Download calib design review - uni

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts

Printed circuit board wikipedia , lookup

Transmission line loudspeaker wikipedia , lookup

Transcript
ATLAS
Status of 8ch calib prototype
N. Dumont-Dayot, G. Perrot, P. Perrodo, I Wingerter-Seez
LAPP Annecy
U. Schafer, D. Schroff,
Univ Mainz
C. de La Taille, N. Seguin-Moreau, L. Serin
LAL Orsay
8ch prototype status
 Slice of 8 analog channels + complete digital part
 3 boards received in apr 02

2 in Orsay (-> 1 to Mainz), 1 in Annecy
SPAC2
Calolgic
TTCRx
Delay
8 outputs
21 jun 2002
DAC
Opamps
& switch
C. de La Taille
calibration board design review
2
8channel prototype chips
 DAC V2

Improved reference
 123 chips received may 02
 110 work OK
 QFP100
 Opamp V2

offset trimmed to 15 µV
 HF switch integrated
 463 chips available QFP44
 Calogic V2

40 chips available in QFP100
 Tested to SEE in Louvain
21 jun 2002
C. de La Taille
calibration board design review
3
Digital part
 Worked OK at once

TTCRx + Calogic + Delay chip + SPAC3
 Jitter visibly improved
21 jun 2002
C. de La Taille
calibration board design review
4
DC uniformity
 Correction of output line resistance
Rout (Ω)

t = 35 µm R = 0.5 mΩ/•
 ΔL = 17 cm W = 200 µm
 Expect : ΔRline = 0.4 Ω
 ΔVout = 0.8 % effect
 Correction by ΔR (5Ω)
10*ΔR (5Ω)
decreasing Iout by increasing R(5Ω)
 Exact correction for W = 2mm : ΔR5 =40 mΩ
 Measured value = 60 mΩ -> overcompensation

output layer
CH0
CH7
VP6 layer
21 jun 2002
C. de La Taille
calibration board design review
5
DC uniformity
 DC current output

1.4% variation between CH0 and CH7
 Consistent with +5.5V additionnal resistance
 5 Ω resistors
All in ± 0.1%
 Rms = 0.04%
 2 resistors where low by 1% : DAMAGED ?

21 jun 2002
C. de La Taille
calibration board design review
6
Lots of unexpected problems…
 Board very different from mod0

Channels no longer aligned but staggered in depth
 Lots of problems with the 8 channels











21 jun 2002
Ground bounce
2V change with enabled channels
80 µV DAC offset
DAC change with all channels on
Overshoot
Signal uniformity
DC uniformity
Damaged chips
Oscillations
Ripple noise
Linearity
C. de La Taille
calibration board design review
7
Example : Ground bounce
 Observation





100 ns bounce on small signals
Worst on far channels (->7)
Clean on channel 0
Present even with collector open
Dissapears if termination resistor (R0)
removed
 Explanation

Large digital current (200 mA) in
inductive ground
 Produces 1 mV ground bounce, coupled
to the output through R0
 Cure

Try to reduce digital current (difficult)
 Move R0 in quiet zone near inductor
 Minimize loop area around NE856
21 jun 2002
C. de La Taille
calibration board design review
8
Parasitic injected charge (PIC)
 Parasitic signal when DAC=0
Proportionnal to CGS and ΔVGS
 ringing with parasitic inductance (L ~ 3 nH)
 Reducing VG can reduce the PIC

• Effect very visible, scaling with VGS
• Below VG= - 0.4V, some PMOS are not
completely OFF : not acceptable

Improved by connecting the nwell to the source
 Referred to DAC value

Peak of PIC after shaping
 1 V DAC makes 5 V output pulse
 1 GeV (η=0,MID) corresponds to DAC=500 µV
L = 0, 22 nH, 100 nH
21 jun 2002
C. de La Taille
calibration board design review
9
Parasitic injected charge (PIC)
 Parasitic injected charge extraction

DAC set at \x0004 = 64 µV
 Signal due to offset needs to be substracted
21 jun 2002
C. de La Taille
calibration board design review
10
Parasitic injected charge (PIC)
 Parasitic injected charge very uniform on 8 channels

Equivalent to DAC=0006=100 µV (180 MeV in the barrel middle) in peak
 Subsequently reduced by a factor ~3, very close to noise level (50MeV)
 Well differentiated
Injected charge x 10
= 500 µV pulse
~DAC=100µV
21 jun 2002
C. de La Taille
calibration board design review
11
Linearity
 Measured on each gain
1-10-100
 Pulse measurements

In red
 After shaping
 DC current measur.

In black
 With Keithley
 Several problems

DAC referenced to
VP6
 Bad 5Ω resistor brand
 Well below 0.1%
21 jun 2002
C. de La Taille
calibration board design review
12
Linearity
 Fit residuals show opamp offset

Very small contribution of parasitic injected
charge
Offset of High gain range in
agreement with low offset
Opamp measurement
Effect too small to be seen
in other ranges
21 jun 2002
C. de La Taille
calibration board design review
13
Linearity
 Fit slopes give uniformity

Very small contribution of parasitic injected
charge
Slope different by 5 % in high
gain range due to attenuator
21 jun 2002
C. de La Taille
accuracy
= 0.3 % but 5 V stripline
compensation deteriorates
calibration boardthe
design uniformity
review
14
Pulse uniformity
1.5 % effect from 5.5 V
compensation foreseen when 50 
near switch
0.6 % from different output lenghes
from 50  to connector (on 128
boards connector position will add
a few per mille
5  resistor at 0.1 % as expected
slope uniformity 0.3 %
21 jun 2002
C. de La Taille
What should we compensate
on 128 boards ?
calibration board design review
15
Oscillations
 Observation 1

Large oscillation at 350 MHz when channels 2,3 or 4 are pulsed together with
DAC>0.25V
 Barely visible oscillation at 350 MHz on all the channels, increasing with DAC>0.25V
 Was not present on single channel test-board
 Explanation 1

None (as usual for HF oscillations)
 Cure 1

Add 150 pF to ground on Cp (gate of current source output transistor)
 Observation 2

180 MHz oscillation on large pulses, alters decay time
 Explanation 2

Delay chip loaded by coax for triggering purposes
 Cure 2

21 jun 2002
Add 50 Ω in series for trigger output
C. de La Taille
calibration board design review
16
Damaged opamps
 Observation





+ 2V moves up to 5.5 V
Some chips (4/18) show large current (5-50mA) from 7.5V to +2V
They still work correctly, but with degraded offset
All chips in parallel on supplies, common 100 Ω protection resistor on +2 V
Difficult to locate bad chips
 Explanation

None, could only be reproduced with 7.5V raised to 15V
 Several power supplies mishandlings have been tried without damage
 Cure

Individual current limitation resistors on +2V and +7.5V
 Series resistors on input and common voltage busses (Vref)
21 jun 2002
C. de La Taille
calibration board design review
17
Damaged DAC
 Observation

1 chip damaged on brd 3
 DAC = 0 gives 2 mV
 4 µA leakage current on MSB
 Explanation

None
 Chip could not be remounted on DAC test board (pins damaged)
 Cure

21 jun 2002
To be examined
C. de La Taille
calibration board design review
18
Enable schematic
 Common emitter between VDD and VSS
VDD

Channels enabled flow 1 mA in VSS
 Channels ON by default (Enable=VDD)
100K
Calogic
0-5V
Enable
2-7V
5K
VSS
 Go to common base configuration

Powered between VDD and GND
 No current in VSS
 Base connected to VCC through 47 kΩ
 Channels OFF by default
VDD
100K
50K
Calogic
0-5V
Enable
2-7V
50K
VCC
21 jun 2002
C. de La Taille
calibration board design review
19
Robustness

Power supplies

Effect of missing +7.5V : maximum current flowing (0.3A/ch) -> add 1MΩ to 5.5V on
output transistor
 Effect of missing +9.5V : maximum current flowing (0.3A/ch) -> Produce +7.5V from
+9.5V
 Effect of missing +5.5V : none.
 Effect of missing -6 V : barely visible, can be almost any (negative) value
 Tests of robustness

The chip starts to deteriorate at VDD = 12 V
 Permanent damage in protection diodes, if unlimited current for V DD = 14 V
 Offset variations

Observation : offset changes by 40 µV when all channels are ON
 Explanation : Offset sensitivity to Vss power supply : dVoff/dV2V = 100 µV/V = 0.01%
 Cure :
• Reduce +2V impedance by decreasing resistance in series with the 3 diodes
• Maintain robustness by individual large resistive decoupling
• Use regulator on +7.5V
21 jun 2002
C. de La Taille
calibration board design review
20
Summary of proposed modifications
 General

Power supplies reduced by 0.5V to VP6=+5V, VDD=+7V and VP9=+9V
 Individual channels











21 jun 2002
R0 = 50 Ω moved away from switch, close to inductor
Output line impedance decreased to 25 Ω
Bulk connected to source instead of VDD via 1 MΩ, 100 nF decoupling
Gate connected to GND, no VG adjustment
150 pF to ground on Cp and 1 MΩ to VDD
1 kΩ in series with Vss = + 2 V (I = 100 µA)
200 Ω in series with fuses to Vss (I = 2 mA)
100 Ω in series with Vdd = +7V (I = 2.1 mA)
Enable translator changed from common emitter to common base configuration, with
no connection to Vss
Short path between emitter of switching NE856 and decoupling capacitor of preceding
emitter follower.
Series resistor of follower increased from 51 Ω to 82 Ω (910 Ω was really too much !)
C. de La Taille
calibration board design review
21
Summary of proposed modifications
 Groups of 8 channels





Remove 50 Ω parallel termination on command line
Change 0 Ω series termination to 50 Ω // 33 pF
Add fuse on VP6 if RFUSE < 50 mΩ
Have local +2V with 3 diodes
Keep possibility of local VG = -0.4 V ?
 DAC

Correct star disribution for VP6 on resistor ladder
 Keep bandgap and reference possibility
 DAC distribution





21 jun 2002
150 pF to ground on Cp and 1 MΩ to Vdd
Connect drain to ground via 1 k instead of using PMOS switch
1 kΩ in series with Vss = + 2 V (I = 100 µA)
200 Ω in series with fuses to Vss (I = 2 mA)
100 Ω in series with Vdd = +7V (I = 2.1 mA)
C. de La Taille
calibration board design review
22
Pending questions
 Digital part

Clock terminations
 Reset scheme
 Regulator on +5V

Maximum current : 3A =>
• All channels ON with DAC=100mV (1/10 full scale)
• 10 channels with maximum DAC

Filtering
 Use of Bandgap reference

Not yet tested
 Schedule for opamps production

Issue order in june ?
 How many wafers : 12, 13, 14 ?
21 jun 2002
C. de La Taille
calibration board design review
23
Pulse shape before shaping
 Full DAC range

100 µV-> 1V
 Up to 5V pulses in 50 Ω
 Undershoot
DAC=100µV
Due to 50 Ω line between
switch and R0 : should be
25 Ω
 Checked by simulation
 Will be corrected

DAC=1mV 0dB
DAC=10mV 0dB
DAC=0.1V -20dB
 Risetime

< 2 ns
 Varies little with DAC
21 jun 2002
DAC=1V -40dB
C. de La Taille
calibration board design review
24
Pulse shape after shaping
 80 dB dynamic range
 Parasitic injected charge

Equivalent to DAC=30 µV
 Or a pulse of 150 µV
 At peak < 15 µV
 Improvement by >10
compared to module0
DAC=0µV
DAC=100µV 0dB
DAC=1mV -20dB
DAC=1V -80dB
21 jun 2002
C. de La Taille
calibration board design review
25
PIC uniformity
 PIC on 8 channels

Good uniformity
 Singularities :

CH7 had the Nwell tied to
5V, as in the original
configuration : clear
improvement !
 CH4 has a larger offset (35
µV) damaged ?
CH7 : VB=+5V
21 jun 2002
C. de La Taille
calibration board design review
DAC=1mV -20dB
26
Designing the 128ch board
 Large number of modifications !

108 modification on the 8 ch. Board alone
 Improved performance and robustness
 Power supplies decreased by 0.5V
 Goal : have VP6 = 5V uniform for all
channels within 0.1%
5V
ref

Star configuration mandatory
 All VP6 lines equalized in length
 Common reference point on borad center :
dimension 2 x 1 cm = 1 mΩ
 Minimize variation of amplitude with
number of enabled channels
 Many tricky PCB layout details 150 mm
220 mm
21 jun 2002
C. de La Taille
calibration board design review
27
Schematics : power supplies
 Regulators on VP9 and VP7 from +11V. Also on VP5 ?
 Regulators on digital Vcc for delays and logic chips
21 jun 2002
C. de La Taille
calibration board design review
28
Schematics : TTCRx
 Derives 40 MHz clock and calib pulse (CMD)
 No EPROM used
21 jun 2002
C. de La Taille
calibration board design review
29
Schematics : SPAC
21 jun 2002
C. de La Taille
calibration board design review
30
Schematics : calogic
 1 chip for counter, 1 for DAC and 4 for Enables
21 jun 2002
C. de La Taille
calibration board design review
31
Schematics : delay chips
 Delay chip : AC output as it can be stuck to Vcc
 Trigger output on front panel
21 jun 2002
C. de La Taille
calibration board design review
32
Schematics
 CMD distribution : White follower
21 jun 2002
C. de La Taille
calibration board design review
33
Schematics : DAC
 Careful separate
layout of VP5 and
GND
 External reference
to VP9
 Internal low offset
opamp connected as
others
21 jun 2002
C. de La Taille
calibration board design review
34
Schematics : DAC distribution
 Low offset opamp for low impedance DAC fanout to 128 channels

21 jun 2002
External fuses ?
C. de La Taille
calibration board design review
35
Schematics : 1 pulser
 Low offset opamp for current source

HF switch with external NE856
 New enable schematic
21 jun 2002
C. de La Taille
calibration board design review
36
PCB layout
 Top layer : analog components
21 jun 2002
C. de La Taille
 Bottom layer : digital components
calibration board design review
37
PCB layout
Layer 2
21 jun 2002
Layer 3
C. de La Taille
calibration board design review
38
PCB layout
Layer 4
21 jun 2002
Layer 5
C. de La Taille
calibration board design review
39
PCB layout
Layer 6
21 jun 2002
Layer 7
C. de La Taille
calibration board design review
40
PCB layout
 PCB cross section : all inner copper layers 70 µm thickness
Output connector
Inductors
Analog part
Digital part
routing
Layer 1
Layer 2
GND
GND
routing
GND
GND
GND
VP5
Layer 3
Layer 4
outputs
GND
outputs
GND
enables
GND
enables
VP5
Layer 5
Layer 6
outputs
GND
outputs
GND
VP5
VP5
VP5
VP5
Layer 7
Layer 8
outputs
GND
GND
GND
GND
GND
routing
routing
21 jun 2002
C. de La Taille
calibration board design review
41
Summary
 The operation of the 8 channels board has been more difficult than foreseen

Layout very different from module 0
 Coupling of large digital signals
 Many modifications necessary
 Robustness had been overlooked
 Performance is good

Linearity < 0.1%
 Parasitic injected charge ten times better than on module 0
 The 128 channel is being finalized

Difficult layout of uniform VP6 : star configuration
 Many lines equalized in length
 Pending issues

21 jun 2002
Voltage regulator on VP5 (pin 4)
C. de La Taille
calibration board design review
42