M. Betz, Q. Du, B. Flugstad, K. Baptiste, M. Vinco Lawrence Berkeley Labs

LO

20 dBm

FSWP

sig gen

400 MHz

reference

500 MHz

+- 5 MHz

FSWP PN

analyzer

*2

821 MHz

+- 52 MHz

IF

RF

120 MHz

+- 20 MHz

AD9912 evaluation board

DDS to generate 100 MHz +- 1 MHz

800 MHz

DDS clock

Marki

T3-03

Homebrew I: AD9912 DDS + Mixer Homebrew II: Direct synthesis with a modern DAC

f1 = 499 645.0 kHz

* Annual freq. variation: < +- 10.0 kHz

* Chroma. meas. ΔP = 0.45%: < +- 2.0 kHz

* Insertion devices: < +- 0.1 kHz

needs to be adjustable in 1 mHz increments

at 1 Hz update rate (slow orbit feedback)

without phase discontinuity

Device Gbps Gain [dB] RMS Jitter [fs]

(499.6 MHz) (1 Hz .. 1 MHz)

FTLF8519P3BNL 2.125 1.0 103

FTLX8573D3BTL 10 -1.2 60

AFBR-709SMZ 10 -2.8 50

FTLF8528P3BNV 8.5 0.6 49

FTLX8574D3BCL 10 -0.7 47

Vialite + optical splitter -0.5 17

Baluns only (TC2-72T+) -2.0 6

* Distribution over phase stable coax

Andrew LDF4-50A Heliax (16 PPM/K)

* Distribution chassis (x12) based

on obsolete RF power amp. modules

* Operational since 1989, highly reliable

* More channels needed!

e.g., for beamline users

* Workarounds with RF splitters have

caused trouble in the past

* Only single frequency compatible

* Phase noise / signal quality bottleneck

Phase noise and spurs at the in- and output

of the current MO distribution chassis in the ALS

* extremely low cost & wide availability

* usable from 500 kHz to 1.5 GHz (-5 dB)

* surprisingly good PN / spur performance!

2 x Balun

TC2-72T+

PN analyzer

source

PNanalyzerinput

503 MHz+- 8 MHz

15 dB

Loopbackfiber

SFP module

evalu

atio

n boar

d

Phase noise test setup

for SFP transceivers

* good candidate for future MO dist.

* designed for satellite communication /

TV broadcast industry

* redundant power supplies

* hot-plugable modules (blind mate con.)

* telemetry and alarms over ethernet

* active optical x16 splitter available

* standalone receivers available

* excellent phase noise performance

see brown trace in PN plot

Low phase noise (PN) master oscillator (MO)

generation and distribution for ALS and ALS-U

Fiber link chassis

Standalone RX modules

x16 active optical splitter

equal to instrumentnoise floor

Measured on

R&S FSWP

PN analyzer

fc = 499.6 MHz

Linac

Buncher

Gun

Booster

Storage ring

50 MeV

2.0 GeV

Accumulator

SRRF

ARF

BRF

ALS-U

= f1

= f2

/4

*6

Timing sys.

E. gun

Buncher

Linac I

Linac II

Buncher

Storage Ring

f1 = 499.6 MHz+- 1.0 MHz

609

608*

f2 = 500.4 MHz+- 1.0 MHz

Accumulator

Booster

Storage Ring

Masteroscillator

Beamline users

Beamline users

Holzworth HS9001A

ALS-U

(2025)

MO chassis(custom made)

Fiber link chassis(up to 13 TX or RX modules)

f1 f2 f1 / 4

f1 distr. f2 distr. f1 / 4 distr.

x16 active

optical splitter

x16 active

optical splitter

x16 active

optical splitter

AD9164 / AD9174 DACs are capable of

directly synthesizing the MO signal with

excellent phase noise performance

less spurs due to higher resolution (16 bit)

and sampling rate (up to 12.6 GSps) of the

DAC compared to AD9912 (14 bit, 1 GSps)

+

synchronized

updates

FTW2

+

D

DFTW1

JE

SD

20

4B

FM

C c

on

ne

cto

r

128 MHz DSP clockVC707

Phase

accumulator

Frequency

tuning wordCordic

AD9174-FMC-EBZ DAC evaluation board 2x 500 MHz+- 2 MHz

JE

SD

20

4B

FM

C c

on

ne

cto

r

interpolation

to DAC sampling

rate

DAC1

DAC2

N

N

to freq.

distribution

/N fDAC = 5.12 GHz

fixed ref.

numerical

controlled

oscillator

FPGA

baseband Baseband freq. constraints

Spur measurement of AD9164 at fDAC = 5.12 GHz

ALS Master Oscillator

ALS-U MO distribution overview Current state in the ALSCurrent state in the ALS An experiment with SFP transceivers Commercial RF over fiber link (ViaLite)

A clean fixed 400 MHz frequency reference is

split in two channels. One is used as clock for a

DDS chip (AD9912), generating an adjustable

frequency of ~100 MHz. The other one is used

as Local Oscillator for a mixer, to up-convert the

variable ~100 MHz to ~500 MHz

* Extremely low phase noise

* Abandoned due to significant spurs at

sensitive frequencies

* They move with frequency set-point and hence are

hard to control

* Spurs inherent to the limited DAC resolution of the

AD9912 (14 bit)

VC707

AD9174-FMC-EBZ

AD9174 = 2 independent DACs for

ALS / ALS-U dual frequency

generation

DAC sampling rate fDAC chosen to

keep major spurs out of ALS

operational frequency range

Basically no spurs below 2 MHz

offset!

+1.8 dBm

for fc = 499.4 ... 499.8 MHz:

Spurious-free Dynamic Range = 117 dB

no major spurs within +- 2 MHz of fc

499.64, 499.65, 499.66, 499.67Carrier frequency fc [MHz]:

2 MHz

499.4 - 499.8 MHz

AD9164

5.12 GHz

16 bit, 12 GSps DAC

with internal NCO

evaluation board

FSWP

sig gen

FSWP PN

analyzer

Setup for PN / spur measurements

Current ALS MO

Homebrew I

Homebrew II

Most sensitive region for

beamline users ...

LLRF will apply noise to beam

PN measurements of 3 MO candidates RMS Jitter 1 Hz .. 1 MHz

t [1 ms / div]

Δf

[HZ

]

freq.

setpoint

change

R&S SMA100B

Holzworth HSX9002A

Holzworth HS9001A

RF output is blanked for 3 ms

internal PLL re-acquires lock

clean frequency step

current ALS MO

Setup for ALS / ALS-U dual frequency generation