| Review: |
The Electronics Workbench (EWB) appears to be your Virtual Electronics
Laboratory of the 21st. century -useful both in tertiary education as well
as in (virtual!) prototyping.
It is in fact an electronic simulation software developed by a Canadian
company, around the analog SPICE3 simulator kernel and extended by a smart
switch level digital simulator. Version 4 has been on offer in 1996/97 but
it can be upgraded now to an improved version 5.
The product covers a range of options available at different prices. All are
based on the same simulation kernel, have the same simulation power but have
certain limitations.
For a quick review, options within the version 4.1 are:
* EWB engineering pack at around A$950 (A$1600 for ver.5) - intended for
simple verification of electronic circuits and equipped with:
libraries of thousands of commercial components,
facility to import/export standard Spice .cir files, and
offering export to PCB (printed circuit board) design netlist files (for
software like OrCAD, Protel, Eagle).
* EWB educational pack at around A$500 - aimed at tertiary users including
laboratory classes and minor projects, offering only 200 component library
and no extended facilities.
* EWB student pack at A$75 - with a minimum component library and with
limitation of circuit size to 25 analog components.
* ETB (Electronics Test bench) at approx. A$170 - an add on software tool
for lecturers to generate multiple choice tutorials, including circuit files
generated using the main EWB software.
Certainly educational discounts are available.
EWB is very fast, faster than the conventional PSPICE-based simulators in
mixed analog-digital simulation mode and it is very easy to learn and use.
For those reasons only it can make even the Systems and Software stream
students attracted to solving hardware problems - a phenomenon hard to
believe though witnessed by myself on a number of occasions.
Its limitations may not be noticed by users - that laboratory is virtual, ie
it is idealised. The noise is not present and output plots (whether accurate
or not) are produced with a wide choice of simulation parameters. Therefore
engineering users need own good judgment and some caution.
From a teacher's perspective the most important aspects are: the EWB's rapid
learning curve, its simplicity and the fact that it is 'unbreakable'. While
the virtual world is slowly taking over, the maintenance of conventional
laboratories used by multiple classes remains a costly aspect of teaching. A
squad of workaholic technicians lead by heroic academics may be put to rest
by running all the usual instruments off the screen using EWB installed on a
local network. Students will still be able to become familiar with: Waveform
Generator, Oscilloscope, Multimeter, Logic State Analyzer, Logic Vector
(Word) Generator, Bode (frequency response) Plotter and some sophisticated
components like Analog-Digital Converters or an Arithmetic Logic Unit.
They will not be discouraged by noise, faulty wires, dry soldering points
and by difficulties resulting from their impatience or a lack of common
sense (they will be hit by the 'real world' later on when they will have
gained some experience).
EWB 4.1 offers schematic editor and analysis tools (instrument icons) with
analysis options/parameters available from menus. Components, grouped in
categories, unfold from toolbar icons. Semiconductors are backed by full
models but may become idealised by ignoring/zeroing most parameters. Special
devices, like controllable switches, analog multipliers, switchmode
converters or nonlinear devices represent several lines of a background
SPICE listing. Information on their detailed implementation is rather brief
and the user needs to gain own experience.
There are other attractive features, such as: an attached window for
description/comment, nested design hierarchy made available through user
defined macros (subcircuits), functional (ideal) device models and even
symbolic operators (from ver.5) like differentiators, integrators, s-plane
filters etc.
Be aware that DC sources need several milliseconds to produce full output -
like in a real lab (that software is realistic). Therefore to observe short
transients - Transient Analysis option should be avoided or else the
simulation can take an eternity. Use Steady State analysis option instead
and apply repetitive excitation - as one would do in a real laboratory.
For a balanced review I may need to drop a few drawbacks: selection of
components has Northern American bias, only single instruments are available
and the rubberbanding tends to produce fractals.
Other problems have already been removed in version5: the drawing area is
larger and zooming is possible, the transmission line and pulse or piecewise
sources are available and separate DC or Fourier analysis can be run.
Version5 offers also access to most simulation and convergence options, to
make regular engineers happy. For that reason it is very different from
version 4.1 which offers access to two only: the number of iterations per
timestep and the limiting accuracy.
Using either 486DX-66, PentiumC90 or even Pentium Pro200 with small or large
number of monitored nodes had little effect on the performance but the size
of the circuit had. Generally as a circuit was approaching the size of full
screen, the simulator was struggling and would eventually give up
complaining about too demanding iteration conditions (although I was already
at the most relaxed end of the scale). Version5 (32 bit) performed a bit
better. Like in a real laboratory, problems may be avoided by developing
circuits gradually, starting at a minimum complexity and using idealised
models. This is because convergence difficulties are sparked by the
accumulation of demanding or critical nodes in a circuit, especially when
their requirements are contradictory (like the size of iterations for
integrating components vs sharply nonlinear ones).
Which version should be purchased for a tertiary environment?
Get a few engineering packs for projects, both v4.1 and 5 (for U/G and for
more advanced users) and get a quantity of educational packs just to run an
U/G virtual lab.
Let students buy their own cheaper versions so that they could do their work
at home as well. And to let the staff generate tutorials/assignments get a
few ETBs.
Is EWB useful only for Electronic Engineers?
Yes, in the case of version 4 but the new version 5 has general symbolic
operators and may be used to model non-electronic systems. However the user
needs to know the operation of the remaining electronic blocks to be able to
utilise them.
The Electronics Workbench (EWB) appears to be your Virtual Electronics Laboratory of the 21st. century -useful both in tertiary education as well as in (virtual!) prototyping.
It is in fact an electronic simulation software developed by a Canadian company, around the analog SPICE3 simulator kernel and extended by a smart switch level digital simulator. Version 4 has been on offer in 1996/97 but it can be upgraded now to an improved version 5.
The product covers a range of options available at different prices. All are based on the same simulation kernel, have the same simulation power but have certain limitations.
For a quick review, options within the version 4.1 are:
* EWB engineering pack at around A$950 (A$1600 for ver.5) - intended for simple verification of electronic circuits and equipped with:
libraries of thousands of commercial components,
facility to import/export standard Spice .cir files, and
offering export to PCB (printed circuit board) design netlist files (for software like OrCAD, Protel, Eagle).
* EWB educational pack at around A$500 - aimed at tertiary users including laboratory classes and minor projects, offering only 200 component library and no extended facilities.
* EWB student pack at A$75 - with a minimum component library and with limitation of circuit size to 25 analog components.
* ETB (Electronics Test bench) at approx. A$170 - an add on software tool for lecturers to generate multiple choice tutorials, including circuit files generated using the main EWB software.
Certainly educational discounts are available.
EWB is very fast, faster than the conventional PSPICE-based simulators in mixed analog-digital simulation mode and it is very easy to learn and use. For those reasons only it can make even the Systems and Software stream students attracted to solving hardware problems - a phenomenon hard to believe though witnessed by myself on a number of occasions.
Its limitations may not be noticed by users - that laboratory is virtual, ie it is idealised. The noise is not present and output plots (whether accurate or not) are produced with a wide choice of simulation parameters. Therefore engineering users need own good judgment and some caution.
From a teacher's perspective the most important aspects are: the EWB's rapid learning curve, its simplicity and the fact that it is 'unbreakable'. While the virtual world is slowly taking over, the maintenance of conventional laboratories used by multiple classes remains a costly aspect of teaching. A squad of workaholic technicians lead by heroic academics may be put to rest by running all the usual instruments off the screen using EWB installed on a local network. Students will still be able to become familiar with: Waveform Generator, Oscilloscope, Multimeter, Logic State Analyzer, Logic Vector (Word) Generator, Bode (frequency response) Plotter and some sophisticated components like Analog-Digital Converters or an Arithmetic Logic Unit.
They will not be discouraged by noise, faulty wires, dry soldering points and by difficulties resulting from their impatience or a lack of common sense (they will be hit by the 'real world' later on when they will have gained some experience).
EWB 4.1 offers schematic editor and analysis tools (instrument icons) with analysis options/parameters available from menus. Components, grouped in categories, unfold from toolbar icons. Semiconductors are backed by full models but may become idealised by ignoring/zeroing most parameters. Special devices, like controllable switches, analog multipliers, switchmode converters or nonlinear devices represent several lines of a background SPICE listing. Information on their detailed implementation is rather brief and the user needs to gain own experience.
There are other attractive features, such as: an attached window for description/comment, nested design hierarchy made available through user defined macros (subcircuits), functional (ideal) device models and even symbolic operators (from ver.5) like differentiators, integrators, s-plane filters etc.
Be aware that DC sources need several milliseconds to produce full output - like in a real lab (that software is realistic). Therefore to observe short transients - Transient Analysis option should be avoided or else the simulation can take an eternity. Use Steady State analysis option instead and apply repetitive excitation - as one would do in a real laboratory.
For a balanced review I may need to drop a few drawbacks: selection of components has Northern American bias, only single instruments are available and the rubberbanding tends to produce fractals.
Other problems have already been removed in version5: the drawing area is larger and zooming is possible, the transmission line and pulse or piecewise sources are available and separate DC or Fourier analysis can be run.
Version5 offers also access to most simulation and convergence options, to make regular engineers happy. For that reason it is very different from version 4.1 which offers access to two only: the number of iterations per timestep and the limiting accuracy.
Using either 486DX-66, PentiumC90 or even Pentium Pro200 with small or large number of monitored nodes had little effect on the performance but the size of the circuit had. Generally as a circuit was approaching the size of full screen, the simulator was struggling and would eventually give up complaining about too demanding iteration conditions (although I was already at the most relaxed end of the scale). Version5 (32 bit) performed a bit better. Like in a real laboratory, problems may be avoided by developing circuits gradually, starting at a minimum complexity and using idealised models. This is because convergence difficulties are sparked by the accumulation of demanding or critical nodes in a circuit, especially when their requirements are contradictory (like the size of iterations for integrating components vs sharply nonlinear ones).
Which version should be purchased for a tertiary environment?
Get a few engineering packs for projects, both v4.1 and 5 (for U/G and for more advanced users) and get a quantity of educational packs just to run an U/G virtual lab.
Let students buy their own cheaper versions so that they could do their work at home as well. And to let the staff generate tutorials/assignments get a few ETBs.
Is EWB useful only for Electronic Engineers?
Yes, in the case of version 4 but the new version 5 has general symbolic operators and may be used to model non-electronic systems. However the user needs to know the operation
|
