3/7/05A. Semenov Batch-by-Batch Intensity Monitor 1 Two-Channel Batch by Batch Intensity Monitor for Main Injector BBI.

3/7/05 A. Semenov Batch-by-Batch Intensity Monitor

1

Two-Channel Batch by Batch Intensity Monitor for Main Injector

BBI


2

Two-Channel BBI Device

BNC Input # 1 From Wall Current

BNC Input # 2 From Wall Current

Beam/No Beam Output signal #1

Analog output #1

Analog output #2 Beam/No Beam

Output signal #2

TCLK

MDAT

EXT RF CLOCK

EXT SYNC

#AA Marker


3

BBI device location – MI60

BBI

WCM fanout

MI60004


4

NuMI batches

pbar batch

What are the main Goals?

One batch to pbar, the rest

to NuMI

Data at 720Hz of beam intensity for 7 different gates (Fast Time Plot)“beam/no beam” logic signal if beam is present after extractionAnalog signal of beam intensity for selected gate


5

Main Injector Harmonics 0 82 122 208 294 380 466 588 0

Placement of gates 1 Pbar batch Bucket 0(Synched with RF) 2 1500 ns before NuMI Batch 1 3 NuMI batch 1 Bucket 122 4 2 208 5 3 294 6 4 380 7 5 466 8 Possible 6th NuMI batch

Gate1Pbar batch

Gate2Pbar Kicker

firing

Gate3NUMIbatch1

Gate4NUMIbatch2

Gate5NUMIbatch3

Gate6NUMIbatch4

Gate7NUMIbatch5

Gate8?NUMIbatch6

Kicker


6

Project Critical Point #1Short ~5ns pulses from WCM

Figure 1 Bunch structure and a 5 bucket gap between batches, measured with the WCM

94ns


7

Project Critical Point #2 Varying baseline

Figure 2 Varying baseline of the RWM during multi batch mode. Figure 3 Expanded measurement of the multi batch structure during 3 turns with the RWM. The varying baseline is evident.

Over full cycle (1.3 secs) Turn by turn


8

Project Critical Point #3Gates Timing

Figure 4 RWСM measurement of the multi batch structure with the pbar batch being isolated by larger gaps

(~43 buckets)

Integration gates 500-1600nsMinimum timebetween NuMI batch windows3 buckets ~55nsTime between pbar and NuMI batches:43 buckets ~800nsTwo main integration gates:Batch (83 x 18.8ns) 1560ns NuMI Kicker rise time 1300 - 1500ns Maximum rise/fall time of gate 10ns


9

BBI Hardware

FPGACYCLONE

53 MHz, TCLK, RevMarker..

VGA212MHzADC

Cable from Tunnel

400MHzDAC

12

Ethernet

INPUTs: resistive wall current monitor

OUTPUT:

Pulse stretcher

DSPSHARC

10Mbit ETHERNET

RS232,RS485Channels: 1,2


10

Direct Sampling at 200MHzThis signal istoo fast for a

measurement by200MHz Digitizer

~4ns

(Imax-Imin )/ I ~42%

1000

400

Samples at 200MHz

phase between beam signal and digitizer clock


11

Analog Pulse Stretcher

OPA OPAPA

+

-

Td = 5ns cable

Input

Output

Spreads signal +/-5ns in time so it will not be missed by

ADCAt 200MHz

Reduces ADC

Dynamic Range

requirement

~4ns

5ns

PreAmp

Operational

Amplifier


12

Analog Pulse Stretcher Input Pulse Width = 2ns

Tdelay = Tsample

No phase errors


13

Analog Pulse StretcherTdelay > Tsample (20%)

(Imax-Imin )/ I ~1.6%

In Pulse 5ns

Output

phase errors!!!


14

FPGA Baseline CorrectionHow define baseline?

• it is natural for our case to define "baseline" as a line “not containing peaks"

• It means that for base line correction algorithm we should use at list 2 parameters – threshold (to detect a peak amplitude above baseline) and window ( to separate a slow fluctuation of base line signal and peak signal in time domain region)


15

Two Step Baseline Correction Algorithm

Step 1 : fix the “floor” of signal by averaging the lowest signal point from a sliding set (64) of programmable windows (100 buckets) and use to define the “minimum signal base line” used to define input signal threshold Step 2 : calculate natural baseline by averaging all signals points below the threshold

Step 1: 100 bucket window

Step1: Minimum points

Step1: minimum signal base line

Step2:Threshold

Step2: final baseline


16


$23 step1: input signal – minimum signal baseline Threshold


17


$23 Step2: output signal


18


Hardware Implementation

ComparatorTemporary Min Value Register

SlidingAverage Filter

Write new value at the

end of window

Window

INPUT:

Data from

RF 1 bucket Integrator

~18.8ns

Rank 64

-

+ Minimumsignal points

Current minimumsignal point inside time window

minimum signal baseline

Step1: averaging a minimum signal points inside programmable windowOUTPUT:

Input Data with restored MinBaseLine

Input Pulses


19


Hardware Implementation (continued)

WindowComparator

Baseline Register

Moving Average FilterFinal

Baseline

Threshold

-

+

OUTPUT:

Input Data with

Restored Base line

Rank 64

INPUT:Input Data

with restored MinBaseLine

Input Pulses

Threshold

Baseline

Step2: averaging all signals points below programmable threshold


20

FPGA Design

BLC Gate Integrator #0

Start Stop Mean#

SHARCSPORT32bit,

10Mbit

Gate Integrator #6

Start Stop Mean#

Threshold>= Beam/No Beam

TCLK

RF,53MHz

TCLK

FromADC

Rev Marker

PLL Multiplier

ADCCLOCK

Timestamp counter

Intensity (gate #0)

DACIntensity #0

FIFO

Delay

RF INT

To SHARCDSP


21

FPGA FUNCTIONS

What we do in the “CYCLONE”:1) Base Line Corrections

2) Beam Intensity calculation - Multi-Turn Averaging

3) Generate the correct position of the batch gates. It is possible to program the position of the gates and the beam threshold for NuMI gate through ACNET

4) A FPGA PLL multiplier (x4) is used to synchronize the digitizer with the external RF or internal (53MHz) clock


22

FPGA FUTURES

What we do in the “CYCLONE” (continued):5) Ability to trigger on TCLK events

6) Multi-Turn “Scope Trace” buffer to analyze input and post-processing signals (4K samples)

7) 10 Mbit,32bit serial interface to SHARC DSP


23

SHARC Digital Signal Processor

SHARC21065L

EthernetRTL8019AS

2MBFLASHRAM

AM29LV017D

2MBSDRAM

48LC1M16A1

FPGA IO Controller

MAX7128

Front EndControl:

VGA, ADC, DAC

FPGASPORT interface

RS232,RS485

ACNET,ETHERNET


24

SHARC’s role on BBI board

• Interface board to ACNET (OAC) via TCP/IP– Includes remote firmware/software update

• Read/write BBI CYCLONE “control” registers, so that BBI processing can be controlled and monitored via network

• additional “offline” data processing and filtering ( fast time plots, base line corrections, FE dynamic range control by VGA,…)


25

SummarySignal Processing Steps:

1) Analog pulse stretcher 2) Digitizing by 12 bit ADC at 53x4=212MHz3) FPGA base line correction4) FPGA gate intensity calculation5) FPGA Multiturn Averaging 6) FPGA Beam/NoBeam checking7) “Scope Trace Buffers” for all gates8) SHARC signal processing, FTP calculation9) SHARC TCP/IP data transfer to OAC

InsideFPGA


26

SUM of Gates Intensity Fast Time Plot

FPGA Constructed Total Intensity

Full NuMI Cycle

Gates Intensity


27

Gates Intensity Fast Time Plot

$23 1.9 e13

NuMI Cycle

0 1 2 3 4 5 6

Gates (Batches)


28

Batch by Batch Intensity $23

stacking batch

Numi veto buckets(no beam just before NuMI extraction

NuMI Batches

slip stack batch


29

$21 Cycle

stacking batch

Numi veto buckets(no beam just before NuMI extraction

slip stack batch


30

BBI LabView Interface


31

BBI ACNET General page


32

BBI ACNET page for channel #1


33

BBI ACNET page for channel #2

3/7/05A. Semenov Batch-by-Batch Intensity Monitor 1 Two-Channel Batch by Batch Intensity Monitor for Main Injector BBI.

Documents