High Voltage High Frequency Pulse Transformer, A Design Analysis

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

High Voltage High Frequency Pulse Transformer,

A Design Analysis
Dmitri Ivanov
Department of Industrial Electrical
Engineering and Automation
Faculty of Engineering
Lund University
July 7, 2015

Special purpose, high voltage, pulse transformers lytic models are used in order to acquire rough estima-
frequently have exceptionally high requirements tion of the parameters using mathematical expressions,
associated with their design. Such is the case for while finite element simulation provides a more precise
the prototype 1:25 kV step-up pulse transformer estimation. Finally, the measurements on the pulse
being developed for the klystron modulator system transformer are performed through short-circuit and
at the European Spallation Source. An insulation no-load tests. The results of the three methods are
requirement of 150 kV as well as strict specifica- compared in order to verify their credibility.
tion regarding the voltage transformation that the In figures 2 and 3 the no-load and short-circuit con-
system performs, makes the construction of the ditions are displayed in finite element program FEMM.
device highly challenging. Evaluation of the elec- Among other parameters Lmag , L1leak and L2leak are
tromagnetic, electrostatic and geometric design of obtained through the simulations.
the transformer therefore provides a much needed
insight into vital parameters of the device.
I. C ORE OF THE ANALYSIS
At the center of the performed transformer analysis
lie two main models; high frequency adjusted equiva-
lent circuit model and finite element based electrostatic
model.
The equivalent circuit model condenses magnetic and
electric qualities of the transformer to a set of circuit
components in such way, that when voltage is applied
to the circuit, it exhibits qualities similar to those of the
real transformer. The equivalent circuit is seen below Fig. 2. Open circuit FEMM simulation
in figure 1.

Fig. 1. Equivalent scheme of the pulse transformer

The component values are estimated using analytic,


measurement and finite element based methods. Ana- Fig. 3. Short-circuit FEMM simulation
Both plots display magnetic field lines, in addition to
magnetic flux density, which is shown in color across
the simulation region.
The results of equivalent circuit simulations all point
toward two main results. Primarily, the values of
leakage inductances, L1leak and L2leak are of great
importance due to large portion of the voltage expected
to fall over the load falling over these parasitic elements
instead. In the figure 4 importance of reducing the
leakage inductance is shown.

Fig. 5. Current waveform measured using oscilloscope

Fig. 4. Influence of leakage inductance on output waveform


Fig. 6. Current waveform acquired through nonlinear Simulink
simulation
As is apparent, the voltage over the load does only
reach its intended top value during each half period if
the leakage inductance is low enough. For the prototype reached through asymmetry of the excitation waveform,
pulse transformer the leakage inductance is expected to lower operation frequency or higher input voltage.
be around 4.5 µH, which threatens the voltage capabil- Electrostatic analysis of the system offers another
ity of the system at nominal load. Different estimation point of view on its functionality. Namely, high voltage
methods predict a output voltage between 18.3 kV and devices are prone to develop corona effect and electrical
22.4 kV for the transformer, with 19.2 kV being the arcing if need for sufficient creepage and clearance
minimal allowable value at nominal load of 1 kΩ. distances as well as avoidance of sharp edges in the
Secondarily, saturation tendencies are observed in design are overlooked. A clear indication of such
the core throughout the simulations. This is supported effects can be found through electrostatic model of
by measurements performed on the prototype. Mild the device. Corresponding model for the transformer
saturation behavior can be observed in figures 5 and is seen in figure 7.
6. First one represents oscilloscope measurement at no The simulation shows that no electrical arcing is
load while second one displays simulation results in expected when the transformer resides in oil with
Simulink under same conditions. dielectric strength around 12 MV/m. Notably, if pockets
As seen in the figures, current waveform values start or bubbles of air are present in the submerged trans-
to rise significantly toward end of each half period. If former the risk of arcing increases due to dielectric
the saturation of the core is allowed to reach critical strength of air being 3 MV/m. Corona effect, which
values, very high current spikes are to be expected occurs at lower field strengths can however be present
in the input current, threatening to damage the device in the design, resulting in modest power losses and
and destabilize its output. High saturation levels can be possibly solid dielectric degradation.
Fig. 7. Electrostatic simulation of transformer cross section in
Finite Element Method Magnetics

II. C ONCLUSIONS OF THE ANALYSIS


Following the evaluation of the transformer proto-
type a conclusion is drawn that the design is functional
in all aspects, however, room for improvement and
safety margins is present.
Primarily the leakage inductance is to be minimized
by introducing measures which reduce the leakage flux.
Such measures include reducing number of transformer
turns on each winding, increasing the length of both
primary and secondary winding and decreasing the area
between the two. This way reduction of voltage drop
can be achieved for the whole design.
While altering the construction, it is important to
remember that change of one parameter can uninten-
tionally compromise other aspects of the design. This
results in need of a fine balance between sufficient
clearance between the windings and saturation of the
core on one side and reduction of leakage inductance
on the other.
A viable approach to the matter is to reduce the
saturation levels through increase in the core cross
section by use of an additional magnetic core. This
way number of turns and thereby leakage inductance
can also be reduced.
Among other improvements use of low permittivity
insulators in order to decrease the parasitic capacitance
is advised for the primary winding. Another alteration
is to increase the radius of the anti-corona devices,
which not only reduces corona effect but also decreases
the chances of potentially very hazardous electrical
arcing.
While the performed analysis revolves around a pro-
totype transformer, the extracted results can be applied
to a wide range of similar devices.

You might also like