10/25/06 PEP-II LLRF Status PEP-II MAC Review October 25, 2006 D. Van Winkle, John Fox, Themistoklis Mastorides, Claudio Rivetta, Dmitry Teytelman, Jiajing Xu
10/25/06
PEP-II LLRF Status PEP-II MAC Review October 25, 2006
D. Van Winkle, John Fox, Themistoklis Mastorides, Claudio Rivetta, Dmitry Teytelman, Jiajing Xu
10/25/06
Outline
• RF Status – Abort Status– Key contributors to aborts– RF Tuning
• Klystron Linearizer Wrap-up
• Klystron Pre-Amp Investigations
• Plans and Conclusions
10/25/06
RF Status (Aborts and Tuning)
10/25/06
RF Status (Aborts)
• Understanding the root causes and patterns of RF station aborts is critical for operating efficiency– RF Aborts are tracked on a daily basis in a an attempt to find
root cause.– Sorting out the true RF aborts from the reported RF aborts is an
ongoing issue.– The goal of doing this RF abort tracking is to reduce the number
of RF aborts.– The pool of knowledge for this RF abort tracking is very limited
(1 person). – New staff is being identified/hired in the accelerator operations
department to share this RF diagnosis skill as well as increase operations efficiency in identifying and resolving RF related issues.
10/25/06
RF Status (Aborts)
Station Faults
her 566
ler 479
HR21 110
LR44 84
LR43 82
HR25 81
HR26 80
HR22 80
HR83 78
LR42 77
HR81 75
HR41 52
HR85 41
HR23 32
LR45 25
RF fault analysis over 483 days (04/23/2005 - 08/18/2006), 1942 faultsAverage of 4.0 aborts per day (1.9 (RF events))
Last event Count Stations Description
03/03/2006 425 ler(4),her(421) IP Vacuum event with/without radiation
01/27/2006 234 ler(215),her(19) didt with transverse
09/05/2005 89 multiple Transverse instability (TFB)
08/14/2006 81 multiple Unknown
08/14/2006 66 multiple Cavity 1 arc (RE reported)
07/16/2006 51 multiple Klystron arc/HVPS Arc
08/17/2006 48 ler(29),her(19) Longitudinal instability
07/30/2006 43 multiple Power dip (Region 8 or 4)
08/05/2006 41 multiple Slow or stuck tuners
08/16/2006 40 multiple Site Power Dip
Top 10 events account for 1118 faults (58%)
Color code: being debugged not actively pursued non-RF
Subsystem Faults
HER RF 629
LER RF 268
HER other 566
LER other 479
7 RF related aborts per day for Run 3
2.9 RF related aborts per day for Run 4
1.9 RF related aborts per day for Run 5
10/25/06
Average Daily Integrated Luminosity and Average Aborts/Day for 448 Day Period
0
100
200
300
400
500
600
700
800
900
Date
pb
-1
0
2
4
6
8
10
12
14
16
18
Ab
ort
s/D
ay (
avg
) o
r (A
mp
s^
2)
Significant Events
Integrated Lumi/Day
Total Avg Aborts/Day
Average Peak Current Product
RF Aborts Avg
RF Status (Aborts)
230kV line Down
BPMs Fall Off
RF Flange Region 4
IR2 Work Q2 chamber and bellows
IR2 Work Q4 chamber
10/25/06
RF Status (Aborts)Key Abort Contributors
Cavity Arcs We are limited in what we can do to address cavity arcs. Our strategy to date has been to
re-partition the gap voltage so that arcing cavities are running at lower gap voltagesStuck Tuners
This has been an ongoing issue for many years. The tuner motor drivers have been self shutting down on hot days sometimes generating an abort. Claudio Rivetta dug in and has figured out exactly what the problem is (Voltage drop to driver logic) and has designed a fix to resolve this issue permanently.
HVPS DipsWe have been seeing HVPS dips in two regions of PEP. At this point it is unclear what is
causing these dips. We have also seen an increase in site wide dips. The SLAC HVPS people are currently working to understand and resolve this issue.
Transverse Feedback Issues (see Ron Akre’s talk)We saw an increase in aborts do to the transverse feedback system this run. Ron Akre
should have some insight into what caused these.Longitudinal Issues
At high currents we saw loss of control in the low modes. This was likely due to the HER being run very close to the RF power limit at the end of the run. The higher order modes are in control but there has been thermal issues with the LER kicker cable plant and absorptive filter connections. These issues should be resolved for the next run.
10/25/06
RF Status (Aborts)
• Longitudinal Issues– Low-modes (cavity fundamental) are fastest growing instabilities in HER
and LER – HOM driven instabilities are 1/3 to 1/5 as fast - well controlled by
broadband feedback– Continued measurement of growth rates as currents increase,
estimation of adequate HOM damping margins in the future– In cavity modes -higher currents in the LER require LLRF system
improvements per Claudio's talk. (LGDW max damping rate may be exceeded otherwise)
– Thermal management • Cavity-style kickers in LER work well - still have had connector and
absorptive filter connector thermal problems.• HER- installation of "old" LER drift tube kickers with better cooling in
progress.• Monitoring of load powers, amplifier powers, etc. is very important in
operation of the systems.
10/25/06
RF Status (Tuning)RF Tuning
• The LER and HER RF stations are routinely “tuned” as currents are increased.
• A measure of how stable the RF system is running is how often the RF stations need to be tuned.
10/25/06
RF Status (Tuning)• What is meant by RF tuning?
– The LLRF system uses a time-domain excitation to inject noise and measure the closed-loop frequency response with the system in operation (and beam in the machine).
– A Model-based technique is used to fit the closed loop data. The model is open-looped and adjusted for best gain and phase margin ( best RF station stability).
• The model based corrections are applied to the station and measured again to ensure convergence.
• The Klystron operating point is not constant with current (resulting in small-signal gain and phase response variations). We configure the loops to operate best at the maximum running current with the idea that that is where the impedance needs to be controlled the most.
• As the loops cannot be opened with beam in the machine, this technique allows adjustment as currents increase, and brings insight into the dynamics of the station operating point.
10/25/06
RF Status (Tuning)• Example of poorly and well configured RF
feedback loops.
Poorly Configured Well Configured
10/25/06
RF Status (Tuning)
• Why is Tuning important?– If stations are mistuned, low mode longitudinal growth
rates can increase beyond our capability to control them with the low group delay woofer.
– Stations may oscillate causing beam jitter or loss.– Gap voltage regulation could be degraded.
• Other loops need tuning as well– Tuner loops, Drive set point loops, Gap voltage loops,
Tuner position control, etc. These loops have variations station to station because of Klystron and other implementation-specific details.
10/25/06
RF Status (Tuning)• During Run 5 and especially in the last 3-4 months of running, the
RF system required very little tuning.• Also during run 5, the LLRF hardware was much more reliable
(Much fewer module replacements than prior runs)
It is essential that the diligence of maintaining a known good set of spare modules and the replacement procedures be maintained
throughout the remaining running of PEP-II. This will be even more important as currents are pushed to new limits!
It is also essential the tuning and monitoring of the LLRF system continue. If the LLRF system is neglected, we will likely see an increase of beam aborts attributable to the LLRF system. The accelerator operations department is the right place for these
activities to be based.
10/25/06
Klystron Linearzier
10/25/06
Klystron Linearizer• In the last MAC review I gave an in depth talk on a klystron linearizer we
had been developing with the goal or reducing the low mode longitudinal growth rates.
• The last slide of that talk was “Next Steps”.
• MD2 with beam (higher klystron Saturation).– Modeling used to specify LER operating point to show saturated
effects (similar to HER).– Careful measurements of growth rates with and without linearizer.
Klystrons will be in heavily saturated state.• This MD will be the decision point for before final production linearizer
development.
10/25/06
Klystron Linearizer• The goal of the linearizer MD was to run
the LER with all stations linearized, and to carefully compare the station dynamics and instability growth rates with and without the linearizer.
• To do this we took grow/damp data at various beam currents with and without the linearizer at the same klystron operating points.
10/25/06
Klystron Linearizer• Final MD Results
10/25/06
Klystron Linearizer
Key Findings from Final MD
• The action of the linearizer as an small signal gain ( amplitude) linearizer was confirmed, though we found that one station could not be linearized. The MD was done with 1 parked station.
• The linearizer does help with the loop frequency response, and loop stability margins ( RF station stability).
• The linearizer does NOT seem to effect the low-mode growth rates ( when consistent operating points are chosen).
10/25/06
Klystron Linearizer• This discrepancy led to studies using the
nonlinear station simulation, and in hindsight we can now explain why the technique helps with station stability via the gain and phase margins, but does not improve growth rates. This understanding came from the analysis of the linearizer MD data.
• With this insight, we have decided not to invest resources in developing production linearizers.
10/25/06
Klystron Linearizer• We are still investigating the HVPS ripple reduction possibilities in
SPEAR. This is possible because spear does not push it’s LLRF system nearly as hard as PEP-II.
Linearized
Bypass
10/25/06
Klystron Linearizer• During the final MD, we had troubles configuring the
linearizer with LR4-2. It was unclear at the time as to what was causing the problem.
• In addition and around the same time Claudio Rivetta speculated that something in the LR42 transfer function was helping to increase the low mode growth rates.
• Some of our noise file measurements had confirmed this result.
It finally became apparent that something was wrong with the pre-amp in LR42. As we investigated further we found some interesting effects with most all of the Klystron pre-amps in both the LER and HER.
10/25/06
PEPII Klystron Preamps
10/25/06
Pre-Amplifiers• The pre-amplifier used in the PEP-II RF
system is under specified. The Klystron Pre-Amp was specified and selected based on CW operation.
• We operate the preamp in a mode which demands a large amount of dynamic range. Large CW signal with small modulation sidebands.
10/25/06
Pre-Amplifiers
• Typical Klystron output with beam
• Sidebands at +/- n*136 kHz.
• 50 to 60 dB down from carrier.
10/25/06
Pre-AmplifiersSignificant Amplitude and Phase Distortion exists in a +/1 MHz band around the carrier.
When carrier is removed (Blue Trace), gain looks flat as it should.
Small Signal
Large Signal
Large Signal
Small Signal
10/25/06
Pre-Amplifiers• All Amps in LER, HER and SPEAR have been measured
10/25/06
Pre-Amplifiers• Intermodulation Performance (MPE Amp)
– This characteristic is not well controlled (nor specified)
Pre-Amplifier
Intermodulation Distortion
10/25/06
Pre-Amplifiers• LR4-2, HR12-3 and HR12-6 are especially bad in
small signal response
10/25/06
Pre-Amplifiers• Pulse Response is undesirable
10/25/06
Pre-Amplifiers
• LR4-2 Distortion affects ability to implement comb rotation (see Claudio Rivetta’s talk)
LR42 is nearly unstable with 20 degrees of comb rotation. Simulations show this is due to non-linear pre-amplifier response.
10/25/06
Pre-Amplifiers• What to do?
– Run as is…– Modify/Replace existing amplifiers
10/25/06
Pre-Amplifiers
• Run as is?– Not a good scenario.– Modeling shows that the RF stations will be
difficult to configure at beam currents above 3100 mA.
10/25/06
Pre-Amplifiers• Modify/Replace existing amplifiers
– To evaluate required changes we have tested:
• Class A amplifier • Modified existing (Class AB) amplifier• Hybrid (replace RF part of existing amplifier)• Other class AB amplifier
– In the works:• Second hybrid option• Second class AB amplifier
10/25/06
Pre-Amplifiers• Conclusions/Plans
– We are actively searching for a replacement/upgrade path for the existing klystron pre-amplifiers.
– A detailed specification will need to be worked out which includes an intermodulation distortion spec.
– We plan to complete this work in the next three weeks and then generate a plan/proposal for pre-amplifier upgrades.
10/25/06
Conclusions
• The RF system is performing fairly well. Reducing the number of aborts below current rate of 2/day will require consistent operations support and careful attention to system configuration.– Transfer of tuning and fault analysis to operations is key to maintaining
the LLRF performance• The modeling effort in conjunction with what we have learned from
the Klystron linearizer effort has improved our ability to predict growth rates and understand the overall RF system operation– The pre-amps have been identified as a source of low mode growth rate
increase.• We are currently working on a modification plan for these preamps.• We hope to have amplifiers retrofitted before the down is over, but it may
end up being a swap and go program.• More testing is required before submitting plan.
– The RFP module has it’s own set of distortions which are just beginning to characterize. This may or may not end up being a significant contributor to low mode longitudinal growth rates.
10/25/06
End
10/25/06
HR21 Aborts by Type
1
42 2
10
5
23
1 1
11
5
1 13 2 1
5
13
1 13
16
2 1 1 1 1 1
0
5
10
15
20
25
30
HV
PS
Arc
Kly
str
on
Arc
Slo
w o
r S
tuck
Kly
str
on
or
HV
PS
Cavity 1
Vacu
um
Cavity 2
Vacu
um
Cavity 1
Arc
Cavity 2
Arc
Cavity 1
Arc
Cavity 1
Arc
(R
E
Cavity 2
Arc
(R
E
Kly
str
on
Vacuu
m
Tra
nsfo
rmer
Arc
Kly
str
on
Bod
y
Tra
nfo
rme
r/C
row
ba
r
Oscilla
tio
n a
t 15
-
Kly
str
on
Sid
eb
an
d
Kly
str
on
Bod
y
Hig
h R
efle
cte
d
Difficu
lt t
o
Cro
wba
r (S
puri
ous)
Ong
oin
g D
iag
nostic
Unkn
ow
n
Sm
all O
pe
ration
al
Pilo
t E
rro
r
Exte
rna
l A
bo
rt
Netw
ork
/IO
C issu
es
HV
PS
Rip
ple
Oscilla
tio
n o
n r
un
up
HR21 Aborts by Type
Extra Info (Aborts)• HR12-1 Aborts
10/25/06
Extra Info (Aborts)• Cavity 1 Arcs (RE reported)
# of Cavity Arcs
0
2
4
6
8
10
12
14
16
18
20
LR45 LR44 LR43 LR42 HR85 HR83 HR82 HR81 HR41 HR26 HR25 HR23 HR22 HR21
Station
Co
un
t
# of Cavity Arcs
10/25/06
Pre-Amplifiers• Amplifiers specified for only CW operation.• We devised a way to test the small signal in presence of large signal response and
found un-expected results.
10/25/06
Pre-Amplifiers• Intermodulation Performance of existing amps
10/25/06
Pre-AmplifiersEvaluations