1. Beam Secondary Shower Acquisition System: Front-End RF Design
Student Meeting
Jose Luis Sirvent
PhD. Student
19/08/2013

2. 1. The dynamic range and the three lines
Two or Three lines to cover all the needed Dyn. range (1-1e6MIPS)
Scheme valid for QIE as well as ADC
Technique based on splitting the signal (Detector current or Charge -> Voltage)
The idea is to use 1 detector and 2 or 3 lines and sample in parallel
Profile reconstruction Bunch By Bunch based on the combination of lines

3. 1. The dynamic range and the three lines
Initial proposal of architecture:
3 Signal cables (Tunnel – Detector) : 34dB, -6dB & DC
2 Power supply cables (Tunnel - Detector)
1 Splitting + Amplification in Tunnel
1 Splitting + Attenuation in Surface HV
Cividec
Amplifier
34dB
Cividec
AC-DC Splitter
-6dB
Cividec
Diamond Detector
Mini-Circuits
DC-4GHz
Splitter AC
40dB
pCVD
-6dB
12V DC
Tunnel DC
-6dB
-12dB
The proposal can be Optimized:
1. The pCVD from Cividec has high output impedance
RF equipment needs to be well adapted for optimal transmission
2. Line DC could be avoided
We can split in surface x3 and lowpass one output
Less Installation needed
-6dB
Mini-Circuits
DC-4GHz
Splitter
-32dB
Cividec
Attenuator
-6dB
-20dB
Surface

4. 1. The dynamic range and the three lines
To adapt… or not to adapt? (CK50)
Better frequency response if adapted in both sides, see green line

5. 1. The dynamic range and the three lines
Updated proposal:
Lines well adapted
Less components in the tunnel (Simpler layout) HV
Cividec
Amplifier
34dB
Cividec
Diamond Detector
-6dB
DC-4GHz
Splitter
40dB
pCVD
-6dB
12V
Termination
50Ω
Tunnel
-6dB
Not yet clear:
If DC measurement is necessary then -9dB splitter:
Needed lines 34dB, -15dB & -35dB
-12dB
-6dB
DC-4GHz
Splitter
-32dB
Attenuator
-6dB
-20dB
Surface DC
Low Pass
Filter
Fc= 5 Hz

6. 1. The dynamic range and the three lines
Updated proposal:
Lines well adapted
Less components in the tunnel (Simpler layout) HV
Cividec
Amplifier
34dB
Cividec
Diamond Detector
-6dB
DC-4GHz
Splitter
40dB
pCVD
-6dB
12V
Termination
50Ω
Tunnel
-6dB
Look Out!!
Worst case:
1e6 MIPS in 1ns
Charge = 1.6e9 C
pCVD peak current =1.6A!!
Needed to calculate
the dissipated power
per device and make
wise decisions or … 
-12dB
-6dB
DC-4GHz
Splitter
-32dB
Attenuator
-6dB
-20dB
Surface DC
Low Pass
Filter
Fc= 5 Hz

7. 2. Going back to Pspice Model of the proposed Front-End
!!!

8. 3. To shape or not to shape 1. Shaping is filtering: Band-Pass
2. A correct shaping avoids bunch pile-up (Cable effect): Recover base-line
Criteria:
Ti= 1/2 Tbunch (0.5e-9 for 1ns bunches)  Fc low= 2GHz
Td= 1/5 Tbetween bunches (5e-9 for 25ns spaced bunches)  Fc high= 200MHz
3. Shaping “cleans” the signal: Better SNR
Working Freq. Range
(200e6 – 2e9 Hz)

9. 3. To shape or not to shape By using our Matlab GUI: Observations:
Shaper returns base-line (Cable independence)
Shaper provides better results in fitting sigma with ADC
On  Mean error: 2.95e-3%
Off  Mean error: 8.55e-3%
If Shaper used, not possible to use QIE10 (mean charge = 0 C)
QIE10 results justified by Digitalization scheme (eqr 0.7 – 1.4%)
When integrating the Pile-Up effect is more evident.
Errors maybe too small ( <0.1% ) for taking a decision.
Timing evaluation of ADC’s to see if possible peak detection at 40Mhz. (Synchronization Issues??)

10. 3. To shape or not to shape Peaks With Shaper Integration With Shaper
Peaks Without Shaper
Peaks With Shaper
Pile-Up: %
Pile-Up: %
Integration Without Shaper
Integration With Shaper
25ns bunch charge tends to 0
Not possible!
Pile-Up: %

11. 3. To shape or not to shape Example of an Slow shaper:
Band-pass 1e3 - 2e9 Hz
Ti:0.5e Td: 1e-3
Example of an Quick shaper:
Band-pass 200e6 - 2e9 Hz
Ti:0.5e Td: 5e-9
In case of using a shaper this must be quick enough (bunch level) to not modify the Gaussian shape.
But remember, If shaper is used not possible to use QIE10!!
Option A (No problem with RH. Amplif) :
Quick shaper (200e6 – 2e9 Hz)
Clean signal
No impact on Gaussian profile
Only compatible with fast ADC’s or standard well synchronized
Option B (Some developments are needed):
No shaper
‘Dirty’ signal
No impact on Gaussian profile (if not much noise)
Compatible with QIE10 and ADC’s