New 30 kA power system at Fermilab and its use for measuring the effects of ripple current on the performance of superconducting high field magnets

Abstract

A new 30 kA, 30 V dc Power System was designed, built, and commissioned at Fermilab for testing Superconducting High Field Magnets. This system has been successfully supporting operations at the Fermilab Magnet Test Facility since April 2002. It is based on six commercial 150 kW Power Energy Industries power supply modules and the following in-house modules: six 720 Hz filters, two 15 kA/1 kV dc solid-state dump switch, and a 3 MJ/30 kA/1 kV dc dump resistor. Additional in-house electronic components were designed and built to provide precise current regulation and distribution of current and current rate of change. An industrial-type Programmable Logic Controller system was used to provide equipment interlocks and monitoring. This paper summarizes studies on the influence of characteristics of this new power system-such as ripple current-on the performance of High Field Superconducting magnets.

Full text

Page 1

1LL06, Page

1
New 30 kA Power System at Fermilab and its
Use for Measuring the Effects of Ripple Current
on the Performance of Superconducting High
Field Magnets
R. Carcagno, S. Feher, J. Garvey, W. Jaskierny, M. Lamm, A. Makulski, D.F. Orris, H. Pfeffer, M.
Tartaglia, J. Tompkins, D. Wolff

Abstract—A new 30 kA, 30 V dc Power System was designed,
built, and commissioned at Fermilab for testing Superconducting
High Field Magnets. This system has been successfully
supporting operations at the Fermilab Magnet Test Facility since
April 2002. It is based on six commercial 150 kW Power Energy
Industries power supply modules and the following in-house
modules: six 720 Hz filters, two 15 kA/1kV dc solid-state dump
switch, and a 3 MJ/30 kA/1 kV dc dump resistor. Additional in-
house electronic components were designed and built to provide
precise current regulation and distribution of current and
current rate of change. An industrial-type Programmable Logic
Controller system was used to provide equipment interlocks and
monitoring. This paper summarizes studies on the influence of
characteristics of this new power system—such as ripple
current—on the performance of High Field Superconducting
magnets.

Index Terms—Superconducting Magnets, High Field, Power
Systems, Ripple

I. INTRODUCTION
HE High Field Magnet Program underway at Fermilab
required a new power system to extend the dc current
capability at Fermilab’s Magnet Test Facility (MTF) to 30 kA,
30 V dc. Other requirements for this system included: < 100
ppm of full scale absolute current readout accuracy, < 10 ppm
of full scale current regulation and stability, and a 3 MJ/30 kA
energy extraction system. In addition, the system must
withstand a maximum voltage of 1000 V dc during a magnet
quench. A system capable of meeting these requirements was
designed and built based on existing power components at
Fermilab. This system was commissioned in April 2002 and
has been successfully supporting the following testing
programs at MTF: (1) The Nb3Sn High Field Magnet (HFM)
program; (2) production testing of LHC Interaction Region
Quadrupoles; and (3) production testing of LHC High-

Manuscript received October 5, 2004. This work was supported by the U.S.
Department of Energy under contract No. DE-AC02-76CH03000.
All authors are with the Fermi National Accelerator Laboratory, Batavia,
IL 60510 USA (R. Carcagno phone: 630-840-3915; fax: 630-840-6724; e-
mail: ruben@fnal.gov).
Temperature Superconducting (HTS) Power Leads. Loads
connected to the 30 kA system include resistive loads (HTS
Power Leads), and superconducting magnets with an
inductance range of 35 µH to 30 mH. Current ramp rates were
as fast as 500 A/s. The maximum HFM magnet quench to date
was at 24 kA.
The new 30 kA Power System includes several features to
provide precise current monitoring and regulation with small
ripple. These features will be described in more detail below.
Ripple voltage and ripple current characterization results will
be presented for different loads. Ripple current at the level
observed in this system does not appear to have an adverse
effect on the performance of Nb3Sn High Field Magnets.

II. 30 KA POWER SYSTEM OVERVIEW
Fig. 1 shows a block diagram of the 30 kA Power System.
The 30 kA bus is shown as solid lines, and electrical signals
are shown as dashed lines.
A. Power Supplies
The system includes six commercial 150 kW Power Energy
Industries (PEI) power supply modules in a master/slave
configuration. The Master Power Supply (MPS) provides the
Silicon Controlled Rectifiers (SCR) firing signals for all
modules. Each power supply has four firing modules with
three SCRs each (one per phase), for a total of 12 SCRs.
Current regulation is accomplished by an external precise
current regulator cascaded to the MPS internal regulator set to
“Voltage” mode. Each PEI supply is capable of delivering
5,000 Amps dc at 30 Volts. There is also the option of tapping
the supply at higher voltages with the corresponding decrease
in maximum current. The system is highly modular, and it can
operate with just the master supply or with the master supply
plus any combination of slave supplies.
B. 720 Hz Filters
To reduce the 720 Hz ripple of the twelve-pulse, unfiltered
PEI supply, a Praeg style filter [1] was designed, built, and
added to each supply. Each filter includes a serial 250 µH
T
FERMILAB-CONF-04-245-TD

Page 2

1LL06, Page

2
choke cooled by Low Conductivity Water (LCW), followed
by two parallel capacitor banks: a 5.5 mF bank and a 27 mF
bank in series with a 200 mΩ resistor.
C. Dump Switch
Stored energy extraction from a magnet is accomplished by
quickly switching a load resistor in series with the bus. Fast
switching is accomplished with two 15 kA 1000 V dc solid
state dump switches. These switches were originally made for
the SSC project [2], and include six SCRs mounted in
watercooled heatsinks. The SCRs continuously carry their
share of the rated current, and they are commutated off from
stored energy in capacitors. Once the SCRs turn off, the bus
current flows through the dump resistor connected in parallel
with the dump switches. The original current capacity of the
dump switches was 10 kA, and they were upgraded to 15 kA
by replacing the SCRs with higher rated components. The
switch opening time is approximately 25 µs after detection of
a fault or a trip command.
D. Dump Resistor
The dump resistor dissipates the stored energy from the
magnet after a quench or trip. The resistor was built from air-
cooled stainless steel elements, and has adjustable taps at 10,
15, 30, 60, 90, and 120 mΩ. It is rated for a maximum current
of 30 kA, a maximum energy dump of 3 MJ, and a maximum
voltage across the resistor of 1000 V dc.
E. Zero-Flux Current Transformer
The current from each PEI module is measured by a bipolar
zero-flux current transformer. The transformer is HITEC
system TOPACC type 60-8 bipolar, rated for a current of
6,000 A. DC accuracy error is less than 30 ppm, with a
temperature coefficient less than 1.5 ppm/K and drift less than
10 ppm/year.
F. Summing Amplifier
The output from each zero-flux current transformer is sent
to a precision summing amplifier. The summed signal is sent
to a precision current regulator and to a VME current
transmitter module for distribution and monitoring purposes.
The summing amplifier was built with precision resistors with
an absolute tolerance of ± 0.001%, and an absolute
temperature coefficient of ± 0.8 ppm/Deg C. Stability tests
made on this unit showed less than 2.7ppm drift in 90 hours.

Fig. 1. 30 kA Power System Block Diagram
G. VME Current Transmitter/Receiver modules
An in-house designed and built VME current transmitter
and receiver module is used to distribute via fiber optic cable
the current (I) and current rate of change (Idot) data to up to 8
different locations at the Fermilab’s Magnet Test Facility. The
maximum length of the fiber optic cable is 250m. The
measured accuracy of the current readout is ± 10 ppm
(integrated over one Power Line Cycle), and the resolution is
± 2 ppm. The current receiver module accepts the fiber optic
signal from the transmitter module and provides six buffered
analog I outputs and three analog Idot outputs. The measured
analog current value accuracy is ± 25 ppm, with a resolution
of ± 2 ppm.
H. Current Regulator
A precision current regulator module was designed and
built for this system. The setpoint for this external current
regulator is provided by an existing in-house designed and
built power supply controller. The output of this regulator is
the drive voltage input for the Master Power Supply internal
regulator set on “Voltage” mode. Since the 30 kA Power
System is intended to be used with a wide range of load
inductance and resistance values, the current regulator
includes independent settings of load resistance and
inductance for tuning the control loop to each particular load.
The Resistance setting can be adjusted to 0.1, 0.5, 1, and 2
mΩ. The Inductance setting can be adjusted to 0.05, 0.5, 2, 4,
10, 25, 50, and 100 mH. Stability measurements performed at
10 kA show less than 3.5 ppm drift of the current receiver
output over a period of 2 hours (a typical magnetic
measurement duration.)
I. Temperature-controlled rack
To meet the strict current regulation and stability
requirements, the following modules are located in a
temperature-controlled rack: the HITEC zero-flux current
transformer electronics, the summing amplifier module, the
current regulator module, and the VME current transmitter
module.
J. Personnel Safety
In adherence to Fermilab’s safety regulations, personnel
safety is provided by a hardwired interlock chain. This system
is called the “Electrical Trip System” (ETS). In the event that
any cabinet door or any cover is opened while the system is
powered, the ETS system will take three simultaneous actions
to protect personnel from electrical hazards: (1) immediately

Page 3

1LL06, Page

3
initiate a power system phaseback; (2) fire the dump to
quickly extract stored energy; and (3) remove the external
interlock permissive to each PEI supply.
K. Equipment Safety
A Programmable Logic Controller (PLC) provides most of
the equipment safety interlock and system status monitoring.
This system is based on the 1500W Helium Refrigerator
Control System [3], and includes the series 505 Siemens
module for PLC and Input/Output modules, Ethernet
communication to the control room, and GE/FANUC FIX32
Human Machine Interface. The PLC monitors the status of
various temperature, flow, and other switches and devices.
There are 135 physical Input/Output points in this system.
In addition to the PLC system, a module called the “Fast
Trip Module” monitors the signals from each zero-flux current
transformer to detect a positive or negative current imbalance
caused by failures such as a shorted SCR. In the event that an
imbalance is detected, this module initiates a phaseback that
takes place in less than 16 ms to avoid damaging more SCRs.
Fig. 2. Half Coil voltage ripple measurements for SR02 (L = 0.038 mH) and
HFDA05 (L = 1.1 mH)
III. RIPPLE CHARACTERIZATION
Ripple from the 30 kA Power System is primarily a
function of load inductance (L). For instance, the performance
of the 720 Hz filter decreases rapidly for very small
inductance loads, and higher current ripple is expected for
such loads. Voltage ripple can be directly related to current
ripple by considering the load inductance dependence on the
current excitation frequency.
Voltage and current ripple measurement results are
presented for two magnets from the Nb3Sn HFM program:
SR02 (L=0.038 mH) and HFDA05 (L=1.1 mH). Inductance
values are for warm measurements at 20 Hz. Warm
inductance measurements at 1 kHz are a factor of two or more
lower. Voltage ripple was measured directly from half-coil
voltage taps (whole coil voltage ripple is twice the measured
amount.) Current ripple was measured from a current receiver
channel. Signals were digitized at 16-bit, 100 kHz with the
same system used for HFM Spike Studies [5].
Fig. 3. Half Coil voltage ripple FFT for SR02 and HFDA05

Fig. 4. Current ripple measurements for SR02 and HFDA05
A. Voltage Ripple
The voltage ripple was measured at several excitation
currents. Fig. 2 shows typical voltage ripple measurements
for SR02 and HFDA05 at 12 kA. Fig. 3 shows a Fast Fourier
Transform (FFT) of the signals from Fig. 2. In addition to the
expected 720 Hz component, there are several substantial
subharmonic ripple components at 60, 120, 180, 240, and 360
Hz. These subharmonic components are caused by imbalances
such as firing circuit imbalance, line imbalance, and
transformer imbalance. As shown, the larger inductance
magnet HFDA05 has a much larger voltage ripple. The levels
of each subharmonic is a function of excitation current and
overall the voltage has a minimum at approximately 7000 A.
B. Current Ripple
Fig. 4 shows the current ripple corresponding to the voltage
ripple of Fig. 2. The largest ripple corresponds to the lower
inductance magnet (SR02), which shows approximately 20 A
peak-to-peak current ripple. Frequency components of the
current ripple are similar to the frequency components of the
voltage ripple (Fig. 3).

Page 4

1LL06, Page

4
Fig. 6. HFDA05 Quench Current vs. temperature with and without 720 Hz
filter.
IV. RIPPLE AND HFM PERFORMANCE
A question often asked is whether power supply ripple may
affect the performance of Nb3Sn High Field Magnets. The
Nb3Sn reaction process can facilitate low interstrand
resistance. High frequency sub harmonics can then generate
Eddy currents in the conductor leading to AC loss heat
generation, which in turn could result in a reduction in the
quench current.
Several magnets from the Nb3Sn HFM program have been
tested using this 30 kA power system [6], [7]. SR01 and
SR02 are magnets with the same geometry and inductance but
have Nb3Sn conductor made from difference manufacturing
process. HFDA05 (HFDB02) have the same cable as SR01
(SR02) respectively but have a different coil geometry and
larger inductance. SR01 and HFDA05 were able to reach
their critical current under the ripple conditions presented in
Section III. For HFM magnets that did not reach their critical
current (e.g. SR02 and HFDB02), conductor instabilities
appear to be the reason for this performance shortcoming [4].
To investigate whether power supply ripple played a role in
the quench performance, further quench tests were performed
on HFDB02 and HFDA05. First, HFDB02 was powered in
series with a paralleled connection of two Fermilab main ring
dipoles. The added inductive load to the system reduced the
voltage and current ripple on HFDB02 by approximately a
factor of five. No improvement was observed in quench
performance. Next, HFDA05 was tested under extreme ripple
conditions by disabling the 720 Hz filter for each PEI supply.
This was accomplished by disconnecting the capacitor bank
from the bus. Fig. 5 compares the voltage ripple of HFDA05
with and without the 720 Hz filter. The voltage ripple without
the filter is approximately five times higher than with the
filter. When the magnet was ramped to quench with the filter
disabled, it again quenched at its critical current. This test was
performed at two different temperatures, and Fig. 6 compares
the quench current with the filter disabled with other quenches
at various temperatures with the filter enabled. As shown, the
performance of HFDA05 under extreme ripple conditions is
no different than under normal ripple conditions, following
the same temperature dependence as expected for a magnet
that has reached its critical current. These
suggest that the power supply ripple at the level produced by
the 30 kA Power System does not have an adverse effect on
the performance of High Field Magnets.
two results
V. CONCLUSION
Voltage and current ripple characteristics of a new 30 kA,
30 V dc power system at the Fermilab’s Magnet Test Facility
were measured for different inductance loads. The quench
performance of Nb3Sn High Field Magnets was analyzed
under different ripple conditions and it appears that power
supply ripple does not have an adverse effect on the
performance of these magnets.
ACKNOWLEDGMENT
We wish to acknowledge the contributions of Julius Lentz
to the design, fabrication and test of several new components
required for the 30 kA Power System.
REFERENCES
[1] W. Praeg, “A High Current Low Pass Filter for Magnet Power
Supplies,” IEEE Trans. on Industrial Electronics and Control
Instrumentation, vol. IECI-17, no. 1, pp. 16-22, February 1970
[2] A. Visser, “A 10,000 A 1000 VDC Solid State Dump Switch,” 4th
European Conference on Power Electronics, September 3-6, 1991,
Florence, Italy.
[3] R. Carcagno and R. Rabehl, “Controls Upgrade of the Fermilab Magnet
Test Facility 1500 W Helium Refrigerator,” Advances in Cryogenic
Engineering, vol. 45, pp. 1795–1802, Plenum Press, New York, (2000)
[4] S. Feher et al., “Sudden Flux Change Studies in High Field
Superconducting Accelerator Magnets”, Submitted to this conference.
[5] D. Orris et al., “Spike Detection: A System for Detecting Rapid
Superconducting Magnet Flux Changes”, Submitted to this conference
[6] A.V. Zlobin et al., “Development and Test of Nb3Sn Cosine(Θ) Dipoles
based on PIT strands”, Submitted to this conference.
[7] G. Ambrosio et al., “Fabrication and Test of a Racetrack Magnet Using
Pre-Reacted Nb3Sn Cable,” IEEE Trans. Appl. Supercond., vol. 13, no.
2, pp. 1284-1287

Fig. 5. HFDA05 Voltage ripple with and without the 720 Hz filter (The filter
disabled signal briefly reached input channel saturation in the positive peaks.)