conform-emma-rev-mtg3-talk-0023v2.0-diagnostics-FNAL ... - ASTeC

shrubflattenUrban and Civil

Nov 25, 2013 (3 years and 11 months ago)

95 views

1

EMMA

Diagnostic devices FNAL

Jim Crisp

December 2007


(updated for 3
-
10psec sigma bunch length)

2

overview


3 WCM’s (Wall Current Monitors)


beam charge


bunch phase or timing


bunch shape (limited 4GHz bw = 40ps sigma t)


89 BPM receivers (Beam Position Monitors)


4 injection, 82 ring, 3 diagnostic


YAG:Ce screens (profile monitor screen)


Yttrium Aluminum Garnet activated by Cerium


Convey few words



3

WCM details


Basically one turn transformer


ceramic vacuum break and magnetic core


40mm aperture, 110mm length


Microwave cutoff in beam tube defines upper frequency
limit


4.4GHz for 40mm ID round pipe


difficult to separate signals propagating along beam pipe from those
induced by beam charge


microwave chip resistors and flexible circuit board used to form
resistive combiner are good for ~10GHz


Tested with 50ohm structure with higher cutoff frequency


4 equally spaced pickoff points provide a measurement
reasonably independent of beam position and rejects
non TEM modes by symmetry

4

WCM ferrite


2ohm gap would provide reasonable signal


Terminates modes in ceramic


0.32Vp for 32pC (40psec response)


desire droop time constant much longer than 830nsec


(15 turns X 55.3nsec/turn)


Ceramic Magnetics MN100 ferrite core


24uH = N^2 uA/L (for N=1turn)


2ohm 24uH = 12uSec (13.4KHz)



1024 turns (55usec) will show significant droop


Could restore background in software



15KHz to 4GHz bandwidth


5

WCM bandwidth


4.4GHz microwave cutoff for 40mm ID


13KHz to 4GHz WCM bandwidth


4GHz ==> 40psec risetime


Can’t resolve bunch structure


3psec (0.9mm) sigma bunch length


53GHz sigma frequency content


80pC in 3psec would have 11Amps peak




Will microwave power launched by beam passing
discontinuities in the beam pipe cause problems?


arcing or interference with diagnostics?


6

Fermi Linac WCM

7

Frequency Response


Measured with 50ohm test fixture to overcome
microwave modes in pipe


‘wiggles’ are from small mismatch in test fixture


(2ohms/50ohms =
-
26db)

Linac RWCM [output/pipe]
-30
-28
-26
-24
-22
-20
0
2
4
6
8
10
Freq [GHz]
dB
8

possible oscilloscopes


Tektronix ‘DPO71254’


Windows based operating
system


4 channels


50Gs/s


10Meg


200usec of memory


12.5GHz bandwidth


23psec risetime


10mV to 1V/div


10ps to 1000s/div


$88500


4 week lead time


Agilent ‘DSO81204’


Windows based operating
system


4 channels


40Gs/s


4Meg (opt 001)


100usec of memory


12.0GHz bandwidth


26psec risetime


1mV to 1V/div


10ps to 1000s/div


$92538 (+$5184 opt 001)


5 week lead time



9

WCM notes


Develop algorithms in LabView


Could be moved to EPICS


Bandwidth limited to <4.4GHz by microwave cutoff


50Gsps scope = 20psec/sample


need to limit bandwidth of signal


4
th

channel on scope could digitize rf fanback signal


wideband buffer amp to share WCM signal


short low dispersion cables require good termination


Required charge and phase accuracy


1% charge, 2 deg at 1.3GHz (4.3psec)


Waiting for approval to spend money

10

BPM details


4 injection, 82 ring, 3 diagnostics


position, charge, time of flight


measure single bunch each turn for 15 turns (55.3nsec/turn)


1000 turns for commissioning?


bunch charge 16pC min / 32pC max


50um / 25um resolution


bunch length 3 to 10psec sigma


longer bunches? (may affect design)


48mm ID with four 20mm buttons


3.7GHz microwave cutoff for 48mm round pipe


Pos = 12.65(A
-
B)/(A+B)


25um desired alignment error



Evaluate with ERLP beam (summer 2008)


Install Feb


May 2009


Beam July 2009

11

BPM approach


Digitizing 3 to 10psec bunch directly is not practical


Diode detector requires matched diodes and careful
temperature compensation


Switches require accurate, stable timing


Downconverters need to be carefully matched and stable
(used on atf at KEK)


(~10um single bunch single turn resolution with 50e8)


Started looking at their design



Consider RLC to produce a decaying sinewave from
bunch transient


Use 500MHz 12bit adc

12

estimate

frequency content


Bunch length and button shape determine frequency content of the bpm
signals


bpm peak when button is ¼ wavelength in diameter, zero when ½ wavelength


3.75GHz peak, 7.5GHz zero


3psec sigma t (gaussian) bunch corresponds to 53GHz sigma f (gaussian)


(10psec / 16GHz)


Beam pipe is a complex shape but microwave cutoff in 48mm round pipe is
3.7GHz

0
0.5
1
1.5
2
2.5
0
1
2
3
4
5
6
7
8
GHz
amplitude
estimated frequency response for 20mm button, 10ps
bpm
beam
beam x bpm
microwave cutoff
13

estimate

button current


difference of two gaussian 20pC 10psec sigma bunches displaced
in time by 20mm/c = 133psec


2.5pC in each half of the doublet


Independent of bunch length for bunch << button length

-100
-75
-50
-25
0
25
50
75
100
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
nsec
Ibutton [mA]
for 20pC (1.25e8), 10psec sigma t, 20mm button
2.5pC in each half
14

400MHz 100ohms


100ohms, 12nsec, 60pf, 2.64nH, Q=15


20pC, 10psec, 20mm button


2.5pC in 60pf = 41.7mV


6.65mVpk of 400MHz


Independent of bunch length for bunch << button length


Rms value averaged over 55ns


0.233*6.65mVpk = 1.55mVrms

-8
-4
0
4
8
0.00
20.00
40.00
60.00
nSec
mVolts
400MHz 100ohms
-50
-40
-30
-20
-10
0
10
0.00
0.25
0.50
0.75
1.00
nSec
mVolts
400MHz 100ohms
2.5pC/60pf = 41.7mV
15

ringing filter


signal must decay to 1% in 55nsec (1 turn)


Insures signal from 1 turn does not corrupt the next


12nsec 1/e time constant


A 400MHz signal sampled at 500MHz looks like 100MHz


-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
0
10
20
30
40
50
60
nsec
with 500Msps 400MHz looks like 100MHz
16

signal processing


For each turn (27 samples)


Multiply samples by sin(

t) and integrate (As,Bs)


Multiply samples by cos(

t) and integrate (Ac,Bc)


Could use decaying sinwaves but advantage is small



t

= {atan(As/Ac)

+
atan(Bs/Bc)
}/ 2



Intensity = A+B


(A = sqrt(As^2+Ac^2) for plate A)


Position = 12.65mm(A
-
B)/(A+B)

17

expected errors


for a full scale peak signal

(2.2Vp
-
p) (gain of 110 for 32pC)


rms of decaying sinewave, 27 samples, R = 12.65mm (cal or gain)


2.2Vp
-
p* ½ *0.233 = 0.26Vrms


SINAD

(signal to noise and distortion)


4.3um


0.14psec


INL

(integral nonlinearity error)


2.0um


0.07psec


AU

(aperture uncertainty)


3.4um (proportional to frequency)


0.11psec



thermal noise for 100ohm 400MHz RLC (25MHz bandwidth)


Cannot be improved with an amplifier


2.5um


0.08psec



Totals (larger with smaller signals and larger positions)


Allowing for 20mm position requires 15db or 5.62 headroom


36um


1.1psec







2
1
2
2
2
2
1
2
2
2
2
2
2
2
233
.
0
1
1
2
1
2
2
2
2
233
.
0
1
1
2


















































N
INL
N
INL
f
AU
SINAD
N
psec
f
AU
SINAD
N
R
m




18

Fermi digitizer board


Currently designing ATCA board for ILC test linac


ATCA


‘Advanced Telecommunications Computing Architecture’


FPGA with enough memory to sample all 15 turns at 500MHz


Field Programmable Gate Array


16 dac channels AD97736


12 bit 1200msps


16 adc channels ADS5463


12 bit 500msps


SINAD 64.3dbc (signal to noise plus distortion)


AU 0.16psec

(aperture uncertainty)


INL 1lsb

(integral nonlinearity error)


2.2Vpp max input


Prototype board expected 3/2008


Test with ERLP beam?


Final design 6/08


EMMA Installation 2/09


EMMA Beam 7/09

19

ATCA Digitizer Development

19

10/100/1,000

ETHERNET


Base Interface

10/100/1,000MB

Ethernet


Fabric

Interface

Full Mesh

Digital IO:

Ext Clock,

Ext Trigger,

Ext Gates,

Front
-
End

Control

Analog Inputs


Ch1
-

ch16







Analog Outputs



Ch1
-

ch16


Ext CLK


Int CLK

Clock Synthesizer

and

Distribution


AD9510


FPGA

STRATIX II

EP2S60F1020


Fabric & Base
Ethernet

Interface



NIOS II


DDR2

SDRAM

1GB

Hot Swap Power Controller,

Low Noise, Slew Control

Spread Spectrum

DCDC

Power

48V DC

200W

FLASH

256MB

DDR2

Controller

ATCA

Backplane

48V A

48V B

ATCA

Rear Transition

Board

10/100/1,000

ETHERNET

IPMC

Intelligent
Platform
Management Bus
IPMB
-
A

IPMB
-
B



4 +4 Channels

ADC&DAC

Module



4 +4 Channels

ADC&DAC

Module



4 +4 Channels

ADC&DAC

Module



4 +4 Channels

ADC&DAC

Module













1GSPS

Serial Link

Full Mesh














ATCA Digitizer Diagram

20

bpm system


Each ATCA crate would have a front end or ‘slot
zero controller’ that processes and delivers data
to an EPICS data base via ethernet


Each ATCA crate would likely have one timing
distribution card


The EPICS application is important and
represents significant effort carefully coordinated
with physics requirements of the machine.


Scott Berg, Shinji Machida have some excellent ideas

21

conclusions


A simple resonant filter was explored.


peak signal independent of bunch length for bunch << button length


Need bpm to proceed


A down converter should be considered


Still use resonator on button


Looking at KEK atf design


The bpm signal amplitude should match the adc input range


need amplifier (gain of ~100), antialiasing filter before adc


Small interface board in ATCA crate


A/B will change by up to 10db (factor of 3) with position


bunch charge can change from 20 to 80pC (factor of 4)


Bunch length likely to change from 3 to 10psec sigma


12um rms error for full scale inputs could increase to 144um rms



Make sure modeling program requirements are met


22

for example

Recycler digital receivers


Simulate receiver inputs to model constant position with changing intensity


14 bit 80MHz adc’s, 120 samples, 63mm radius


each position measured 100 times


Mean is plotted on the left and the standard deviation is on the right

-50
-25
0
25
50
-60
-50
-40
-30
-20
-10
0
A+B [db]
Pos [mm]
0
0.5
1
1.5
2
-60
-50
-40
-30
-20
-10
0
A+B [db]
Pos Noise [mm]
23

Recycler adc linearity


Difference between previous slide measurements and horizontal lines


Position depends on linearity of A and B adc’s


Dotted lines from

0.3bits typical for the adc


Solid lines indicate error from previous slide

-0.5
-0.25
0
0.25
0.5
-60
-50
-40
-30
-20
-10
0
10
A+B [dbm]
pos [mm]
24

YAG:Ce

profile monitor screens


Harvested from the brain of Ardin Warner:



www.crytum.com


YAG:Ce
-

Yttrium aluminum garnet activated by cerium
is a fast scintillator. The material's mechanical properties
enable production of thin screens down to 0.005 mm
thick.


They can drill holes or apply scribe lines for optical
reference


Adjusting the doping can provide relaxation times that
allow separation of turns (<50nsec)


30mm diameter 100um thick $100

25

YAG:Ce & Fermi eCool


Arden Warner and others use YAG:Ce 100um
thick 30mm diameter screens to measure
4.36MeV electron beam of 500mA, 2usec
(6.25e12) with 1Hz rep rate.


use gated CID camera 800x600 pixels $8k


gate can be 20nsec


50um resolution


National Instruments 'Vision' for processing data


'imageJ' is also very good and it’s freeware

26

eCool beam

Picture of YAG:Ce used in the electron cooler

A couple of images taken with a gated CID camera. The
gate width here is 100ns but the camera system can be
gated as short as 200 ps which is not necessary in
cooler application. The electron beam is a 2µs pulse at
1Hz rep rate, 100
-
500 mA.

YAG/Pepper
-

pot image of the beam
before adjustments.

Photo’s courtesy of Arden Warner, Fermilab

27

Summary


Ready to start spending money on WCM’s



Need bpm to develop analogue front end


4 feedthroughs would allow me to build prototype at Fermi


Bpm time line is tight


Fermi has deployed digital bpm systems in Recycler, Main Injector, Tevatron, and atf at KEK.


They are expensive but work well


A commitment would help leverage resources to insure time requirements are met


Perhaps a commitment to purchase 1 digitizer, ATCA crate, and controller? (~$24k US)


384 heavy stiff bpm cables attached to expensive/fragile vacuum feedthroughs


Need to provide for interface bracket attached near bpm


Can they follow the girder?



Spares?



I don’t know anything about screens


Arden Warner, Alex Lumpkin do.

28

resolution vs accuracy


25um resolution


very surprised if physical bpm is linear to that
accuracy even with mapping


hor/ver positions are coupled


12.65mm(A
-
B)/(A+B) ~ 0.72mm/db


25um = 0.035db


Cable aging, bending, cable temperature, humidity,
changes in termination, integrity of connections,
number of connection cycles, corrosion



Should incorporate some ‘calibration’ feature