CERN-ATS-2012-057
15/03/2012
EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH
CERN - ACCELERATORS AND TECHNOLOGY SECTOR
Performance Improvement of a Measurement Station
for Superconducting Cable Test
P. Arpaia
1,2
, L. Bottura
2
, G. Montenero
1,2
, S. Le Naour
2
1 Engineering Department, University of Sannio, Benevento, Italy
2 CERN, Geneva, Switzerland
A fully digital system, improving measurements flexibility, integrator drift, and current control of superconducting
transformers for cable test, is proposed. The system is based on a high-performance integration of Rogowski coil
signal and a flexible direct control of the current into the secondary windings. This allows state-of-the-art performance
to be overcome by means of out-of-the-shelf components: on a full-scale of 32 kA, current measurement resolution of
1 A, stability below 0.25 Amin
-1
, and controller ripple less than 50 ppm. The system effectiveness has been
demonstrated experimentally on the superconducting transformer of the Facility for the Research of Superconducting
Cables at the European Organization for Nuclear Research (CERN).
Submitted to "Review of Scientific Instruments"
Geneva, Switzerland
CERN-ATS-2012-057
March 2012
Abstract
1/21
Performance improvement of a measurement station for
superconducting cable test
Pasquale Arpaia,
1,2
Luca Bottura,
2
Giuseppe Montenero,
1,2
Sandrine Le Naour
2
1
Engineering Department, University of Sannio, 82100 Benevento, Italy.
2
European Organization for Nuclear Research (CERN), 1217 Geneva, Switzerland
A fully digital system, improving measurements flexibility, integrator drift, and
current control of superconducting transformers for cable test, is proposed. The
system is based on a high-performance integration of Rogowski coil signal and a
flexible direct control of the current into the secondary windings. This allows state-of-
the-art performance to be overcome by means of out-of-the-shelf components: on a
full-scale of 32 kA, current measurement resolution of 1 A, stability below
0.25 Amin
-1
, and controller ripple less than 50 ppm. The system effectiveness has
been demonstrated experimentally on the superconducting transformer of the Facility
for the Research of Superconducting Cables at the European Organization for Nuclear
Research (CERN).
I. Introduction
Superconductivity is a technology with a declared interest in several fields of physics
and engineering
1
. A key design parameter for any large-scale application of
superconductivity is the current carrying capacity, also referred to as “critical current”. Its
assessment needs first and foremost for an accurate measurement of the voltage-current
characteristic of the sample, in general, function of temperature, current, and magnetic field
2
.
This is a relatively delicate task for single wires, nowadays carried out through quasi-
industrial standards. On the other hand, for large-size cables, facilities of appropriate
dimensions and functionality are few, mainly owing to the difficulty and cost of providing a
large and complex set-up for assessing the device properties as a function of the
abovementioned parameters
3-8
.
Cable critical current tests commonly involve current levels in the order of the tens
of kA, in principle supplied by large power converters. However, main drawbacks are
significant equipment cost, and the need for large current leads, resulting in high cryogenic
loads and operating costs. For this reason, in this range of current, an interesting alternative is
2/21
to use superconducting transformers
9-17
. A low current is fed to a superconducting primary
winding with a large number of turns, inductively coupled to a superconducting secondary
with a much smaller number of turns and directly connected to the sample under test. The
modest primary current, usually in the range of 100 A, can be generated with relatively
inexpensive and standardized power supplies, and the current feed-through into the cryogenic
environment can be optimized to have lower cryogenic load by orders of magnitude. Such a
device provides test capability at moderate capital and operating cost.
Beyond the obvious issues on the performance and protection of the primary and
secondary windings, one of main concerns in the operation of a superconducting transformer
for cable test is a suitable control, with a precise and accurate measurement of the secondary
current. Indeed, current measurement is a key component of the control loop. Resistive losses
due to the interconnections between the secondary and the sample lead to an unavoidable
decay of current, unless the primary current is adjusted continuously to compensate and
maintain the sample current at the desired set value. The control loop, on the other hand, must
account for the physical characteristics of the coupled system formed by the primary winding,
and its power supply, the secondary, and the current transducer.
In
10
and
11
, advanced systems for measuring the secondary current and for
compensating its decay are proposed. Both are based on similar concepts for the
improvement of the current measurement quality and the implementation of a suitable
control. The sensing elements for the current are two Rogowski coils connected in anti-series,
providing a good rejection to parasitic couplings. The transducer signal is integrated in time
by using digital integrators, and the integral is converted to current by means of a suitable
calibration coefficient. The signal conditioning and the control are then implemented by
means of customized analog electronics. In
10
, main drawbacks are the resolution and the
offset of the digital integrator. In
11
, these issues are addressed both by implementing a
custom FPGA-based integrator with higher resolution and by minimizing the residual offset
via a dedicated procedure. However, the control loop is still analog (proportional action).
Furthermore, in both the systems, instead of directly monitoring the secondary current, a
voltage produced via a function generator is used as reference signal on the primary power
supply, by giving rise to an indirect control strategy. This makes rather complex both the
common operation of defining an arbitrary cable current cycle and the detection of system
faults: a test cycle defined in terms of secondary current has to be converted in a
corresponding voltage. This requires that the closed-loop transfer function is known with
satisfying accuracy for all the operating conditions.
3/21
In this paper, a flexible system improving the state-of-the-art of superconducting cable
testing based on transformers is proposed to overcome state-of-the-art solution by meeting
the overall target of a secondary current control for a 32 kA full range, with resolution better
than 3 A, relative ripple less than 10
-4
, and stability below 0.5 Amin
-1
. The secondary current
measurement is improved by compensating the residual offset of a high-performance digital
integrator by adjusting the self-calibration time. The measurement system is integrated in a
fully-digital control loop, with all the related benefits
18
, i.e. good noise margins, easiness in
the implementation/modification, and so on. In particular, in Section 2, the proposed
measurement system with the associated digital control algorithm is illustrated. In Section 3,
experimental results from the on-field characterization and validation of the proposed system
at CERN, in the Facility for Research on Superconducting Cables (FReSCa)
4
are illustrated.
II. The Proposed System
In Fig. 1, the architecture of a measurement station for superconducting cable test,
based on a transformer for supply and Rogowski coils for current measurement, is reported.
The objective is to generate a test current in the sample I
s
, i.e. in the transformer secondary,
ideally equal to the set point I
ref
. At this aim, the main issue is to provide a direct control of
the secondary current I
s
. The control block, according to the feedback signal I
m
from the
measurement system, acts to provide a reference voltage V
ref
to the power supply of the
primary, a voltage-controlled current source. The source drives the current I
p
into the
primary, by inducing a secondary current I
s
. In turn, I
s
is sensed by means of two Rogowski
coils, suitably positioned in phase opposition and connected in anti-series in order to reject
parasitic electromagnetic couplings to stray fields. The coils voltage signal V
RC
is
proportional to the secondary current rate dI
s
/dt. The secondary current is then obtained by
integrating the differential signal V
RC
from the Rogowski coils. The measured current is
finally compared to the reference I
ref
in order to generate the feedback signal I
m
compensating
for resistive losses.
FIG. 1. Architecture of a transformer-based measurement station for superconducting cable test.
4/21
In this architecture, the fundamental elements are the measurement system and the
control strategy. The former has to provide high-quality measurements, by minimizing
undesired uncertainty effects, such as integrator drift and noise. The latter has to follow
closely the reference signal, i.e. the desired secondary current I
s
, by accurately driving the
current into the primary I
p
. Moreover, the overall system should be flexible, in order to allow
quick modifications of the desired waveforms according to the test needs.
The above-mentioned requirements are met by exploiting high-performance numerical
integration and digital control algorithms, as described below.
A. The measurement and control system
In Fig. 2a, the architecture of the measurement and control system is shown. The
control reference V
ref
is generated by a digital waveform generator, with at least 16 bits of
resolution in the input range of the voltage-controlled current source (Fig. 2b) in order to
accurately control I
p
. The Rogowski coils signal V
RC
, acquired by means of a digital
integrator, is related to the secondary current in the ideal case as:
(1)
where K
RC
(VsA
-1
) is the sensitivity coefficient of the Rogowski coils in anti-series
configuration. Then, after digital integration, the measured magnetic flux is:
(2)
where n stands for a discrete time instant, and
offset
(n) is the undesired flux contribution
arising from the voltage offset on the data acquired from the integrator. This integration
equivalent offset signal arises from both all the circuit before the integrator input as well as
from the integrator itself. In practice, design objectives are achieved by imposing the
following requirements on the integrator:
Signal-to-Noise Ratio (SNR) better than 90 dB, for resolution in current readings
better than 1 A on a full-scale of 32 kA;
