Do we need another SPICE?


.....YES!!! if you are negatively strucked with components and/or node limitation of commonly available Evaluation Versions for Windows environments..

Surfing on the web I have found a wonderful tool coming from   Linear Technology Corporation.

The name of this free tool is SwitcherCAD III.

From the on-line guide the main highlights: 

Presentation:
SwitcherCAD III is a new SPICE that was developed for modeling board level switching regulator systems. Incorporated into the new SPICE are circuit elements to model practical board level components. Capacitors and inductors can be modeled with series resistance and other parasitic aspects of their behavior without using sub-circuits or internal nodes. Also, a simulation circuit element was developed for power MOSFET's that accurately exhibits their usual gate charge behavior without using sub-circuits or internal nodes. Reducing the number of nodes the simulator needs to solve significantly reduces the computation required for a given simulation without compromising the accuracy or detail of the switching waveforms. Another benefit of these new simulation devices is that convergence problems are easier to avoid since they, like the board level component the model, have finite impedance at all frequencies.
Fun advices:

SwitcherCAD III is designed to be used by three different types of design engineers: those who know what they're doing, those who think they know, and those who are sure they know absolutely nothing about switching regulator design. The experienced designer needs a "what if" program that allows him to quickly alter aspects of a circuit to find an optimum design. The neophyte needs a cookbook approach that yields a reliable design based on the simplest of inputs. The "loose cannon" designer needs a program that will allow him to exercise his free will, but will be intelligent enough to alert him to fatal design flaws.
 Wisdom:
 Please be aware, however, that SwitcherCAD III is not intended as a total solution. It is only a tool to ease the design procedure, which must also include breadboarding and testing. Use common sense with the results obtained from simulation.
Hardware Requirements:
LTspice/SwitcherCAD III runs on PC's running Windows 95, 98, 2000, NT4.0, Me, or XP. It doesn't work under Windows 3.1 or DOS. Since a simulation can generate many megabytes of data in a few minutes, free hard disk space (>200MB) and large amount of RAM (>128MB) are highly recommended. Basically, the program can run on any PC with Windows 95 or above, but the simulation may not finish if there is not enough hard disk space.
LTspice/SwitcherCAD III will also run on Linux. The program has been tested on Linux RedHat 8.0 with WINE version 20030219.

I have spend a relaxing day to explorate the potentialities of this spice simulator by simulating a fully balanced (and at moment virtual) audio preamplifier based on the supertriode concept developped during 2K days. For this analysis I have included  into the simulation flow some circuities often keepen out for the above mentioned limitations of available demos. Further I have explorated also the possibility about hierarchycal design offered by the tool.

Fig.1

Fig.1 is the hierarchycal  view of the preamp. The blocks include the  preamp, thr bias correction, the shunt regs, the reference regs as shown in Fig 2,3,4 and 5 respectively and a  model for the signal cable. 

Fig.2

 

Fig.3

The preamp unit is build around my vt25920 signal supertriode. The supertriode is a new device characterized conceptually but not fully developped in a practical way although prototypes of it has been already realized. This preamp unit is fully balanced and takes adavantage by the use of my SOT1 magnetic-unit in mu-metal core specifically conceived for use in extreme fidelity audio systems. The use of a high perveance core requires total care in the bias currents othervise non linearity are present during reproduction. The first stage is a cathode follower with a plate  voltage of 6 volts. The output stage (a classical push-pull unit) runs at 16Vpp. The secondary of first stage's output transformer is used to inject a correction dc voltage able to equalize the bias plate voltages in the output stage. This correction is obtained by a reference voltage of 0.6V, Fig.5,  and performed  by the circuit in Fig.2.

Fig.4

Plate voltages are regulated by shunt regs. The shunt used mimic the behaviour of a TL431 unit but the pass element is build with a supertriode device. I will detail this unit in my Device Section later.

Fig.5

At this point I have explorated both transient and AC capabilities of  SWCAD III.  The deck in Fig.1 although not complex creates a quick failure in the PSpice evaluation enviroment for reached limits of components and nodes.

Fig. 6 is a screenshot of a transient analysis showing thr output voltage with  its FFT and the LOG file with a summary of  the evaluated THD.

Fig.6

On the same deck, the Fig.7 shown the result of an AC analysis. The simulation captures very well the high  frequency resonance peak, typical for a mu-metal core, at  70kHz, while Fig.8 is the circuit response on a 70kHz square-wave (response close to the resonance peak) .

Fig.7

 

Fig.8

 

Fig.9

The schematic in Fig is used to simulate the power-up conditions about plate voltages, vcc and vref. Current generator emulates the current load of attached blocks such as the preamp and bias correction units. You can use also a more complex deck having blocks in Fig. 3, 4 and 5 but on a 450MHz-Pentium II-512Mb Ram, simulation speed is about 80..10ms/s. Plate voltages are obtained starting from a choke input rectifier obtained from a Wheastone's bridge build with superdiodes, Fig. 10. In order to speed-up simulation internal nodes must be initialized to zero.

Fig.10

Results of transient analysis are shown in Fig. 12.

Fig.11

 

Fig. 12

 

Fig.13

Fig.13 evaluates the behaviour of heater power supplies.

Fig. 14

 

Fig.15

 

Fig.16

Fig. 16 depicts  a parameterized transient analysis based on variation of the the base's capacitors of the Fig. 15. Resistors involved in the arrangment of the LM317 can be tuned to obtain an exact 6.3V. The LM317 is not macro-modelled but modelled with a transistor level description as shown in Table 1, so the simple deck in Fig. 13 contains 116 bjt.

At last, Fig.17 is a mixed-signal deck used to simulate the behaviour of relay units. Involed circuitries are LM7805, NE555, diode-bridge, inverters, switches and  relay macromodels.

Fig.17

 

Fig.18

 

Fig.19

 The summary Table shown in the following link contains all my models used for my audio evaluations.

MODELS

Free availability is welcome with the exception of  Supertriodes, Superdiodes and Supertriode-Based Shunts.