Study and Engineering Practice of Modeling IC Controllers for Switch Mode Power Supplies in SIMPLIS Environment Runxin Wang, Student Member, IEEE, Jinjun Liu, Member, IEEE, Pu Zhang, Junfeng Hou School of Electrical Engineering, Xi’an Jiaotong University, Xi’an, CHINA E-mail: [email protected] Abstract—This paper introduces some most recent results in a research project that aims at developing computer simulation models of integrated circuit (IC) controller chips inside switch mode power supplies. These models play a very important role to smooth operation of the electronic design verification test (eDVT) system. Based on engineering practice, the challenges to model power supply controllers are discussed and some strategies and methodology for such modeling activities in SIMPLIS environment are proposed in this paper. The way to model NCP1230A, an IC controller delivered by ON Semiconductor, is taken as an example to show the modeling process. Simulation waveforms of several established models and the related experimental results are given to verify the validity of the proposed strategies and methodology.

I. INTRODUCTION Accompanied by the increasing capabilities of software and improvements in hardware e.g. higher CPU speed, computer simulation is taking a more and more important role in switch mode power supply (SMPS) design process. Based on broad researches and discussions that have lasted for more than a decade, the ideas that integrate circuit design and test procedures in simulation and thus uncover most of the potential design deficiencies before the first prototyping, such as electronic design verification test (eDVT) system, are gradually realized in engineering and began to bring benefits for some power supply manufactures by saving time and money [1-3]. In eDVT procedure, where a power supply system is emulated in a “hierarchical” or “multilevel” way, the system is usually partitioned into device/function blocks, after which each block will be modeled individually. Each integrated circuit (IC) controller chip inside a modern power supply falls into one of such individual blocks naturally due to its powerful control functions. Modeling such complicated IC chips not only play a very important role but also occupies a big proportion of efforts in modeling the whole system. Based on concrete work and from viewpoint of practical engineering, this paper introduces some most recent results in a research that aims at developing computer simulation models of IC controllers for an eDVT system. After showing some brief background information regarding to the eDVT system in

Section 2, the challenges in this system to model SMPS IC controllers are illustrated and some strategies and methodology for IC modeling are proposed in Section 3. In Section 4, the IC controller NCP1230A is taken as an example to show the modeling process. Some key experimental and simulation waveforms from other typical IC chips are given in Section 5, in order to show the effective of the established models and verify the validity of the proposed strategies and methodology. Conclusions are listed in Section 6. II. THE INNOVATIVE TOOL OF EDVT SYSTEM AND SIMPLIS The eDVT system is an innovative engineering tool for power supply manufactures that is capable of using automated circuit simulation to emulate the complete final test performance during a production power supply design. This web-based system facilitates the efficient sharing and reuse of analog design IP (intellectual property) blocks within and among design sites even they spread geographically over different continents [2, 3]. Fig. 1 and Fig. 2 show the conventional and eDVT-based product development procedure respectively [1]. Compared with the conventional product development procedure, where any errors found in the electrical prototype test are fed back to the designer who would then have to go thorough another design iteration, the eDVT-based product development procedure builds a virtual prototype firstly by using a simulation tool with adequate models. By running a set of simulation tests in this earlier stage, most of the errors found in a later stage in the conventional product development procedure can be detected before electrically prototyping. A more detailed eDVT procedure is shown in Fig. 3 [1]. SIMPLIS, a PC-based software package that is specially developed to analyzing switching systems, is used as simulation engine of the eDVT system. It typically characterizes a switch mode power supply into a cyclical sequence of linear circuits and this makes simulations converge much quickly by taking advantage of the repetitive piecewise linear structure of such systems. Due to this feature, SIMPLIS has more competition capability in industry applications compared to other simulation software packages [4].

This work was supported by ASTEC Custom Power (HK) Ltd. and Transim Technology Corporation.

0-7803-9547-6/06/$20.00 ©2006 IEEE.

1216

Fig. 1. Conventional product development procedure.

Fig. 2. eDVT-based product development procedure.

Fig. 3. eDVT procedure.

III. CHALLENGES AND PRACTICAL STRATEGIES IN IC CONTROLLER MODELING As shown in Fig. 3, adequate simulation models for device/function blocks or a reliable source that is capable of providing such models is very important. Compared with

discrete components such as diodes and transistors, there exist more difficulties to get proper models of IC controllers. Each IC manufacturer explains the function of their products in their own way. Not every vender supplies simulation models to endusers, and even they did so the ready-made models, which were doomed not for generic usage in any simulation environments, may be incompatible with a specific simulation engine such as SIMPLIS. In this case, to make the simulation system run smoothly, it seems that one has no way to choose except developing models by him- or herself. But is this feasible? If feasible, what the models will base on and how to implement such models in SIMPLIS environment? Being a user rather than the designer of the IC chips and then, generally speaking, with less enough knowledge and information about the internal circuits inside ICs at semiconductor level, one can not develop a model that represent all detailed physical properties inside the chip. However, just as mentioned before, eDVT is aimed to emulate a power supply system in circuit level and find the design error so that one can make the design modification. It is more concerned with circuit behavior rather than physical performance inside IC, and only in this way could a simulation be finished in an acceptable time range. Furthermore, SIMPLIS tries to achieve a tradeoff between accuracy and simulation speed by adopting piecewise linear models for each component. Usually this type of models must be topologically simple comparing to those based on physical properties. Therefore, lack of detailed components information inside ICs such as parasitic parameters is not a big problem. Even if provided, they will be ignored safely. What is really necessary is a very clear and comprehensive description to the function and features of IC chips, which can define their behavior precisely. Datasheets, sometimes accompanied by application notes from the venders, are usually regarded as the main source of such information. With these authorized data, the target of the modeling activity becomes to develop a simplest behavioral model of the IC controller in SIMPLIS environment, partly based on one’s knowledge and experience. The established model should represent features that are concerned by SMPS designers as much as possible. Schematics can be organized into multiple levels in a hierarchy in SIMPLIS environment. Like that in the eDVT procedure, as an optional strategy, one can also partition the IC internal circuits into function blocks and model and test each block individually. This could greatly facilitate the modeling and trouble shooting process. Like other modeling process, the established IC model should be tested by experiment. Summarizing all discussions above, we can sketch out strategies and basic steps to model IC controllers in SIMPLIS environment as following: 1) Understanding the datasheet thoroughly and partitioning internal circuit inside IC into function blocks, which can match schematic blocks in SIMPLIS environment as close as possible;

1217

2) Modeling each function block one by one and connecting them after making sure every blocks work well by testing them individually; 3) Testing the whole IC model in simulation environment; 4) Testing the IC model by comparing the simulation and experimental results. IV. MODELING EXAMPLE OF CHIP NCP1230A This paper takes the chip of NCP1230A, a low-standby power soft skip mode controller delivered by ON Semiconductor in 2004, as an example to show the proposed

modeling process. Besides all functions of a current mode PWM controller with internal ramp compensation, NCP1230A distinguish from conventional controller chips by having an event management scheme, which disable the front-end PFC circuit during standby, and offering another feature that it goes into soft skip mode with limited peak current when the power supply is at light loads. Fig. 4 shows the internal circuit architecture from NCP1230 datasheet. According to the function each component undertakes, the internal circuit inside NCP1230A are roughly divided into eight function blocks i.e. oscillator and PWM Event Management

Ctrl Signals

PFC Ctrl

Time Delay

Driver Feedback

Oscillator& PWM Logic Soft Start

Fig. 4. Internal circuit architecture of NCP1230A.

Switch PF C Vcc

Event Management VC C HV

11.7

VC C

R2

C urrentLim iter

Soft Start

10Meg R1

U1

U nit1 VC C ok VC C

U nit4 VC C ok SSram p

HV

OVP

Skip

BlnkdC S

R TN

Stby F lag

SSram p

R TN

R TN

D riv e U nit2 VC C

U3

Vdd

CS

CS

U nit6 PW Mout

STOP

R TN

Setpoint

SSram p R TN

U4

U2 T clk

QN

FB

PFC Ctrl

U nit3 ErrorF lag

U nit5 ErrorFError125m lag s

U nit7 Error125mPF s C on

Sk ip

Stby FStby lag 125m s

Stby 125m s

R TN

Leav Stby SSram p R TN

Setpoint

100Meg R3

Driver

Oscillator& PWM Logic

Q

Feedback& Ctrl Signals Vdd

FB

R TN

R TN

U5

Voltage Reference R TN

DR V Lim iter

BlnkdC S

GateC trl

F lipF lop S

Q

R

QN

ErrorF lag

R TN

VC C ok GateC trl

Time Delay

Fig. 5. Top-level schematic of the established SIMPLIS model for NCP1230A.

1218

PF C Vcc

PF C Vcc

HV

HV

FB

FB

CS

CS

VC C

VC C

R TN

DR V

D RV

GN D

logic, event management, soft start, feedback circuit, time delay circuit, driver circuit, PFC control and so on. Fig. 5 shows the top-level SIMPLIS schematic model that is organized into multi-levels in a hierarchy structure. It contains several blocks, each of which represents a sub-circuit realizing relatively independent functions within the controller chip. The blocks are modeled and tested individually, only after which they are connected to make up the final entire chip model. The labels in both Fig. 4 and Fig. 5 are used to indicate the function division process and attentions should be paid to that Fig. 5 only shows the top-level schematic, in which each block is descendible if being clicked in the software environment. Fig. 6 to Fig. 8 illustrates some experimental results and the

corresponding simulation waveforms based on the established chip model. Fig. 6 shows the waveforms during a single-chip test on the NCP1230A. Fig. 7 and Fig. 8 show the steady-state and transient responses respectively, which are obtained from a 120 W, 16 V dc output universal ac input adapter power supply that is currently under design and employs the controller NCP1230A. Above comparison between simulation and experimental results clearly showed that the established simulation model is able to represent the behavioral property of the IC chip although they are not strictly same at the component level. The simulation also has an acceptable speed that make it be suitable to eDVT, where a very complicated model that needs

Experiment waveforms

Simulation waveforms

Two startup sequences

Two startup sequences

Fig. 6. Single-chip test waveforms for testing the dynamic self-supply feature.

Vin=180Vdc, load=2.5A Experiment waveforms

Simulation waveforms (POP)

T=11uS

T=10uS

Fig. 7. Steady-state response (POP) of a power supply under design.

Vin=180Vdc, load=0.05A Experiment waveforms

Simulation waveforms (POP)

Fig. 8. Steady-state response (POP) of a power supply under design at light load.

1219

been submitted to industry application successfully.

a long time to simulate a simple issue seems less valuable. V. EXPERIMENTAL VERIFICATIONS OF OTHER TYPICAL CHIPS

ACKNOWLEDGEMENT

Some more simulation and experiment results are given in this section, in order to show some newest example and verify feasibility of the proposed methodology of behaviorally modeling SMPS IC controllers again. Fig. 9 shows the steady-state response of the UCC3817 Demo board, where UCC3817, delivered by TI in 2004, is a conventional PFC controller that employs a multiplier in its circuit. Fig. 10 shows the startup process of the Demo board for IR1150S, which is another PFC controller but based on onecycle control theory and has a totally different operation principle compared to UCC3817. International Rectifier delivered IR1150S in 2005.

Authors thank Dr. Thomas G. Wilson, Jr. and Dr. Ronald C. Wong in Transim Technology Corporation, Mr. Laurence McGarry and Dr. William Lee in ASTEC, for their guidance to work in the Power Electronics Simulation Modeling Center. Authors thank Mr. Gavin Gan in ASTEC for his great help and significant contribution to Fig. 6, Fig. 7 and Fig. 8 in this paper. REFERENCES [1]

[2]

VI. CONCLUSION

[3]

Based on practical work in engineering, this paper introduced strategies and methodology of modeling IC controllers in the SIMPLIS environment for an eDVT system. The experimental and simulation results verified that this methodology is effective and actually dozens of models have

[4]

Q.M. Li, F.C. Lee and T.G. Wilson, “Design verification and testing of power supply system by using virtual prototype,” IEEE Trans. Power Electron. , vol. 18, pp. 733-739, May 2003. L. McGarry, K.Y. Qu, W. Lee and T.G. Wilson, “A virtual prototyping process for power supplies using an electronic design verification testing (eDVT) system,” in Proc. IPEMC’04 Conf., vol. 3, August 2004, pp. 1689-1694. L. McGarry, K.Y. Qu, W. Lee and T.G. Wilson, “Quality assurance verification using a virtual prototyping system,” in Proc. IEEE APEC’05 Conf., vol. 2, March 2005, pp. 1325-1331. Transim Technology Corporation, SIMPLIS Reference Manual, UK: Cantena Software Ltd, 2003.

Demo Board of UCC3817: Vin=220V 50Hz, load=1.25kohm Experiment waveforms

Simulation waveforms (POP) 1

Iin / A

0.8

0.6

0.4

0.2

0 300

250

Vin / V

200 150 100 50 0

0.9

0.92

0.94

0.96

0.98

1

time/Secs

20mSec s/div

T=10mS

T=10mS

Fig. 9. Steady-state response (POP) of UCC3817 Demo board circuit.

Demo Board of IR1150S: Vin=220V 50Hz, load=1.2kohm Experiment waveforms

Simulation waveforms

400 350 300

Vo / V

250 200 150 100 50 00

100

200

time/mSecs

Fig. 10. Startup process of IR1150S Demo board circuit.

1220

300

400

500

600 100mSecs/div