11
TWEPP – Vienna – September 27th 2011
Wideband (500 MHz) 16 bit Dynamic Range Current Mode Preamplifier for
the CTA cameras
D. Gascóna, A. Sanuya, J. M. Paredesa, Ll. Garridoa, M. Riboa, X. Sieirob
Universitat de Barcelona
Institut de Ciències del Cosmos ICC-UB (a)
Departament d’Electrònica (b)
1
22
I. Introduction
II. PreAmplifier for CTA (PACTA)
III. Input stage
IV. Current gain stage
V. Transimpedance stage
VI. Noise
VII. Test results
VIII. Conclusions.
Outlook
3
~ 10 kmAir shower
~ 1o
Che
renk
ov li
ght
~ 120 m
Gamma ray
I. Introduction: Cherenkov telescopes
3.5o FOV càmera 577 PMTs3.5o FOV càmera 577 PMTs
MAGICCamera by IFAE
M. Martinez
Stereoscopy provides better:•Angular resolution.•Energy resolution (height).•Background rejection.•Sensitivity.
4
Artist view of CTA-NorthKari Nilsson
I. Introduction: the Cherenkov Telescope Array (CTA) observatory
10-14
10-13
10-12
10-11
10 100 1000 104 105
E x
F(>
E)
[TeV
/cm2s]
E [GeV]
Crab
10% Crab
1% Crab
GLAST
MAGIC
H.E.S.S.
CTA
Exploring the cutoff regime of cosmic
accelerators
Population studies, extended sources and,
precision measurements
High redshift AGNand pulsars
10-14
10-13
10-12
10-11
10 100 1000 104 105
E x
F(>
E)
[TeV
/cm2s]
E [GeV]
Crab
10% Crab
1% Crab
GLAST
MAGIC
H.E.S.S.
CTA
10-14
10-13
10-12
10-11
10 100 1000 104 105
E x
F(>
E)
[TeV
/cm2s]
E [GeV]
Crab
10% Crab
1% Crab
GLAST
MAGIC
H.E.S.S.
CTA
Exploring the cutoff regime of cosmic
accelerators
Population studies, extended sources and,
precision measurements
High redshift AGNand pulsars
How CTA aims to extend energy range and increase sensitivity? Large array (>1 km2)of Cherenkov telescopes (50-100): over 100K channelsDifferent sizes: dish from 6 to 24 m
Camera and electronics must be optimized in terms ofPerformanceCost and reliability: integration
5
Preamplifier SignalConditioning
PhotoSensor
Digitization
Camera Pixel Front end electronics
Read-out
I. Introduction: the camera
Front end electronics: Pixel: fast photosensors
High QE PMTs (baseline) // SiPM Modularity: cluster of 7/8 pixels Front end electronics in the camera Digitization & trigger
Huge dynamic range: 16 bits Signals up to 5 Kphe Single phe resolution for calibration:
Noise 1/8 phe PMT gain 40Ke ENC 5Ke (10 ns integration)
PM tube
Voltage dividerDC-DC converter
HV control cable
HESS cluster
5
High BW (>300 MHz, full
chain):Night Sky Background:
Up to 100 MHz Minimize integration time
Simulation (S. Vorobiov)
This work
66I. Introduction: requirements
To develop a generic preamplifier valid for any CTA camera Common component: low cost and reliability for mass production
This component must fulfil a set of demanding requirements:
Low noise Good single photoelectron resolution at PM gain of 4·104
High dynamic rangeFrom < 1/8 of phe to 5 Kphe: 16 bitsGood linearity (< 3% nonlinearity)
High BW500 MHz. Total BW including FE must be 300 MHzFor fast read-out option
Low input impedanceLow pick-up noise and high BW: very close to PMT Compatible with SiPM
Low power About 100 mW with low impedance driver (cable, tline)
Reliability/compactness Mass production, Integrated circuit, ASIC
77
II. PreAmplifier for CTA (PACTA): circuit design
• Basic circuit: Super common base input
Low noise / High DR: used in LHC calorimetry Cascode current mirror with CB feedback Fully differential transimpedance amp.
(TIA)
• Performances BW > 500 MHz Low Zi < 10 Ohm up to > 500 MHz Low noise (in=10 pA/sqrt(Hz)) Differential: optimal CMRR and PSRR
• But the current mirror can not stand a 1000 phe pulse Limited to 12/13 bit DR Saturation at 500 to1000 phe
• Not enough for 16 bit…
7
Simplified schematic
Q1
Re
Ib1Ii+
Q2
Rc
M2M1
M2cM1cVcas
CC
Vb
Iba
QF
Q1
Re
Ib1
Vcc
Ii-Q2
Rc
M2 M1
M2c M1cVcas
CC
Vb
Iba
QF
Rf Rf
Vo-Vo+
+
+
-
-
88
II. PreAmplifier for CTA (PACTA): circuit design
• Previous circuit is modified to split the input current:– Current is divided in the common base stage– Different current mirrors for high and low gain
• Each can be optimized for BW / linearity– Dedicated saturation control circuit is added to the HG
mirror• Current division remains operational even if HG mirror saturates• Saturation threshold of HG mirror can be controlled
– Range: > 6000 phe • True delta pulse with 500 MHz BW • No arrival time effect considered
8
Patent pending
Simplified schematic
Single ended and differential versions
M2M1
High Gain
M2M1
Low Gain
SATURATIONCONTROLCIRCUIT
+
-
+
-
Common base (gate) stage with
current division n:1
n 1 High Gain
Low Gain
99
II. PreAmplifier for CTA (PACTA): prototypes
AMS SiGe 0.35 um techno:– High speed, low noise and
offset HBTs– 5 V / 30 GHz NPN HBTs– 3.3 V power supply:
• DR vs power trade off
9
PACTAv1.2 chip 2 mm2
QFN32 packageSubmitted June 2011
• An input stage + current amp. block
• A complete differential PACTA:– No low impedance driver
• 2 single ended PACTAs• 1 differential PACTA• Main modifications
– New OpAmp (s.e. and fully diff. versions): • Low output impedance stage• Higher Slew Rate (1 V/ns)
– Improve compensation of input stage and current amplifier
PACTAv1.1 chip 2 mm2
QFN32 packageBack from foundry Oct. 2010
10
1i PAD Qf mQf f CC C c g R C
1
1
1 1 1
( )1 1 1
1 //
mQmQf E i
Q
i mQ E Q Rf f Q Rf E
gg s R C
cT s
s s sC g R c c R c c R
Q1 Q0
IPD
Node a Node bHigh Gain Low Gain
QC1 QC0
VbCas
Rbias
RS
CPAD
x14
x14 x1
x1
Qf
RESD
CCRf
Vccb
10
III. Input stage
• Super common base: low Zi (<15 )• Two unbalanced current outputs:
– Vbe of Q1 and Q0 is the same– Q1 = 14 * Q0: current splitting
• Cascode transistors to improve linearity:– Minimize variation of Vc of Q1 and Q0– Minimize Early effect
• HF feedback loop (Return Ratio T(s)):– Miller compensation Cc: pole splitting– Q1/Q0 emitter degeneration RE:
• Linearizes Zi and improves matching• Limited by DR and noise: tradeoff
10
T(f) (mag) vs Cc
T(f) (phase) vs Cc
Phase Margin of T(f) vs vccb
GBWP of T(f) vs vccb
Cc=0
Cc=750 fF
1
1
1
EmQ
imQf f
RgZ
g R
With pole splitting:
Dom. Pole Zero in Zi(s)At f >2 GHz
1111
III. Input stage
• Input impedance depends on Vccb
– gmQf
11
• No impact of dominant pole on Zi
– Dominant pole also affects open loop Zi
– No inductive effect in the BW of interest
Zi(f) (mag) vs Cc
Zi(f) (phase) vs Cc
Zi(f) (mag) vs vccb
Zi(f) (phase) vs vccb
1
1
11
1
1
( )
Q Rf
Q Rf f
i
mQmQf Q
Q
c c sc c R
Z sg
g c sc
Using Blackman´s impedance formula:
zZi
pZi
pZizZi
1212
III. Input stage
• Bonding inductance must be considered: – QFN32 package: < 1mm bond wires– Series resonance: CPAD*Lb (input induct.)– Rs to increase (damping factor) – On chip decoupling cap. also need damp. res.
• Ground inductance Lg is critical:– Direct feedback to Qf emitter
• Output driver provides large current pulses: – Ringing or oscillation possible!
– Downbonds to package cavity • < 0.5 mm wires
– Multiple grounds, different grounds for:• Input stage• TIAs
12
T(f) (mag) vs Lb
T(f) (phase) vs Lb
T(f) (phase) vs Lg
T(f) (mag) vs LgT(f) (mag) vs CC
T(f) (phase) vs CC
Nominal inductances
1313
IV. Current amplifier: saturation control circuit
• Low voltage cascode current mirrors – Local feedback: common base HBT Qcb
• Minimize input impedance
• Minimize Voltage variation of VCQ1
• Wideband amplifier with high DR:– BW > 500 MHz
– Current gain (AI): 2.5
– > 12 bit DR
• But M1/M1c to ohmic region for large signals: – Feedback and low input impedance are lost
• Saturation ctrl circuit “quenches” large signals:
– Qoc1 is controlled by the voltage Vc-Vm
– For low currents Vc-Vm<<Vbe_On
– If drain current of M1 increases, Vc increases and
Vm decreases
– Turn on point of Qoc1 can be tuned through Vlim
13
M2M1
M2cM1c
Iba
Vb Vcas
Qcb
Qoc1
Saturationcontrolcircuit
Vm
Vc
M3
M3c
VlimTo collector of Q1
(Input stage)
1
1
1inMIRmQcb cF mM
Zg Z g
1414
IV. Current amplifier: saturation control circuit
• Feeback loop must be compensated
– Two main poles related to M1 gate and input nodes
– Add Cc for dominant pole comp• Drawback: limits amplifier BW
– BW limited by Cin• Must increase Cc to compensate
• BW > 500 MHz provided that:– Cin < 700 fF– With PM > 65 deg
• Source degeneration (RSM) in v1.2
– For large bias currents (or with a
high pulse rate) gmM1 increases
– GBW increases and PM decreases– Source degeneration limits the
effective value of gmM1
14
0
100
200
300
400
500
600
700
800
900
0 500 1000 1500 2000 2500 3000 3500
Cin [fF]
BW
[M
Hz]
0
500
1000
1500
2000
2500
3000
Cc
[fF
]
BW [MHz]
Cc for PM=65 deg [fF]
1( )
1 1
mM cFMIR
INcF cF
mQcb
g RT s
CsR C s g
1 2 ...cF C gsM gsMC C C C
1515
V. TIA
• Single ended and fully differential versions• OpAmp architecture:
– HBT input pair + folded cascode– Miller gain stage– Output stage:
• None for PACTAv1.1• Low impedance class AB (v1.2)
• Differential OpAmp with CMFB: – Gain > 65 dB– GBW> 700 MHz– PM > 70 deg– SR = 1 V/ns (PACTAv1.2)
• Low output impedance push-pull stage:– Based on NPN HBTs– Feedback loop must be extremely fast– Compensation is critical:
• Operates in closed loop (OpAmp output stage)
– Drives 50 Ohm loads (AC coupled)
15
x14
Qfol Qfols
x2
RSENSE
MPs
Q1
Vbef
x16
CC
Class AB output stage
Provides up to 20 mA peak current with 5 mA quiescent current
Thanks to J. Lecoq, E. Delagnes and P. Moreira
Compensation of the local FB loop Return Ratio T(f) versus Vbef
Post layout simulation including bonding inductances
PM of T(f)
GM of T(f)
GBW of T(f)
UnityGainFreq of T(f)
1616
VI. Noise
• For the single ended version• Series Noise
– Dominated by the input stage• Paralllel Noise:
– Significant contribution of• Input stage• Current mirrors
• Parallel noise dominates for Cin<5 pF
16
22 1 2
4 8n bbQf S ESDmQf
e KT r R R nV Hzg
22
2 1
1 1
2
1 21 4
1 3
4.2
mMnMIR SM
I mM SM mM
gi KT R
A g R g
pA Hz
22 24
2 8.4n BQf nMIRbias
KTi qI i pA Hz
R
1717
VI. Noise
• Differential version: uncorrelated noise sources add in quadrature
– Noise should be sqrt(2) higher• Integrated output noise referred to an
input current noise source– Equivalent noise current (ENI)– Minimal for low cap. sensors (PMT): in– Still quite ok for high capacitance sensors
17
Equivalent Noise Current (ENI)
0,0E+00
2,0E-07
4,0E-07
6,0E-07
8,0E-07
1,0E-06
1,2E-06
1,4E-06
1,6E-06
1,8E-06
0,10 1,00 10,00 100,00 1000,00
Cin [pF]
EN
I [A
rm
s]
Single Ended
Differential
en=1.13 nV/sqrt(Hz)
in=11.8 pA/sqrt(Hz)
PMT SiPM, LAAPD
1818
VII. PACTAv1.1 chip test results
Pulse shape
Input current pulse
-1,20E-02
-1,00E-02
-8,00E-03
-6,00E-03
-4,00E-03
-2,00E-03
0,00E+00
2,00E-03
10 12 14 16 18 20 22 24 26 28 30[ns]
[A]
-1,49E-04-2,79E-04-4,17E-04-5,58E-04-6,97E-04-8,38E-04-1,67E-03-2,48E-03-3,35E-03-4,11E-03-5,49E-03-6,80E-03-8,21E-03-9,60E-03-1,10E-02
PACTA high gain output for different input pulse amplitudes
-1,60E+00
-1,40E+00
-1,20E+00
-1,00E+00
-8,00E-01
-6,00E-01
-4,00E-01
-2,00E-01
0,00E+00
2,00E-01
4,00E-01
10 15 20 25 30 35 40[ns]
[V]
150 uA pp280 uA pp420 uA pp560 uA pp700 uA pp840 uA pp1.67 mA pp2.48 mA pp3.35 mA pp4.11 mA pp5.49 mA pp6.8 mA pp8.21 mA pp9.6 mA pp11 mA pp
PACTA low gain output for different input pulse amplitudes
-7,0E-01
-6,0E-01
-5,0E-01
-4,0E-01
-3,0E-01
-2,0E-01
-1,0E-01
0,0E+00
1,0E-01
10 12 14 16 18 20 22 24 26 28 30
[ns]
[V]
150 uA pp280 uA pp420 uA pp560 uA pp700 uA pp840 uA pp1.67 mA pp2.48 mA pp3.35 mA pp4.11 mA pp5.49 mA pp6.8 mA pp8.21 mA pp9.6 mA pp11 mA pp
1919
VII. PACTAv1.1 chip test results
Pulse shape (normalized waveforms)
Normalized input signal
0,00
0,10
0,20
0,30
0,40
0,50
0,60
0,70
0,80
0,90
1,00
10 12 14 16 18 20 22 24 26 28 30[ns]
-1,49E-04-2,79E-04-4,17E-04-5,58E-04-6,97E-04-8,38E-04-1,67E-03-2,48E-03-3,35E-03-4,11E-03-5,49E-03-6,80E-03-8,21E-03-9,60E-03-1,10E-02
PACTA normalized high gain output for different input pulse amplitudes
-0,4
-0,2
0
0,2
0,4
0,6
0,8
1
1,2
10 15 20 25 30 35 40
[ns]
150 uA pp280 uA pp420 uA pp560 uA pp700 uA pp840 uA pp1.67 mA pp2.48 mA pp3.35 mA pp4.11 mA pp5.49 mA pp6.8 mA pp8.21 mA pp9.6 mA pp11 mA pp
PACTA normalized low gain output for different input pulse amplitudes
-2,00E-01
0,00E+00
2,00E-01
4,00E-01
6,00E-01
8,00E-01
1,00E+00
1,20E+00
10 12 14 16 18 20 22 24 26 28 30
[ns]
150 uA pp280 uA pp420 uA pp560 uA pp700 uA pp840 uA pp1.67 mA pp2.48 mA pp3.35 mA pp4.11 mA pp5.49 mA pp6.8 mA pp8.21 mA pp9.6 mA pp11 mA pp
2020
VII. PACTAv1.1 chip test results
Frequency response Small signal BW exceeds 500 MHz both for high gain and low gain Minimize peaking for LG in final version
Gain for different signal levels (Low Gain)
0,00
10,00
20,00
30,00
40,00
50,00
60,00
70,00
10,00 100,00 1000,00
Frequency [MHz]
Gai
n [
dB
Oh
m]
66 uApp
133 uApp
266 uApp
533 uApp
750 uApp
Gain for different signal levels (High Gain)
0,00
10,00
20,00
30,00
40,00
50,00
60,00
70,00
10,00 100,00 1000,00
Frequency [MHz]
Gai
n [
dB
Oh
m]
33 uApp
75 uApp
110 uApp
160 uApp
230 uApp
21
Transimpedance gain (charge)
1,0E-12
1,0E-11
1,0E-10
1,0E-09
1,0E-08
1,0E-07
1,0E-06 1,0E-05 1,0E-04 1,0E-03 1,0E-02 1,0E-01
Input peak current [A]
Inte
gra
l o
f th
e o
utp
ut
pu
lse
[V
s]
High Gain
Low Gain
21
VII. PACTAv1.1 chip test results
Transimpedance gain (integral of the pulse) and linearity HG about 1 KOhm LG about 50 Ohm Relative non-linearity error < 2 %
100x(Meas-Fit)/Fit
1 phe
100 phe
Transimpedance gain (charge)
0,0E+00
1,0E-09
2,0E-09
3,0E-09
4,0E-09
5,0E-09
6,0E-09
7,0E-09
8,0E-09
0,0E+00 5,0E-03 1,0E-02 1,5E-02 2,0E-02 2,5E-02 3,0E-02
Input peak current [A]
Inte
gra
l o
f th
e o
utp
ut
pu
lse
[V
s]
High Gain
Low Gain
Relative error of the integral of the output pulse
-10
-8
-6
-4
-2
0
2
4
6
8
10
1,0E-06
Input peak current [A]
Re
lati
ve
lin
ea
rity
err
or
[%]
High Gain
Low Gain
2222
VII. PACTAv1.1 chip test results
Dedicated board with PMT (R81619mod) Additional voltage gain (14.4 V/V)
Total gain TIA gain is 14.4 KOhm To minimize the impact of following stages on the input ref. noise
Differential to single ended conversion + 50 Ohm driver Lot of work on grounding and shielding Many thanks to:
R. Mirzoyan, D. Fink, P. Nayman and F. Toussenel
x14
x14LG
PACTA
Drv
Drv
HG 50 Ohm
50 Ohm
3.3V
1.8V
LDO
PM + Divider
100 nF
Prot100 nF
Diff to single ended+
50 Ohm driver
2323
VII. PACTAv1.1 chip test results
Single photoelectron spectra at 900 V PMT at 900 V, gain about 4x104
New analysis method (R. Mirzoyan) with less systematic under development Integration time: 10 ns ENC of about 5000 electrons (11.8 pVs / q x 14.4 K) S/N = 8
PACTA + 20GS/s DPO Scope PACTA + NECTAr0 chip (see poster session)
2424
VII. PACTAv1.1 chip test results
Input referred noise as function of the integration time Theoretical and measured noise (with and without PMT) Still some few hundred e contribution of pick up noise Differential configuration to minimize CM noise
With single ended configuration theoretical ENC should be 1/sqrt(2)
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Integration time [ns]
EN
C [
e]
PM + PACTA +THS (HV 900V)
PM + PACTA + THS (HV OFF)
PACTA +THS (no PMT)
Theo. Parallel (white) (PACTA)
Theo. Parallel (white) (PACTA + THS)
Theo. Par+ Ser(HF) + 1/f (PACTA)
Theo. Par+ Ser(HF) + 1/f (PACTA+THS)
2525
VII. PACTAv1.1 chip test results
Preliminary tests with SiPM Low Zin current mode circuit are well suited for SiPM readout
DC coupling without external components We just took an available MPPC (S10931-050P)
1 V overvoltage Recovery time seems to be dominated by internal SiPM time
constant · DC coupled· Possible to ctrl each
SiPM bias with on-chip circuitry
LG
PACTA
HG
100 nF
SiPM
10 K
Vb
- HV
Vop=Vb-HV
20 ns
2626
VIII. Conclusions
• Wideband and high dynamic range current mode preamplifier for CTA cameras
• PACTAv1.1 meets most of the requirements :– Input Referred Noise< 400 nA rms– SNR for SPE spectra: 8, at the nominal PM gain (40K)– Input range: > 20 mA peak– Dynamic range: 15.9 bits– Relative linearity error for charge measurements: < 2 %– BW: 500 MHz. Both for HG and LG– Input impedance: 10 to 15 Ω. For full BW.
• A cable / tline driver has been implemented in PACTAv1.2 – Single ended and differential versions– Power consumption: 150 mW
• Factor 3 or 4 smaller than current prototypes build with COTS
• New versions of the circuit for SiPM readout are under development
– Potentially, good time resolution
262626
TWEPP – Vienna – September 27th 2011 D.Gascón
2727
Back-up
2828
V. Noise: noise variance at the output of the system
• Noise PSD at the output depends on the transfer f unction of the system:
• The noise power is obtained integrating the output PSD:
• I n HEP the noise is usually studied in time domain (useful).• Noise process (white): series of random (Poisson) Dirac
impulses (t).• Noise weighting f unction: contribution at the output at the
measurement time to of noise impulse at t i:
– For time invariant shaper it is the mirror of the impulse response:
• The noise variance (or power) at the output is:
2 22 2 2
0 0 0y yy x xE G f df G f S f df G f e f df
2
yy xxG f H f G f
2 2 20 0
1 1
2 2G w t dt G h t dt
0
0 , i i t tw t t h t t h u
0 , iw t t
2 20 0 0
1,
2 i it G w t t dt
f or time invariant:
2929
V. Noise: time domain analysis
• Time variant system: compute w(to,t i): use definition!• Pre-shaper f unction p(t): source impedance and preamplifier!
– Diff erent f or series and parallel noise– But all fi rst order system with < 4 ns.
• w(to,t i) at the end of integration (to=t1+tR):– Shaded area of the impulse arriving at t i.
• Analytical expression f or our system (tR=T):1.Noise impulse arriving af ter end of integration (ti>t1+T ):
w(to=t1+T,t i)=0.
2.Noise impulse before start of integration (ti t1 ):
3.Noise impulse af ter start of integration (ti > t1 ):
-60 -40 -20 20
0.2
0.4
0.6
0.8
1w(ti)
ti [ns]
RR
tQ CT
time-invariant pre-shaper
p(t)b
aA gated
integratorp(t)
p
R
p t1 t1+ R
ti=t1-tp
ti=t1+tR
p
pW(to=t1+ tR,ti)
to=t1+ tRti
ti
ti=t1
1 1
1
10 1 1,
i ii
i
t t t T tt T t
i it ti i
A Aw t t T t p x dx e e t t
1
1
0 1 1 10, 1
ii
t T tt T t
i ii i
A Aw t t T t p x dx e t t t T
1( )
t
p t A e u t
< 4 ns
10 ns 5 ns 1 ns 0,5 ns
0,1 ns
w(t0,ti) fordifferent
TIME [ns]
3030V. Noise: time domain analysis
• The noise variance is:
• For T>>:
2
2 20 1
11
2
T
white niwhitei
At t T e T e
2
2 20 1
1
2white niwhiteTi
At t T e T
Approximation error as
function of
3131II. Noise requirements
Good single photoelectron resolution: S/N > 10 in the charge spectra
How to translate to typical specification for amplifiers? Series (en) and parallel (in) noise power spectral densities
Assumptions Flicker noise is negligible Current preamp: RPM open Voltage preamp: ZT resistive (RPM) CPAR is small enough: series noise negligible for I amp RPM is small enough: parallel noise negligible for V amp
Noise variance at the output of gated integrator is (approx): T is the integration time
2 2 21
2no niG Te
ZT
eZT
RSeRs
en
in Zi
Iin
IPMT
I preamp: VO=ZT·Iin
V preamp: VO=G·Vin
Vin
ZT = RPM // CPAR
2 2 21
2no T niZ Ti V preamp: I preamp:
32
0,00E+00
5,00E-12
1,00E-11
1,50E-11
2,00E-11
2,50E-11
3,00E-11
3,50E-11
5 10 15 20 25 30
in [A
/sqr
t(H
z)]
Integration time T [ns]
Low gain (40K)
High gain (100K)
32II. Noise requirements
For a voltage preamplifier, the signal at the output of gated integrator is Zi is the PM load impedance
And the S/N is
Max input referred noise of the amplifier (en):
For a S/N > 10
2· ·
·i phe
Vni
Z QSN T e
2· ·
·
i pheni
MIN
Z Qe
ST N
0
2E-10
4E-10
6E-10
8E-10
1E-09
1,2E-09
1,4E-09
1,6E-09
1,8E-09
5 10 15 20 25 30
en [V
/sqr
t(H
z)]
Integration time T [ns]
Low gain (40K)
High gain (100K)
Voltage amplifier Current amplifier
< 1 nV/sqrt(Hz) ! < 10 pA/sqrt(Hz) !
· ·o oA iA i iA iS v dt Gv dt G Z i dt Z G Q
3333III. The ATLAS LAr preamplifier
Low noise Series noise: en=0.36 nV/sqrt(Hz) Parallel noise: in=6.7 pA/sqrt(Hz) Low gain option: 40 K
Low noise preamplifier is needed
High dynamic range: about 14 bits
Super-common base: Small input impedance Photo-detector current is sensed
3434IV. Measurements
Single photoelectron spectra
Pulser LED @ 460 nm < 500 ps FWHM
Afterpulsing observed
LASER @ 640 nm < 50 ps FWHM
Optical attenuator
Hamamatsu R7600 PM To be repeated with R9420
3535IV. Measurements
Single photo-electron spectra @ HV = 500 V : 40 K gain
• Preamp noise: 3-4 Ke, limited by common mode and pick-up
noise
PM HV500 VSingle
photoelectron
3636IV. Measurements
Effect of the integration time
0,00E+00
1,00E+03
2,00E+03
3,00E+03
4,00E+03
5,00E+03
6,00E+03
5 10 15 20 25 30
Integration time [ns]
No
ise
[e]
Measured Noise [e]
Theoretical White Noise [e]
3737
IV. Measurements: conclusions and plans
• Super common base architecture is promising– Low noise
– High dynamic range
– Low input impedance
• However covering full dynamic range with a single preamplifier is still difficult:
– Preamp saturation (for LAr preamplifier) • 5 mA
– Max signal (6 Kphe) • 20 mA peak current @ 40 K gain
37
3838
VII. PACTAv1.1 chip test results
Low inductance QFN socket Test PCB with socket for characterization Noise/single phe measurement:
Dedicated PCB with Additional gain 50 ohm drivers
Test set-up:• Agilent 81155A pulse generator
• HP RF signal generator (DC-1GHz)
• Picoquant Laser Pulser (50 ps FWHM)
• Tektronix TDS7154B scope: • 1.7 GHz
• 20 GS/s
• Active differential probe: 1 GHz
39
Transimpedance gain (peak to peak)
1,00E-05
1,00E-04
1,00E-03
1,00E-02
1,00E-01
1,00E+00
1,00E+01
1,0E-06 1,0E-05 1,0E-04 1,0E-03 1,0E-02 1,0E-01
Input peak current [A]
Out
put
peak
vol
tage
[V
]
High Gain
Low Gain
39
III. PACTAv1.1 chip test results
Transimpedance gain (amplitude) and linearity HG about 1 KOhm LG about 50 Ohm Relative non-linearity error < 3 %
100x(Meas-Fit)/Fit
1 phe 100 phe
Transimpedance gain (peak to peak)
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1,6
1,8
2,0
0,0E+00 5,0E-03 1,0E-02 1,5E-02 2,0E-02 2,5E-02 3,0E-02
Input peak current [A]
Ou
tpu
t p
ea
k v
olt
ag
e [
V]
High Gain
Low Gain
Relative error of the amplitude of the output pulse
-10
-8
-6
-4
-2
0
2
4
6
8
10
1,0E-06 1,0E-05 1,0E-04 1,0E-03 1,0E-02 1,0E-01
Input peak current [A]
Re
lati
ve
lin
ea
rity
err
or
[%]
High Gain
Low Gain
4040
III. PACTA chip test results
PMT signal shape
2.1 ns
PACTA OutputHV=1200 V
PMT directly to scope
PACTA OutputHV=900 V
4141
III. PACTA chip test results
Gain is calibrated for single photoelectron spectra, comparing: Single photoelectron signals with no preamp, direct to scope (only high
gain) Single photoelectron with PACTA Result is quite close to what we expect: 14.3 Kohm
1,00E+04
1,00E+05
1,00E+06
1,00E+07
100 1000 10000
HV [V]
R8
61
9 m
od
ga
in
PM to Scope
PM + PACTA
0,00E+00
1,00E+05
2,00E+05
3,00E+05
4,00E+05
5,00E+05
6,00E+05
7,00E+05
8,00E+05
9,00E+05
1,00E+06
1,10E+06
1,20E+06
1,30E+06
1,40E+06
1,50E+06
850 950 1050 1150 1250 1350 1450 1550
HV [V]
R8
61
9 m
od
ga
in
PM to scope
PM+PACTA
4242
III. PACTA chip test results
S/N ratio for single photoelectron measurements Optimal integration time is the which maximizes S/N: about 10 ns
0,00
1,00
2,00
3,00
4,00
5,00
6,00
7,00
8,00
9,00
10,00
0 5 10 15 20 25 30
Integration time [ns]
S/N
PACTA +THS (No PMT)
PM + PACTA + THS (HV OFF)
PM + PACTA + THS (HV 915 V)
Theo. (PACTA)
Theo. PACTA + THS)
4343
III. PACTAv1.1 chip test results
Preliminary tests with SiPM
Bi-gain is also working
4444
III. PACTA chip test results
Preliminary tests with SiPM
Charge spectrum
4545
PACTA Connections
Different scenarios: location and connection to FE PMT cluster as single board
No pb to place preamp very close to PMT (< 3 cm) PACTA to FE connection:
Differential impedance controlled PCB traces Cheap and robust
PMT cluster in several boards: PMT – cable - PACTA PACTA Zin is 10-20 Ohm: no adapted If distance between PMT and PACTA increases: EMC…
Preliminary test shows than 10 cm could be ok What is the min length that cluster design can achieve?
PACTA to FE connection: Differential impedance controlled PCB traces
PMT cluster in several boards: PMT – PACTA – cable(s)
PACTA can be very close to PMT but… Room for PACTA + test pulse + monitoring … ???
+ CABLES and connectors ??? How many ? NOT RECOMENDED
PMT
PACTA FE
PM cluster in single board
< 3 cm
PMT
PACTA FE
PM cluster in several boards (A)
Length ???
CablePCB trace
PMT
PACTAFE
PM cluster in several boards (B)
How many cables ???
4646
IV. PACTAv1.2
• A fully differential TIA and 2 single ended TIAs Each of them with HG and LG outputs
• New 50 Ohm drivers have been integrated Many thanks to J. Lecoq (LPC), E. Delagnes (CEA/Saclay) and P. Moreira (CERN) Class AB push-pull follower with fast local feedback Dynamic range to directly match std differential ADCs:
1 V for single ended version 2 Vpp for fully differential version
Total power consumption : 120 – 150 mW For 2 (High and Low gain) differential outputs To compare with COTS solution
Preamp FE amplifiers: double gain + level adaptation Power consumption reduction: 500 mW /ch (aprox)
• Single ended input stage for minimal noise According to simulations 1/1.2 wrt fully differential TIA
Ideally should be 1/sqrt(2) Common mode noise (bias current sources, etc)
• Minor changes on the current mode input stage: Minimize high frequency peaking Closer to a first order system response
46
PACTA1.2 chipSiGe BiCMOS 0.35umAMS 2 mm2
QFN32 packageSubmitted on June 6th