Академический Документы
Профессиональный Документы
Культура Документы
Contents:
1. ABSTRACT 2
-
TABLE 3-1 COMPARISON OF SIMULINK, SABER AND PSPICE 8
4. OTHER OPTIONS
4.1 DIY
4.2 Spreadsheet
4.3Analogue Simulation
4.4Finite Element Analysis tools
5. REFERENCES 10
BY
James Moreland
ENOTRAC UK Limited
6’”Floor, Times House, Throwley Way
Sutton, Surrey, England SM1 4AF
Tel. +44 (0)181 770 3501, Fax. +44 (0)181 770 3502
e-mail: james.moreland@enotrac.com
Webs ite : http ://www.enotrac.com
8/ 1
IEE Colloquium “Power Electronic Systems Simulation” ENOTRAC
Choosing a Simulation Tool ENGINEERING-ORGANISATION-TRACTION
1. Abstract
Simulation of power electronic systems offers benefits in the design process by allowing various
options to be tried out before any hardware i s built. It also enables organisations to predict the
behaviour of their equipment under unusual operating or failure conditions which may be difficult or
dangerous to realise in the real world.
However, choosing the right simulation tool for your organisation can be a time consuming and
expensive process. This paper shows some of the factors to be considered when choosing a
simulation tool and gives a comparison, from the author’s experience, of three popular digital
simulation computer packages.
Simulation activity can be broken down into three overlapping areas based on how the power
electronic switching elements are viewed:
a) device level - if your design work uses terms like “electron”, “hole storage”, “charge
density” or “Boltzmann’s constant” then you probably need a physics-based
model. Typical activities would include gate drive and snubber circuit design.
b) converter level - key words here are “volts”, “amps” and “switching “frequency“. Typical
activities are design of modulation and switching algorithms. The most
suitable model is probably an equivalent circuit or a characterised switch
depending on how much detail is required. Switch models can generally be
characterised by on/off resistance and switching times to give a good
approximation to any particular IGBT, GTO etc.
c) application level - key words here are “volts”, “amps”, “kilo/megawatts” and “rpm” for motor
drives. Typical activities are design of complete variable speed drive
systems and static VAR compensation systems. The switching elements
- can be represented as ideal switches or the complete converter may be an
averaged model (this allows frequency domain simulation which is not
possible with switching elements).
In principle, it is possible to build a universal system model covering all three of the above views
and thus suitable for all users In practice, such a simulation would be unwieldy using current
computer hardware, with a time ratio of tens of thousands to one (clock time to simulation time)
and producing hundreds of megabytes of data. It is much more cost-effective to scale the
resolution of the model in keeping with the kind of questions you want to ask it and the’answers
you want to receive. For example, if you want to test the stability of a new variable speed motor
control algorithm you are not concemed with knowing the switching time of the IGBTs to the
nearest nanosecond.
IEE Colloquium “Power Electronic Systems Simulation” ENOTRAC
Choosing a Simulation Tool . ENGlNEERlNMRGANlSATlON-TRACTION
The evaluation process should be run as a mini project with clearly defined time scale, definition of
the system to be simulated, and method of assessment of how well the simulation package meets
your requirements and expectations.
The assessment will be subjective to some extent but it is important to state your requirements in
advance, especially if comparing two or more simulation tools. A table showing marks out of 10 for
each required property of the simulation tool can be a helpful way to make comparisons. The
requirements should also be weighted to reflect your priorities: for example tool A may have a
much bigger range of transistor models than tool 6, but if you only work with GTOs then this is not
important.
8/3
IEE Colloquium “Power Electronic Systems Simulation” ENOTRAC
Choosing a Simulation Tool ENGINEERINGORGANISATION-TRACflON
How good is the technical support? When you phone the help-line do you get an engineer or a
telephonist?
What training is available?
Is there support available apart from the supplier? E.g a user group.
Don’t be inflexible. Once you start using the tool, some of the facilities may suggest different
ways of working. You may be able to do things you hadn’t previously considered because they
were thought to be too difficult.
Talk to existing users with similar requirements to yours. If the investment is large, eg. involving
several licences, or an expensive tool (eg. Saber), it is well worth paying a visit to an existing
user (a serious software vendor should be able to put you in touch with users in a similar field).
This can shorten the evaluation process considerably. The existing user can also answer many
of the questions above from the ‘real-life’ perspective rather than from the marketing slant of the
software supplier/developer.
Although there are areas of overlap between the simulations, each has some particular qualities.
Sometimes the tools are used in a complementary manner: For example, Matlab can be used to
calculate parameters for a Saber model or even, vice versa, Saber can be used to derive
parameters for a Simulink model.
The most important aspect of any model is how closely it represents the real system. The
simulation tools described here may help to improve the range of your simulation work and make it
more productive, but they can only simulate what they are told to.
The simulation engine is based on Matlab’s powerful matrix processing core and several fixed and
variable time step solving algorithms are available.
The full power of Matlab is available for graphing and post processing. Matlab allows 2D and 3 0
graphs but the command interface is not very user friendly.
Simulink is mainly useful for application level modelling since it contains a wide variety of control
system models . It can be used for converter level modelling but it is less suitable since it models
system equations rather than system components. The author is not aware af any device level
modelling using Simulink but it could be done if the equations are known.
3.2 Saber
Saber, from Analogy Inc., was developed in the US in the mid 1980s as an engineering simulation
tool. The library of models (Analogy distinguish between templates which are generic devices,
characterisable by the user, and components which model specific, commercially available,
devices) covers several technologies, not just electrical/electronic but also mechanical, magnetic
8/4
IEE Colloquium “Power Electronic Systems Simulation” ENOTRAC-
Choosing a Simulation Tool ENGlNEERlNEORGANlSON-TRACTION
and hydraulic. All the models can be mixed in the same simulation since the fundamental network
equations are of the same form. This makes it popular in the aubmotive industry which is Saber’s
largest single user base.
Models are created by schematic capture either using Analogy’s own tool, Sabersketch, or by a
third party schematic capture tool. Users can create their own models (if the required function is
not in the library) using Analogy’s proprietary Hardware Definition Language, MAST. This includes
the ability to add models from technologies not included in the standard library e.g. pneumatic.
Saber is also compatible with SPICE models.
The simulation engine has separate simulators for the analogue and digital (event) domains and a
patented algorithm ensures synchronisation of the two simulations when required, without having
to evaluate the complete system at each time step. This is especially important for modelling
switching circuits where you want firing pulses to take effect on the power circuit at exactly the
right time but without having to specify very small time steps during the rest of the simulation. The
analogue simulation engine has a variable time step algorithm with various simulation controls
available to the user, however, the use of these is not well documented and should be left to the
experienced modeller. In particular, the variable time step algorithm does not cope very well with
high frequency oscillations e.g. after a switching transient, and may miss them.
Graphical display of the simulation output is good, with an intuitive user interface. A variety of post
processing tools are available including waveform measurements and a waveform calculator.
The Saber gaphical user interface (GUI) uses AIM, a programming language based on T C W k
(public domain programming language now administered by Sun Microsystems) so users can
modify it or create their own GUls.
Saber is useful for both application level and converter level modelling since it can be used to
simulate control system equations as well as actual components. Some work has been done at the
University of Washington in Seattle on using MASTISaber to develop a range of physics based
power device models (see reference [I I).
3.3 Pspice
Originally developed by Microsim using the SPICE2 (Simulation Program with Integrated Circuit
Emphasis) algorithms developed at UCB in the 1970s, Pspice is one of the most popular of the
commercial SPICE-based simulation packages. It is now owned by ORCAD, suppliers of the
popular PCB layout package.
Models are created by schematic capture or by text editing of a netlist. Users cannot create their
own components but large numbers are available from various suppliers since SPICE models are
interchangeable between all the SPICE based simulation tools . Unlike Saber, Pspice is only
aimed at electronic engineering applications, although it is possible to develop analogous models
using electrical elements (R, 1,C) for non-electrical problems eg. heat dissipation.
Graphical display of the simulation output is good, with a reasonable user interface. A variety of
post processing tools are available including waveform measurements and FFT. Graphs can be
copied to the Windows clipboard.
i=
Z
'c
0
816
c
0
e
a
.-L
a,
r,
a
c
8
2 J -I
818
IEE Colloquium “Power Electronic Systems Simulation”
Choosina a SimulationTool
4. Other Options
4.1 DIY
Writing your own simulator software for a particular system or product may be a valid option if your
organisation has the expertise. The simulation can be tuned to run efficiently for your particular
needs. Schematic input is not required since there is only a limited number of parameters.
Spreadsheets can be used to display the results.
Many organisations have old FORTRAN and C programs dating back 15 or 20 years. If these are
known to give valid results then they should not lightly be discarded. It may be possible to
integrate these with a general purpose simulation tool.
4.2 Spreadsheet
A spreadsheet can be used for simple fixed time-step simulations. It is particularly suited to
switching systems with repetitive waveforms. Graphic display functions are included and, on the
PC,the output will be compatible with other Windows programmes.
A newer alternative is to use high powered Digital Signal Processors to run code downloaded from
the simulation schematic capture tool. With sufficient processing power, this can also be fast
enough fora hardware-in-the-loop simulation but with the added advantage of having the graphing
and post-processing capabilities of the digital system. An example of this is the dSpace simulator
which can be used to simulate Simulink models. See their website, http://www.dspace.de for
further information.
Two examples of such tools are Maxwell by Ansoft (see http://www.ansoft.com) ‘and SLIM by
Alstom Research Centre (phone 01785 274661, fax 01785 274676).
SLIM is available in Unix and Windows 95/NT versions for 2D and 30 analysis and can be bought
outright or hired. The author has used the results of a SLIM analysis to create a frequency
dependent, complex impedance Saber model.
Maxwell is available in Unix and Windows 95/NT versions for 2D and 3D analysis. See their
website for further information.
IEE Colloquium “Power Electronic Systems Simulation”
Choosing a Simulation Tool
5. References
[l].Lauritzen P.0, Subramamanin Y , Bi Y , Green L. ‘An Efficient Way to Implement Electrical and
Thermal Device Models in MAST’, North American ASSURE Meeting, March 1997
- Printed
.
0 1998 The Institution of Electrical Engineers.
8/10 and published by the IEE, Savoy Place, London WCPR 0% UK.