Qualification of a HV-MAPS produced by TSI
Christoph Blattgerste for the Mu3e collaboration
Heidelberg University
26.03.2019
MuPix7
MuPix7 is first fully monolithic HV-MAPS developed for Mu3e Well characterized prototype produced by AMS
40×32 pixels with a size of 103×80µm
Electronics
Amplifier in each pixel
Conversion analog↔ digital at comparator in periphery cell Digital electronics and state machine in periphery
Motivation
AMS process no longer supported & delivery delays
→ Change to TSI HV as manufacturing process?
↔
AMS H18 andTSI HV based on same IBM HV process Identical design in 180nm HV-CMOS
Expect no significant differences between AMS/TSI beside statistical variations
Setup
MuPix7 sensors bonded on insert (5 AMS & 4 TSI) Mupix8 v2 PCB with MuPix7 insert (printed circuit board) Data readout via 8b/10b encoded LVDS signal with Stratix IV FPGA
Breakdown Voltage
I-V-curve for reverse biased diodes
Clear difference between foundries
Higher breakdown allows for higher HV
Hint for TSIs quality
90 91 92 93 94 95 96 97
HV [-V]
102 101 100 101 102 103 104
diode current [nA]
300-1-4 AMS 300-1-5 AMS 300-1-6 AMS 300-1-8 AMS 300-1-9 AMS 300-1-2 TSI 300-1-3 TSI 300-1-7 TSI 300-1-10 TSI
VCO scan
VCO
ChargePump TOVCO VPVCO
VNVCO
Patch signal Reference signal VCO
CLOCK
Divider Dig. Part
Serializer
Readout Timestamps CLOCK
CLOCK/5 SelSlow
CLOCK/10
VPPump
PD
Same shape for TSI and AMS
TSI slightly less jitter (not significant)
VCO scan
VCO
ChargePump TOVCO VPVCO
VNVCO
Patch signal Reference signal VCO
CLOCK
Divider Dig. Part
Serializer
Readout Timestamps CLOCK
CLOCK/5 SelSlow
CLOCK/10
VPPump
PD
Same shape for TSI and AMS
TSI slightly less jitter (not significant)
same DAC value suitable
for TSI chips 0 5 VPVCO [dec]10 15 20
40 60 80 100 120
jitter [ps]
standard value 300-1-5 AMS 300-1-8 AMS 300-1-9 AMS 300-1-2 TSI 300-1-3 TSI
Time resolution
Latency: Hit TS - Trigger TS
No significant difference Right tail due to
time-walk
Latency [8ns]
−60 −40 −20 0 20 40 60 80 100
Normalized Entries [1/8ns]
0.2 0.4 0.6 0.8
1 AMS
TSI
= 20.82 [ns]
σ
= 20.71 [ns]
σ
Power consumption
VDD: digital part VDDA,VSSA: analog pixels
Most power for digital part
Big variances forVDD
VDD VDDA VSSA total
measured voltage 10
20 40 60 80 100120
power consumption [mW]
300-1-4 AMS 300-1-5 AMS 300-1-6 AMS 300-1-8 AMS 300-1-9 AMS
300-1-2 TSI 300-1-3 TSI 300-1-7 TSI 300-1-10 TSI
Power consumption
TSIVDD about 25%
higher than AMS TSIVDDA,VSSA about 10% higher than AMS TSI has higher power consumption overall
VDD VDDA VSSA total
measured voltage 10
20 40 60 80 100120
power consumption [mW]
300-1-4 AMS 300-1-5 AMS 300-1-6 AMS 300-1-8 AMS 300-1-9 AMS
300-1-2 TSI 300-1-3 TSI 300-1-7 TSI 300-1-10 TSI
Power consumption for important DACs
DAC current (AMS) [mA] current (TSI) [mA] TSI/AMS
VN2 3.9 4.4 1.13
VNLVDS 8.0 19.1 2.39
VPComp 3.8 4.4 1.16
VPDac 11.5 12.4 1.08
Power consumption for important DACs
DAC current (AMS) [mA] current (TSI) [mA] TSI/AMS
VN2 3.9 4.4 1.13
VNLVDS 8.0 19.1 2.39
VPComp 3.8 4.4 1.16
VPDac 11.5 12.4 1.08
LVDS output
Need for high resolution spectrum analyzer
“Eye” as overlap of differential signals VNLVDS affects eye height only
eye height measured correctly
LVDS scan
Significant difference AMS/TSI
higher “eye” for TSI chips
→ 4×better signal quality for TSI
0 10 20 30 40 50
DAC [dec]
0 20 40 60 80 100
eye height [mV]
default DAC 300-1-7 TSI 300-1-4 AMS
LVDS power consumption
Significant difference AMS/TSI
better signal quality for TSI needs higher power
→ TSI generates higher signal for same power
0 10 20 30 40 50
DAC [dec]
1 2 3 4 5 6 7
eye height/current [V/A]
default DAC 300-1-7 TSI 300-1-4 AMS
Summary
comparison
Area AMS TSI
Breakdown HVAMS ≈93V HVTSI ≈96V PLL µAMS ≈42.0ns µAMS ≈35.0ns Time res. σAMS ≈20.7ns σTSI ≈20.8ns Power PAMS ≈105mW PTSI ≈129mW LVDS VNLVDSAMS = 15 VNLVDSTSI = 2
Summary
comparison
Area AMS TSI
Breakdown HVAMS ≈93V HVTSI ≈96V PLL µAMS ≈42.0ns µAMS ≈35.0ns Time res. σAMS ≈20.7ns σTSI ≈20.8ns Power PAMS ≈105mW PTSI ≈129mW LVDS VNLVDSAMS = 15 VNLVDSTSI = 2
results
Difference only in power consumption
→solvable with lower DAC VNLVDS for TSI chips
Outlook
future measurements
Signal-to-noise (SNR)
Efficiency measurements from test-beam Evaluation of irradiated chips
Backup slides
Mu3e
Search for decayµ+→e+e−e+ → cLFV Exclude other decays e.g. µ+→e+e−e+νeνµ
P~p= 0 & P mµc2
multiple layers with recurl→ vertex/momentum resolution
Detector
Pixel layers provide good vertex resolution
Scintillating fibers & tiles with precise timing resolution Recurl through applied magnetic field→ momentum measurement
Diodes profile
larger depletion zone for stronger electric field depletion over whole diodes profile
HV scan
scan HV applied to each pixel higher HV→ more hits higher HV→ less noise
0 20 40 60 80
high voltage [-V]
0 1 2 3 4 5 6 7
rate [kHz]
chip:
300-1-2 TSI noise
Fe55
Pulse shape
Detector temperatures Tmin= 0◦C,
Tmax = 70◦C Pulse shape changes with temperature Measurements show temperature dependency
→ DAC settings concerned for different
temperatures 0 200 400 600 800
time [ns]
620 640 660 680 700 720 740
threshold [mV]
chip300-1-3 TSI pulse shape for 0.75V injection
0°20°
40°60°
80°
Injection
quantify injection and source
injection histogram is gaussian
Fe55 histogram is gaussian with left tail tail caused by hits between pixel cells
0 100 200 300 400 500 600 700 800
ToT [ns]
0.0 0.2 0.4 0.6 0.8 1.0 1.2
counts [#]
1e4
gaussian mean = 465.1±2.724 Fe55 mean = 398.4±14.64
injection U=0.75V Fe55
Injection
varying injection creates different pulse heights readout electronics saturate for higher voltages
0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 injection [V]
0 500 1000 1500 2000 2500 3000
ToT [ns]
300-1-9 AMS 300-1-2 TSI
0 1987 injection [#3975 e] 5962 7950
Injection
varying injection creates different pulse heights readout electronics saturate for higher voltages
0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 injection [V]
0 500 1000 1500 2000 2500 3000
ToT [ns]
300-1-9 AMS 300-1-2 TSI
0 1987 injection [#3975 e] 5962 7950
Injection
difference due to statistical variations different absolute values unimportant
0.2 0.3 0.4 0.5 0.6 0.7
injection [V]
0 200 400 600 800 1000 1200 1400
ToT [ns]
Fe55 peak 300-1-4 AMS 300-1-5 AMS 300-1-6 AMS 300-1-8 AMS 300-1-9 AMS 300-1-2 TSI 300-1-3 TSI 300-1-7 TSI 300-1-10 TSI AMS mean TSI mean
0 1987 injection [#3975 e] 5962 7950
VPPump scan
exponential shape same for TSI and AMS TSI slightly less jitter (not significant) DAC value are suitable
0 10 20 30 40 50 60
VPPump [dec]
40 60 80 100 120
jitter [ps]
default value 300-1-5 AMS 300-1-8 AMS 300-1-9 AMS 300-1-2 TSI 300-1-3 TSI
Time resolution
Low delay
High time resolution AMS chips perform better
no significant differences
300-1-2 TSI300-1-3 TSI300-1-7 TSI300-1-10 TSI300-1-4 AMS300-1-5 AMS300-1-6 AMS300-1-8 AMS300-1-9 AMS sensor[#]
2 0 2 4 6 8 10 12
delay [ns]
mean AMS mean TSI
Time resolution
Low delay
High time resolution AMS chips perform better
no significant differences
300-1-2 TSI300-1-3 TSI300-1-7 TSI300-1-10 TSI300-1-4 AMS300-1-5 AMS300-1-6 AMS300-1-8 AMS300-1-9 AMS sensor[#]
22.0 22.5 23.0 23.5 24.0
Time resolution [ns]
sigma AMS sigma TSI