DD

WBS 1.2.8:AFE II Status and Plans

Alan Brossfor the

DZero Central Fiber Tracker GroupDZero Run IIb Detector Upgrade Director’s

ReviewJuly 15, 2004

Fermilab

DD

The D0 Central Fiber Tracker

A little Background

DD CFT Physical Geometry

8 cylinders Outer 6

2.52 m Inner 2

1.66 m 16 Fiber doublet

layers 8 Axial 8 Stereo 1 each on each

cylinder r - .2 to .51 m Axial readout to

north Stereo readout to

south 77,000 channels 100,000 with

Central and Forward Preshower Detectors

DD Waveguide Bundle Routing

Routing the optical signal to the VLPCs which sit under the CC imposes some constraints.

5 sectors Each with

different fiber run

Waveguide lengths

7.7m min 11.5 m max = 8.1 m

Light Yield is a function of

Current Worst Case LY 8 pe

3-FRONT

4-FRONT

D3-G

4

B 1

5A 5B

E 3

R99.000

A2-G

5

A 2

G 5

C 3

E 4

C3-E

4

A 1

H 5

A1-H

5

B 2

F 4

123

ONE HALF CONNECTOR IN SECTOR 3 ONLY WITH FULL SHAREDCONNECTORS AT SECTORS 1 & 2, AND SECTORS 4 & 5.

4 5

B2-F

4

274

296

232

135

177

SECTORS 1 THRU 5 (5 SPLIT)CABLE ROUTES3/4 DIA. BUNDLES HELD AT R58 AND

CONTINUING ON TO THE TOP OF THE NOTCHES575eith3card.dwg

SCALE: 1/8(LOOKING AT SOUTH FACE OF CC)

B 3

H 6

B3-H

6

D4-F

5

D 4

F 5

B 5

B5-H

11

F 9

G 1 0

H 1 1

F9-G

10

D 7

E 8

D7-E

8

C 6

H 1 0

B1-E

3

C 2

F 3

H 4

D 2

C2-H

4

D2-F

3

D 3

G 4

C6-H

10

G 9

A 4

A4-G

9

E 7

G 3

H 3

G3-H

3

E2-F

2

E 2

F 2

F 8

E7-F

8

D 6

H 9

D6-H

9

C 5

G 8

C5-G

8

E 6

F 7

E6-F

7

B 4

H 8

B4-H

8

G 7

D 5

D5-G

7

A 3

F 6

A3-F

6

C 4

H 7

C4-H

7

E 5

G 6

E5-G

6

C1-D

1

C 1

D 1

E 1

F 1

G 1H 1

H 2

G 2

E1-G

1

F1-H

2

G1-H

1

C.P.S.

C.P.S.

H 1 2

H 1 3

H 1 4

H 1 5

H 1 6

H 1 7

G 1 1

G 1 2

G 1 3

G 1 4

G 1 5

F 1 0

F 1 1

F 1 2

F 1 3

F 1 4

E 9

E 1 0

E 1 1

E 1 2

D 8

D 9

D 1 0

D 1 1

C 7

C 8

C 9

B 6

B 7

B 8

A 5

A 6

H 1 8

H 1 9

H 2 0

H 2 1

H 2 2

G 1 6

G 1 7

G 1 8

G 1 9

G 2 0

G 2 1G 2 2

F 1 5

F 1 6

F 1 7

F 1 8

F 1 9

F 2 0

E 1 3

E 1 4

E 1 5

E 1 6

E 1 7 E 1 8

D 1 2

D 1 3

D 1 4

D 1 5D 1 6

C 1 0

C 1 1

C 1 2

C 1 3

C 1 3

B 9

B 1 0

B 1 1 B 1 2

A 7

A 8

A 9

A 1 0

B 1 3

C 1 4

C 1 5

D 1 7

D 1 8

E 1 9

E 2 0

F 2 1

F 2 2

F 2 3

G 2 3

G 2 4

G 2 5

H 2 3

H 2 4 H 2 5

H 2 6

H 2 7

H 2 8

B6-C

8-F

10

-H1

2E9

-G1

1

DD Visible Light Photon Counter

8 channel VLPC is used to readout fibers

QE 80% Nominal Gain =

40k But wide

distribution

DD

Analog Front-End (AFE) Readout Board

And Why We Want to Replace It

DD Analog Front-End (AFE) Board

Approximately 200 AFE boards are needed to readout CFT (&CPS/FPS)

512 channel (2/cassette) Analog output via SVX IIe Discriminator output

every 396 ns for L1 trigger

MultiChip module (MCM) SIFT chip

– Analog buffer to SVX– Discriminator output

SVX IIe

DD AFE I vs. AFE II

There are a number of problems with current AFE board that impact data taking and can significantly compromise data quality for high luminosity running

The problems with the AFE board are associated with the MCM

SVX IIe SIFT – SVX IIe interaction on MCM Functionality limitation (by design)

Threshold setting– Zero suppression (1 setting/64 ch – SVX) (Analog information)– Discriminator (14 or 18 channels per setting)

To fix this our proposal is to build new AFE boards - AFE II

No MCM New pipeline/Trigger Chip – TriP (and TriPt)

Pipeline Discriminator

Commercial FADCs for analog information FPGAs for processing

Online thresholds – one/channel (analog)– 1/16 discriminators

DD AFE I Problems

SVX Saturation At high Luminosity (high occupancy ( 30%) analog

signal for a given crossing will be reduced because of “history” in SVX front-end.

– One reset/superbunch (12 xings) Pedestal width and stability

Tick to Tick variations and problems within a tick At high luminosity discriminator firing effects

pedestal – requiring additional increase in thresholds. Quality of analog information is significantly degraded

At the Highest Luminosity expected for Run II, the degradation in the quality of the AFE analog information will significantly impact offline reconstruction and the Preshower capabilities.

Threshold increase as much as 4-5 pe!

DD AFE I Problems

Discriminator At high occupancy discriminator firing will also

impact the discriminators Threshold shift as L increases.

Compounding these problems we have VLPC and fiber issues at high luminosity (instantaneous and integrated)

VLPC Drop in gain (20%) Drop in QE (10%)

Fiber Radiation Damage induced light yield loss (10-20%)

DD SVX Pedestals vs. Accelerator Tick

Seen in all channels, effect is proportional to gain:

• Additional threshold cuts needed to suppress tick-dependent noise.

accelerator tick

Mea

n pe

dest

al 20 counts

20 counts 1.5-2pe

DD SVX Saturation

The SVX pipeline is only reset once per superbunch (12 xings)

At 30% occupancy 4 Hits integrated in pipeline “Typical” integrated charge on front end 40 pe 40% drop in Signal for “triggered” hits

Loss of signal in main pulse vs. # pe's in early pulse

-0.9-0.8-0.7-0.6-0.5-0.4-0.3-0.2-0.1

00.1

0 20 40 60 80 100 120

N' (corrected #pe 's)

rela

tive

sign

al

Relative Signal Loss

Number pe

DD Pedestal Shift- Single Channel Example

ADC spectra (from SVX) are affected by firing of SIFT discriminators

This example shows

A ped shift to the left of 1.2 pe

ADC resolution degradation

Individual pe almost gone

< 1% occup.

32% occup.

DD Pedestal Shift Global

Ped shift in 4 of the 8 MCMs on an AFE

30% occupancy Inner layer of

CFT at L 1032

Plots represent mean ped change v. channel between 0 and 32% occupancy

2-4 pe Large compared

to 8 pe Structure

MCM layout 32% occup.

DD

Physics Impact

But First – What is the light yield“If we had 100 pe/hit, I would not be

talking”

DD Light Yield

These data show LY v. for =0

Average for each supersector

“Current worst case LY” 8 pe

This is a lower limit due to SVX sat. & ADC dynamic range

Expect Axial LY Stereo LY

Axial Gain > Stereo Gain

SVX Sat. Effects smaller in Stereo

DD

The Light Yield/Signal will get worse

DD VLPC Luminosity Dependent Effects

As luminosity goes up VLPC QE and gain drop

0, 10, 20, 30, 40 % occupancy(expect close to 40% on inner layers at highest RunIIb L)

10% drop 20% drop

QE Gain

DD Fiber Radiation Damage

Dose CFT L1 2 fb-1: 10 krad 30 fb-1: 150

krad 90 fb-1: 450

krad

Represents attenuation length change

0 fb-1 : 5.0 m 2 fb-1 : 3.8 m 30 fb-1 : 2.7 m 90 fb-1 : 2.1 m

Michigan Radiation Test Summary

0

0.2

0.4

0.6

0.8

1

1.2

0 200 400 600 800 1000 1200

Dose Received (kRad)

Rel

ativ

e Li

ght I

nten

sity Cyano HBT

3HF Run II3HF Run I

DD

So, What do we Conclude about Physics Impact

Complicated, but…

DD CFT Data Quality at High Luminosity

Degradation/loss of analog information Tracker

Offline tracking with analog information significantly degraded Efficient tracking may only be possible with discriminators

No chance of using pulse height to help tracking Point set resolution Cluster “splitting”

– MC indicates that the analog information should help– Have not been able to demonstrate this with data– Even at low luminosity, AFE board analog information affects this

capability Central and Forward Preshower Detectors

Potential degradation of analog information could be crippling These may become “Digital” devices

Even if we accept all of the above – IMPACT STILL LARGE On Physics to Tape (NO RECOVERY POSSIBLE) On Manpower needed to make the most of the

tracking/preshower data

DD One Physics Example Silicon Track Trigger Impact on Higgs

Red curves High/Low LY limit

03 Black

High/Low limit 06 Discriminator Hit

efficiency drop due to threshold shift (025% occ.)

94.5% 91% 8 out of 8

63% 47% Zh case

h bb 20% Loss in STT

“tagged” Higgs evts

AFEII recovers this No Threshold shift

Sing

le-H

it ef

ficie

ncy

Discriminator Threshold (pe)

DD

AFE II Status

The Sequel

DD AFE II

Progress on AFE II is covered in D0 Notes 3898, 4009, 4076, 4233, 4497, 4500

3898 “Design of the new MCM” 4009 “MCM II and the Trip Chip” 4076 “Characterization of the TriP Chip Running at 132

ns Using a modified AFE board” 4233 “System tests of the new electronics for the CFT

and Preshower Detectors” 4497 Z-coordinate Measurement in the CFT and the

Track Search” 4500 “Problems with the AFE operation at high rates”

Also AFE II proposal and AFE Design Specification documents

This information is in your handouts

DD AFE II

A Tremendous Effort has already gone into AFE II

PPD EE DepartmentJohn Anderson

Jim HoffAbder Mekkaoui

Paul Rubinov

D0 Alan Bross

Juan EstradaCarlos GarciaPeter Hasiakos

Bruce HoeneisenMarvin Johnson

Mike Utes

DD AFE II AFE II

Full New set of boards No SIFT No SVX – Much simpler architecture Will use instead

Trigger Pipeline Chip: TriP (or Tript - more in a bit)– Discriminators and analog pipeline– TriP chip submission was very successful – meets spec.

Commercial Flash ADCs + FPGA for analog information Integrates completely with existing system

Will Have No Front-end saturation problem

– Reset once per crossing– Also allows for data taking in abort gap (cannot do presently with

AFE I)• VERY useful for Calibration studies

Improved Ped dispersion and stability– Lower and tighter threshold setting capability

• Channel by Channel - analog Improved reliability Improved readout flexibility

– Decreased deadtime– Multi-buffering possible

• CFT Readout may dominate L1 readout rate without multi-buffering Added Functionality with new submission of TriP Chip - Tript

– z information from timing ( 2 ns rms)

DD TriP

The existing TriP ASIC (7000 die on hand) Lots of testing done over last 2 years

DD AFE II –TriP Performance

Measurements on TriP very promising

Noise at 1/5 pe rms Threshold setting at 1.5

pe reliably Very high quality Analog

Data Plot at right was obtained

after 1 day of work on a TriP modified AFE board

To get this good, a plot with the AFE took about 1 month of dedicated work by the same people!

BLACK – DATA RED – FIT

All Discriminators Firing!

DD AFE II – TriP Discriminator Performance

Upper plot Red – all data Blue same data

with threshold and discriminator fires

Black – fit Lower plot

Ratio of analog data with and without discriminator fires (threshold set)

30% occupancy

DD TriP-t: the new idea

About 1% additional components:(time to amplitude converter)

7 bits t-info offline: 120ns full scale=> 2ns time => ~30cm resolution in Z

DD AFE II - TriPt

With a rather simple modification to the TriP design – time stamp for hits can be obtained

Current TriP Electronics resolution was

determined to be 400 ps For 8 pe signal expect

about 2 ns sigma Z measurement – 30 cm

Has impact on reco time and cluster splitting

In Z +15 min bias MC 40-60% reduction in reco

time D0 Note 4497

Improvement in clustering algorithm utilizing z info also a possibility 400 ps intrinsic resolution

DD TriP-t prep work

Bench testing

New package

DD TriP-t current status

The chip designers are chip designing. Ready for submission 23 Aug ’04 The new chip is the “critical path” item for

the project Some prep work going on now

We will need new packaging We had some concerns about features

used on the new chip, but not used on the old (current!) TriP chip, so we are retesting some things.

DD AFE II - Cost and Schedule

Full Project Plan Now Available

DD AFEII Prototype - current status

Schematic design is 100% complete. Internal review was held in early May. Layout of the board is about 90% complete. All parts are here. Req’s for boards are in the system. RFQs for stuffing are on the street. Firmware about 75% complete.

DD AFE II – Cost and Schedule

Cost M&S: Production cost M&S are based on quotes for parts and labor –

contingency estimate is grounds up: Manpower

Estimates from AFE experience Schedule critical path is the Tript chip Completion for the 05 Shutdown is not likely Planning for adiabatic Installation

AFE I and AFEII mixed running Will need to test compatibility once AFE II prototypes are ready

(9/04) Installation will have to be done at the crate level (16 boards)

because of difference in power requirements for AFE I and AFE II Credibility of installation plan is dependent on prototype tests

How quickly they come up Platform test and integration with trigger Software readiness Physical Installation can occur quickly (8 hour)

DD AFE II Milestones

AFE II Prototype under test 9/04 Tript MOSIS submission 9/04 D0 Internal Review 1/05 Director’s Review 2/05 Tript production submission 3/05 AFE II pre-Production 7/05 AFE II production 9/05 AFE II production complete 12/05 First Board Ready for installation 2/06

DD Conclusions – Improvements with AFE II

AFE II will Improve noise floor and pedestal stability which will allow for

consistent and reliable threshold setting with no discriminator feed-through or discriminator threshold shift

Will increase physics trigger efficiency Potential for better hit efficiency and point resolution Maintain capability of Preshower Detectors

No Front-End Saturation problem (reset once per crossing) Added functionality of the Tript (z information)

Large reduction in reco time Possible improvement in cluster finding

– Cluster splitting leading to additional improvement in track quality Readout architecture much more flexible

Less deadtime Multiple buffers can be added to greatly increase L1 capability

Board construction is simpler, more robust No MultiChip modules (MCM) Mostly commercial parts/standard mounting techniques Repairs/rework MUCH easier than in AFEI

AFEII will improve our capability to maintain high-quality and stable operation as the Tevatron Luminosity increases in RunII. This will allow D0 to maintain excellent tracking, trigger, and Preshower performance