Transcript Sector45-HWC-ABPO-2008-02-28

POWER CONVERTERS
D. Nisbet & Y. Thurel [AB/PO]
2008-02-28
Introduction
Sector 4-5 Overview
Converter related topics (/ type)
Issues
Specific Events
Conclusion
4/4/2016
2
Sector 45 Overview (1/2)
• Level of “powering”
60A Circuits:
94 / 94 …………
47 / 94 …………
80-120A circuits:
29 / 29* ……….
12 / 29* ……….
tested @ ±10A
tested @ ±55A nominal
tested @ ±10A
tested @ ±72-80A nominal
(0/20 of ±100 A nominal magnets see > ±10A)
600A circuits:
38 / 38* ..………. tested @ ±45A
38 / 38* ..………. tested @ ±200A
15 / 38* ..………. tested @ ±550A** nominal
* Available circuits, exception of:
Inner triplet & RCO, or blocked
** not 550A for all
4/4/2016
3
Sector 45 Overview (2/2)
• Level of “powering”
2kA Circuits (IPQ & IPD):
16 / 16* ……….. tested @ I
nominal
13kA-18V QD & QF circuits:
2 / 2 ……………… tested @ 10.9kA 12kA nominal
13kA-180V Main circuit:
1 / 1 ……………… tested @ 10.3kA 12kA nominal
* Available circuits, exception of:
Inner triplet
4/4/2016
4
60A: Summary
• Reliability results
– No power converter failure during Sector 45 HWC
– 1 / 94 converters changed for crowbar check
=> We are working on this converter in test hall.
No real crowbar issue up to now.
RCBV26.R4B2
• Some issue to be deeply studied
– Distortion in [0 .. ±3A] range
with nominal dI/dt (0.5A/s)
– Observed on 10% converters
=> We are working in test hall
on this issue
=> We will have to check
on magnets (next sector)
4/4/2016
Ramp from 0A to +5A
Voltage Distortion
2.6V
1A
5
LHC120A: RCBYHS4.L5B1 Quench
A magnet was detected being weaker than others
I meas. Abnormal
“bump” of current.
dI/dt = -0.625A/s just
before.
60mA
b
a
1V
I ref. 1st acceleration
0.25A / s² up to nominal
dI/dt 0.667 A/s.
V ref compensates
“bump” current by
“negative” step (a) , seing
current increasing. Then,
it corrects with a positive
step (b)
V meas follow V ref
4/4/2016
6
LHC120A: RCBYHS4.L5B1 Quench
Some specific tests were performed to analyze better the issue
40ADC
Converter executes a
±5V.. ± 15Vpp 1sec period@ 40A.
These “exotic” patterns were very
useful for spying magnet defaults.
DC current is 40A during this test.
Power converter is in voltage mode and
executes 10Vpp 1sec period
=> Abnormal « Bump » on current
We just catch (by chance?) this event since very close to limit!!!
=> A report of tests done. Analyze still on going (MPP).
=> Do we have to focus more on some tests (not missing this)?
4/4/2016
7
LHC600A-10V: Summary
• Very dedicated Specific Points
– Tests sequences can be optimized (reduced)
=> Save time & better define expected test result
– 0V distortion Issue
• Quench Protection System tripped on this distortion
=> Commissioning parameters had to be re-arranged
• High current reference error reported by FGC (//R)
=> Sequence validation issue
– Magnet inductance variation (RU)
• Key parameter for regulation (operation) and
discharge curve validation (commissioning)
=> Is LHC database format & data valid?
– Crowbar Validation Issue
• 600A Magnet is a wide family
(L, R//, Extraction system, 2 types of converters)
=> Gives a wide variety of discharge curve difficult
to be analysed
4/4/2016
8
LHC600A-10V: 0V Distortion Issue
• Impact on QPS
– QPS Trips, integrating the signal and
assimilating it at a quench signature. 0V/0A
<= A 1ms distortion voltage I-dependant on
power converter (4 quadrant stage)
– This issue was known from AB/PO but
influence on QPS was under-estimated
=> New filter from QPS based on distortion
data to be validated at beginning of next
sector.
100A
QPS Trip!
• Impact on Precision
– Not an issue on pure Magnet current, but
more difficult to deal with // protection
resistance
=> Sequence will be re-arranged to fit with
maximum limit to accept
10mA current deviation
4/4/2016
9
LHC600A-10V: Crowbar Analysis
± 600A
± 10V
± 600A
± 10V
•
•
•
•
± 600A
± 10V
Multiplicity of circuits
Bipolar circuits
Inductance not constant doesn’t help
Quench can be hidden in fast discharge.
=> A better tool based on running simulation MUST be developped
(several groups involved in characterisation of their device)
4/4/2016
10
LHC600A-10V: Magnet Inductance
• Impact on FGC
– Loop Stability
RST Algorithm uses the value of inductance for calculating its inner
current loop parameters:
L [H] RU.R4 Inductance vs I
30% max variation expected!!
5
=> These cases should be
Real Curve
4
known before starting
LHC Layout
commissioning to be better
3
Value (1.2H)
prepared in case of trouble.
2
• Impact on Curve analysis
1
0
I [A]
– Discharging curves are difficult
0
100
200
300
400
to analyze and requires simulators
=> LHC Database format for this Inductance parameter adequate?
4/4/2016
11
2kA & 13kA
• QF, QD & IPQD converters all
operated without any
significant issues
• Some remarks:
– One IPQ circuit operated with a
water fault in a sub-converter
(filter partially blocked) for all
the sector 45 duration
=> However N+1 redundant
topology worked and made this
failure transparent to testing
– RD3.R4 exhibited unexpected
behaviour when firing quench
heaters (current increases)
200A
3.5A
1000A
4532A
(Quench?)
RB, QF & QD: High Precision 22 bits ADC*
• « Idle tones » Issue on Sigma Delta ADC
–These spikes were as big as 20ppm peak-peak and
they were seen around 820A and 900A.
– All units tested (13kA) have the same behavior
although the DC levels vary slightly.
– Measurements in the laboratory reproduced
these idle tones at the same
DC input levels as seen in RB
=> PO is working on this
problem and it may require
some RB test time coordinated
with other HC activities
* These ADC are used to digitalize analog
measurement from DCCT to high
precision current loop on all 13kA.
820A
Up to 20ppm
900A
Up to 20ppm
4/4/2016
13
RB: [Failed => Unexpected successful] test
• “Re-catch” current
Earth Fault Detected
=> RB Trip in Fault @ 9kA
9kA
RB OFF => ON @ 5kA
- Re-Catch current
- Re-take control.
I.meas
5kA
I.ref
=> No Need to open
Switch to save time
=> QPS does not trip
PO In Action
1h30
Load Time constant 13 740 sec (4 hours).
Time 9kA => 0A > 7hours
4/4/2016
PO specialists mobilized
=> Problem diagnosed
as a weak component
Not a TRUE earth flt
=> Make the repair on
LIVE circuit
(5kA circulating)
14
All converters: Earth System Overview
• Earth System & Fuse issue
– Which fuse???
– Why so worrying?
I Magnet
R Cu Cable High
VOUT+
R Cu Cable Low
VOUT-
I earth
Converter
Earth Fault principle:
Active system, based on 10V
common mode on magnet
negative polarity*
- Detection even if converter
is in OFF state.
(up to Vcc)*
*Internal to the Voltage Source
Current Source
4/4/2016
10V common
mode
Fuse
Vearth leak
100mA
BLOWS ONLY IN
ABNORMAL
CASE
OVER VOLTAGE
PROTECTED
R3
Vcc
VS Cde
THIS FUSE 1A
“THE EARTH FUSE”
R1
A
- Earth fault path is energized
by a 100mA controlled current
source, not by current
deviated from the power
converter main output current
L Magnet
C
B
0VVCC
R2
100
Ohms
It is this additional
fuse 0.1A which
created false earth
fault on RB.
* RB exception
15
Powering Group of Circuits
• 138 circuits powered
• Use of machine optics functions
RB, QF, QD - Sector45 - PGC test
10000
1.40E-01
RPHE.UA47.RQD.A45:I_MEAS
1.20E-01
8000
Current (A)
7000
1.00E-01
6000
5000
8.00E-02
4000
6.00E-02
3000
2000
4000
4.00E-02
1000
3500
2.00E-02
3000
Current
(A)
15:36:00.0
tim e
15:21:36.0
15:07:12.0
14:52:48.0
14:38:24.0
-1000
14:24:00.0
0
Maximum Absolute Error (A)
9000
Successful squeeze
of D2 -> Q10
RPHE.UA47.RQF.A45:I_MEAS
RPTE.UA47.RB.A45:I_MEAS
Matching Section L5 - Squeeze
RPTE.UA47.RB.A45:
MAX_ABS_ERR
RPHE.UA47.RQD.A45:
MAX_ABS_ERR
RPHGA.RR53.RQ10.L5B1:I_MEAS
RPHE.UA47.RQF.A45:
MAX_ABS_ERR
RPHGA.RR53.RQ7.L5B1:I_MEAS
RPHGA.RR53.RQ10.L5B2:I_MEAS
RPHGA.RR53.RQ7.L5B2:I_MEAS
2500
0.00E+00
RPHGA.RR53.RQ8.L5B1:I_MEAS
2000
RPHGA.RR53.RQ8.L5B2:I_MEAS
RPHGA.RR53.RQ9.L5B1:I_MEAS
1500
RPHGA.RR53.RQ9.L5B2:I_MEAS
1000
RPHGB.RR53.RD2.L5:I_MEAS
500
RPHGB.RR53.RQ5.L5B2:I_MEAS
time
18:14:24.0
17:45:36.0
17:16:48.0
16:48:00.0
16:19:12.0
15:50:24.0
15:21:36.0
14:52:48.0
-500
14:24:00.0
0
13:55:12.0
Some issues to
investigate for RB,
QF, QD functions
RPHGB.RR53.RQ5.L5B1:I_MEAS
RPHGB.RR53.RQ6.L5B1:I_MEAS
RPHGB.RR53.RQ6.L5B2:I_MEAS
RPHH.RR53.RQ4.L5B1:I_MEAS
RPHH.RR53.RQ4.L5B2:I_MEAS
Inner Triplet Commissioning coming!!!(1/2)
• Complex system with interleaved circuits
– Was never tested on superconductive loads
– Crucial for machine operation
– Will certainly require some time before reaching required
performances (high precision on high complexity system)
=> Sector 56 Planning must integrate testing on Inner Triplet
4/4/2016
17
Inner Triplet Commissioning coming!!!(2/2)
System Complexity:
• Control uses:
- 3 FGCs
- specific Interlock &
control card
• Specific safety devices
never tested on
superconductive magnet:
- 7kA free-wheeling
thyristor
- 5kA free-wheeling diode
=> Time needed for correct
commissionning & Analysis
=> Time needed for High
Precision current
measurements
4/4/2016
18
Processing Data & Tools
• Sequences
– Sequences showed some issues on power converter and magnet
=> Automated tests are efficient and sensitive
– Some sequences can be shortened
=> Move measurement to another step
– Some “sign-off” point can be removed
• Post test analysis
– Not easy to get data
=> Data from MTF? Viewing of Data?
• Discharge tool is absolutely essential
for speeding acceptation process and validate better the curves
=> Up to now, acceptance on discharging curves (crowbar) are: not
very precise, nor quick and easy to do. (they are nevertheless
crucial: Example of R-discharge missing on High current)
4/4/2016
19
Conclusion
– For Next sector circuits should be commissioned quicker
• Sequences optimized for efficiency and for specificities
– A discharge tool is required for analyzing curves / this tool will be
used by PO (and MPP). An efficient & precise tool will require
model from all significant part involved in circuit.
– Planning of next sector is
• A lot of circuits to be commissioned as quick as possible
BUT ALSO
Upgrade
on
known
system
4/4/2016
• 600A 0V distortion management by QPS new algorithm (new
filter)
• 13kA Sigma Delta ADC issue will require some test time
• Some tests on 60A will have to be foreseen
Commissioning of new
• Inner Triplet commissioning !!!
complex system !!!
20