Simulation of Impulse Generator Followed by Practical Verification

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Simulation of Impulse Generator Followed by

Practical Verification
Ludek Pelikan, Michal Krbal, Jaroslava Orsagova
Department of Electrical Power Engineering
Brno University of Technology
Brno, Czech Republic
xpelik15@stud.feec.vutbr.cz, krbal@feec.vutbr.cz, orsagova@feec.vutbr.cz

Abstract— Paper focuses on simulating a real impulse Voltage pulse waves can have many different shapes, but
generator in LTspice in such a way that it is close to real for precise testing, it is necessary to standardize them. The
conditions. There is a description of comparison of measured and voltage curve defined according to the standards is shown in
simulated important monitored values. These are the front time Figure 1.The monitored parameters include front time T1 which
T1 and the time to half value T2, which are the monitored is defined as the time T multiplied by 1/0.6, and time to half
parameter of waves for each atmospheric impulse voltage test. value T2. Indication T1/T2 is usually used to describe the wave.
Keywords— High Voltage Impulse Generator, Simulation, Two types of voltage waves are mainly used for impulse
Impulse Voltage Wave Shape, LTspice, Overshoot. voltage testing. The first and most used type of wave is full
lightning pulse, which is characterized by a short duration with
I. INTRODUCTION indication 1.2/50 µs. On the other hand, the switching pulses
are characterized by a longer duration – T1 is more than 20 µs.
Pulse overvoltage typically arise from indirect and direct Standardly used test switching wave is 250/2500 µs.
lightning strikes and switching in electricity systems. For Unfortunately, the accuracy of these times is difficult to
testing insulation of power equipment against pulse achieve, therefore some tolerances are standardly allowed. For
overvoltage effects are used Impulse generators. These full lightning pulse, the standards specify a T1 front time
generators are the main elements of high voltage test rooms tolerance of 30 %. Thus, T1 must be in the range of 0.84 µs to
and can be designed as voltage, current or mixed. From this 1.56 µs. A tolerance of 20 % is accepted for time T2, that is,
point of view the work will deal only with voltage generators. between 40 µs and 60 µs. The tolerances for switching pulses
Due to the limiting size of the main generator element, i.e. the are different because the difficulty of keeping them is different
active capacity, the devices are manufactured as multistage. in terms of the lengths of times T1 and T2. Therefore, the
This makes it possible to generate voltage waves in the order of tolerance for time T1 is 20 % and for T2 is 60 %. Another
hundreds to millions of volts using commercial electrical tolerance that is monitored is the peak of the voltage wave. In
elements. These devices should generate voltage waves with this case, it is easy to achieve a fairly accurate value, therefore
precise parameters. its tolerance is at most 3 %. The parameters of the wave depend
on all elements of the electrical circuit, i.e. besides the internal
connection also on the type of load. Thus, from certain load
values, the generator may not meet the requirements for
maximum tolerances of the waveform. In these cases, it is
necessary to adapt the connected circuit to the parameters of
the wave. It is possible to change the front and tail resistors,
change the number of stages, or insert other elements into the
circuit. An example of a simple diagram of a one-stage
generator is shown in Figure 2. Cs represents capacitance of
active capacitor, capacitance Cl is a load, Rs frontal resistor and
Rp tail resistor.

Fig. 1 Full lightning impulse according to the standards


Fig. 2 Basic scheme of impulse generator
This research work has been carried out in the Centre for Research and
Utilization of Renewable Energy (CVVOZE) in the research infrastructure
CVVOZEPowerLab with support of the project LM2015092. Authors
gratefully acknowledge financial support from the Ministry of Education,
Youth and Sports of the Czech Republic.

978-1-7281-9479-0/20/$31.00 ©2020 IEEE

Authorized licensed use limited to: Technische Universitaet Graz. Downloaded on February 23,2021 at 09:17:26 UTC from IEEE Xplore. Restrictions apply.
Output voltage shape can be performed using the 2nd order
differential equation. After its solution, the generator output
voltage is determined by the Equation 1.

v2(t) = V1·k· (eα t - eα t)


1 2
(1)

Constants k, α1 and α2 are dependent on values Rp, Rs, Cs and


Cl. However, this often-shown mathematical description does
not consider other elements in the circuit and its parasitic
properties. For the generator user it may not be easy to find the
appropriate wiring for the device under test. The calculation
method is very difficult. Therefore, graphical-numerical
methods are used. Finding the values of resistors R1 and R2 can
be done by Angelini's graphical-numerical method, using the
known values of Cs, Cl, T1 and T2. Another method is used to
monitor the course and measure the time constants from which
Fig. 4 Waveform change depending on load capacity
further calculations are made. However, as mentioned above,
parasitic properties such as circuit inductance are difficult to
respect. But the next possibility to make finding the wiring Figure 3 shows part of the initial design of the simulated
easier is to create a regular Circuit Simulation which can help circuit, respectively two stages of generator, power supply and
solve the problem quickly. its load. The currently used values of the individual
components are listed here. The spark gap in the circuit is
II. SIMULATION simulated by voltage switches S. The Cdiv capacitor and Rdiv
resistor are a voltage divider that is necessary for the practical
The simulation was performed in LTspice XVII. The evaluation of high voltage measurements over time and is
choice of this program was thanks to him its simplicity and its therefore a constant load. The Rse resistor is a non-inductive
functions, which allowed all necessary operations. The resistive band that contributes to the waveform and allows the
disadvantage of this program simulation is the difficult generator to be connected to the load.
reading of values of waves. The simulated object was a
university impulse generator SGΔA 1000 – 100 at the Brno First, a theoretical simulation of all stages was performed
HV laboratory CVVOZEPowerLab. It is a ten-stage device at the highest possible voltage charging voltage. No parasitic
with a maximum voltage rating 100 kV per stage and a effects in the circuit were considered in these simulations. It is
maximum total energy 100 kJ. also considered that the chopping spark is completely
disconnected. The only load is theoretically a capacitive
divider to which additional purely capacitive loads have been
added. As an example of load capacity change, Figure 4 shows
a capacity change in the range of 0 to 3 nF in 0.5 nF steps.
With higher load capacity, the maximum voltage value
decreases, with both T1 and T2 increasing.

With this measurement, the limit values for the current


generator connection could be read. It has been found that the
maximum possible theoretical voltage value is 942.3 kV at no
added load. At this maximum voltage, T1 is 0.857 µs and T2 is
50.32 µs. Both parameters meet the condition of atmospheric
voltage tolerance. Load limits beyond which tolerance limits
can no longer be met can be important in finding the
appropriate circuit connections. These limits can be easily
found through the interaction of the load capacity and the
subsequent reading of the time values. Obviously, the time T2
will not be problematic in terms of high tolerance and that
considering the existing wiring with the divider, the lower
limit of time T1 will not be problematic either. Thus, the
interaction will take place in terms of the maximum time T1,
i.e. 1.56 µs. Table I. shows the simulated and observed values.
This maximum possible theoretical capacitive load is
Fig. 3 Initial simulation of impulse generator approximately 1.14 nF.

Authorized licensed use limited to: Technische Universitaet Graz. Downloaded on February 23,2021 at 09:17:26 UTC from IEEE Xplore. Restrictions apply.
TABLE I. PARAMETERS OF SIMULATED IMPULSE WAVES The measured waveforms were very different from the
C (nF) V (kV) T1 (µs) T2 (µs) simulated waves. This was mainly due to unaccounted
parasitic influences. The most influential parasitic matter is the
1.00 927.1 1.469 51.63
inductance in the circuit. An important characteristic of the
1.10 925.5 1.510 51.76 inductance in this circuit is that it superimposes overshoots in
1.14 924.9 1.558 51.81 the wave form. In particular, the first overshoot indicates the
1.15 924.7 1.567 51.82 magnitude of the wave. The entire residual wave size is
1.30 922.0 1.672 52.04 thereby reduced, and the reading times change. To reduce the
2.00 911.2 2.115 53.04 influence of overshoot, the measured wave is corrected by the
amplitude-frequency function. This correction causes a
III. PRACTICAL MEASURING reduction in the displacement of the entire carrier wave. The
Several measurements with a different capacitive load correction will mainly affect large overshoots with high series
were made for comparison with simulations. A capacitive inductance.
divider is used for the voltage in time measurement, which
represents a constant load of 0.67 nF. Another burden is the Figure 6 and Figure 7 compares simulations without
chopping spark. It is a device with stages divided by capacity, considering parasitic properties with the measured values for
which can be half bypassed. It therefore serves as a constant three, five and ten stages of impulse generator. This
load with a voltage divider. In our test room, a 1.2 nF spark comparison of values is not very accurate due to inaccurate
gap is normally connected even when not in use. Thus, after load capacitance values assigning.
half bypassing, its load is 2.4 nF. High-voltage foil capacitors
were used as additional variable load. The first capacitor had a
capacitance value of 0.6 nF, second 3.33 nF.

Figure 5 shows the connected device in the circuit. On the


left is impulse generator, next to a chopping spark gap, in front
of it a voltage divider. The testing was carried out with
different variations of the loads involved. The selected voltage
value was 200 kV to be able to measure at different numbers
of generator stages. At low voltages there may be a problem
with spark ignition when using a high number of stages.

Fig. 6 Differences of simulated and measured T1

Fig. 7 Differences of simulated and measured T2


Fig. 5 Real view of wired parts of circuit

Authorized licensed use limited to: Technische Universitaet Graz. Downloaded on February 23,2021 at 09:17:26 UTC from IEEE Xplore. Restrictions apply.
IV. SIMULATION WITH APPROACHING REAL VALUES
Due to inaccurate results of the previous chapter, further
adjustments were made in the simulation. First, it was
necessary to find out what inductance occurs in the circuit.
Consideration was given to the size of the measured
overshoot. The measured overshoot at three connected stages
and the added load of 1.2 nF is shown in Figure 8, where the
inductance in the circuit produces an overshoot of about 9 kV.
Obviously, the inductance in the circuit should be in the order
of tens µH. One of the known values was the divider
inductance, as reported by the supplier, which was inserted
into the simulated circuit. Furthermore, an impedance was
inserted into the circuit in which the same overshoot occurs as
in the real measurement. For three-stage wiring, an
approximate value of 34 µH for the series-connected
inductance was found due to overshoot. Unfortunately, with
this simple inductance connection, T1 time is decreased.
Specifically, the value dropped to 0.767 µs, which does not
fully correspond to the measured value of 1.128 µs. Wave
correction using the amplitude-frequency function was
neglected. On the other hand, the time T2 approaches 48.28 µs, Fig. 9 Final simulation of impulse generator
which is closer to the measured value of 48.92 µs.

The next step was to find the inductance for one stage. The
higher the inductance of the stage, the time T1 increases. A
good variant seems to be a parasitic inductance of 5.4 µH per
stage and a rest loop Lrl inductance of 17.5 µH. The resulting
simulated part of the circuit is shown in Figure 9. Using this
assignment, the simulated and measured times T1 and T2 were
compared again. The results are shown in Figures 10 and 11.
The results show that the resulting waveforms are similar to
the real values due to the included parasitic inductances. It can
be also seen that larger errors usually occur with higher
numbers of generator stages involved. This is due to the
omission that, with a lower number of generators connected,
short-circuiting bands are used, which add to the circuit
additional rest loop inductance. Therefore, when using multi-
stage simulation, it is appropriate to reduce Lrl appropriately. Fig. 10 Differences of simulated and measured T1 for final simulation

Fig. 8 Measured voltage curve with overshoot


Fig. 11 Differences of simulated and measured T2 for final simulation

Authorized licensed use limited to: Technische Universitaet Graz. Downloaded on February 23,2021 at 09:17:26 UTC from IEEE Xplore. Restrictions apply.
V. CONCLUSION further design of simulation models, for example when
The simulation of the impulse generator model helps to connecting other elements to a circuit, such as a Glaninger
better understand the circuit and can be used in decision circuit.
making in its modifications. Modeling in LTspice is suitable REFERENCES
due to the quick reading of wave parameters. Unfortunately, at
[1] IEC 60060-1 “High Voltage Test Techniques Part 1: General Definitions
the known maximum value of the output voltage, it is and Test Requirements,” 2010.
necessary to find at least the approximate supply voltage. This [2] A. Veverka, “Technika vysokých napětí,” [in Czech].
paper can help or inspire in creating own simulation models of [3] K. Veisheipl, “Simulation of the High Voltage Impulse Generator,”
impulse generators. The knowledge gained from these International Scientific Conference on Electric Power Engineering
measurements and the creation of this paper is very useful for (EPE), pp. 1-5, 2016.
[4] J. Wolf, “Impulse Voltage Generator Type SGΔ, Operating
Instructions,” 2004.

Authorized licensed use limited to: Technische Universitaet Graz. Downloaded on February 23,2021 at 09:17:26 UTC from IEEE Xplore. Restrictions apply.

You might also like