factor

4. Signal Features and Artefacts

5. Pile-up Signal from dBLM is a Landau distribution:
For large x (time): can be approximated by exponential decay of the signal
Substantial part of bunch signal overlaps with following bunch (longer than 25ns)  pile-up
Two time constants: rise time and fall time
Are the time constants independent of the signal height? Verify!
If so, all the previous pile-ups can be summed to one exponential with this time constant
 correction be subtracting exponential?

6. Baseline droop
From the lower cut-off frequency of the band-pass response of the signal chain – what ever element is limiting (feeding capacitor inside the diamond?)
Bunch trains:
Equal areas
Baseline shift proportional to intensity t
Time constant for droop and baseline restauration is the same?
Droop visible in the yellow and the blue signal
Droop correction in the new VFC card (integrates losses per bunch): average between point before the bunch and point after the bunch subtracted for the next (!) turn:
Ok only if turn-by-turn variation is small (not statistically driven), not the case for 16L2
dBLM (in contrast to FBCT) signals for consecutive bunches overlap!

7. Second signal peak
Farther away (varying distance), probably a signal reflection which can be cured on the HW level (blue trace)
Distance: 2 times the cable length transmission time
Signal speed in coax cable: ~ 2/3 c = 2E8 m/s
1ns ~ 0.2 m
very close to main bunch-peak
same effect or different effect (~2m)?
Or simply the following bunch (below)
 Send me your examples For checking!

8. Saturation Effects
Diamond has high dynamic range, but the other elements in the signal chain can saturate.
Send me examples, if you have and if you know which element was saturating.
Rosy box configuration is limiting the signal height
Arek: limit on green signal (right)?
Amplifier: output limited to 1V: corresponds to 0.1V 1/10 attenuator or less (cable and splitter attenuation) in plot on the right: 0.06 – 0.08 V  above this artefacts
Very high signal on TCP dBLM, when B1 is dumped: Signal does not have Landau form, but sharply breaks down and signal height is cut and followed by ringing
Pink curves next slide
Blue curve next slide?

9. Cont. Right: pink: ringing; blue: ??
Below: pink signal does not look saturated, still there is ringing afterwards
TDI
long.
High gain /
Low gain
TCDI on
MKE rise
trans.
Reflection
due to high peak.
Ignore it!
10/08/2017
F. Burkart - BTP section meeting
10/08/2017
F. Burkart - BTP section meeting 9
Clean:B2 – IP8
With over-injected pilot

10. What is the statistical fluctuation:
Estimate average signal per particle hitting the dBLM  average number of particles  statistical error due to fluctuations
 How many turns (or bunches) need to be averaged for reasonable statistical error? Or: How high does the signal need to be (taking into account attenuation etc.) for reasonable statistical error

11. Questions How to trigger 16L2 diamonds on local UFOs?
E.g. IP7 scope trigger on multiple peaks within several bunches duration  transmit to IR7?
Trigger on UFO buster?  trigger arrives too later
Trigger on mobile IC (not connected to standard readout) e.g. electronics like direct dump BLMs?

12. From Arek’s email Integrated loss /per bunch
-> cross correlating it to the IC could give some real conversion factors. V -> Gy/s -> p/s
-> loss structures over beam/batch/train
Waterfall plot bbb/turnbturn
-> indication of some resonances, we (Daniel, Bjorn and myself) already have seen some 3rd and 7th turn... -> This waterfall that Daniel showed last BLM WG... however still some understanding is needed.
Correlation to (divided by): intensity, emittance, brightness, (position, tune ?)
ADTBox / HeadTail
-> correlated to a) over certain turns could give some indications of most active bunches with the amplitude of this activity.
Spectrum analysis of the bbb losses
-> I got a separate note on that that is in preparation
-> understanding the various peaks in the loss spectrum (I've got many NON 16L2 related as well for comparison)

13. End

14. IR7: 1 count (bit) is about 1 mV
16L2:
1 count is about 15 mV
Baseline: 1-2 counts
Bunch signal: up to 8 counts
dBLM:
Readout: 9 sec of deadtime
CH0: B1
CH1: B2
CH2: trigger
CH3: turn clock: bunch zero in index number of CH0 and CH1 + offset (cable length etc.)
ATTRS: list of attributes (metadata in each file!)
Time structure: 1.6 ns; unit signal: V
ICs:
BLM dose value: 1 per hour, logging DB DOSE_HH (only during beam, but offset included), 83 sec RS every 83 sec

15. CVD Diamond response to radiation as ionisation chamber
Diamonds for Beam Instrumentation, E. Griesmayer, B. Dehning, TIPP 2011
Ionization energy: 13 eV/eh pair
Stopping power for MIP: 615 eV/um
pCVD 500 µm thickness, Size: 1 cm2 (all the same?)
Drift time with applied E field of 1V/um (bias voltage 500V): about 6 ns; drift velocity 10^5 m/s
Dynamic range (1 – 5E9 particles)
Photons:
Neutrons: 6MeV – 1 GeV (and above)
Protons: 100% efficiency
, ions, electrons

16. LHC BLM System Main purpose: prevent damage and quench
3600 Ionization chambers
Beam abort thresholds:
12 integration intervals: 40μs to 84s (32 energy levels)
 1.5 Million threshold values
Each monitor aborts beam
One of 12 integration intervals over threshold
Internal test failed
Requirements and Challenges
High Dependability (Reliability, Availability, Safety)
Threshold precision (factor 2)
Reaction time 1-2 turns (100 – 200 μs)
Dynamic range: 108 (at 40µs 105 achieved – 106 planned)
Radiation hard: currently at CERN development of kGy radiation hard readout to avoid noise from long cables

17. Beam Abort Threshold Determination
Relate the BLM signal to the:
Number of locally lost beam particles
Deposited energy in the machine
Quench and damage levels
Extensive simulations and experiments during system design and beam tests in the LHC
Proton loss locations (tracking codes: MAD-X, SIXTRACK)
Hadronic showers through magnets (GEANT, FLUKA)
Magnet quench levels as function of beam energy and loss duration
Chamber response to the mixed radiation field (GEANT, FLUKA, GARFIELD)

18. Diamond Detectors Fast and sensitive Small and radiation hard
Used in LHC to distinguish bunch by bunch losses
Dynamic range of monitor: 109
Temporal resolution: few ns
Test system installed in cryo magnet at LHC