A short summary of the important features of current-mode control, and links to the definitive free book on the topic.
38 Years of Current-Mode Control
20016 marks the 38th anniversary of the publication of a landmark paper on current-mode control. Since then, current-mode has become the industry standard for controlling switching power supplies. To mark the occasion, we are making available, for the first time ever, a free downloadable copy of the definitive book on current-mode control.
In 1978, Cecil Deisch published a paper that introduced current-mode control in a practical and easy-to-understand circuit. While this was not the first publication on current-mode control, it was the one that popularized it in industry. Up until that time, power supply designers had used voltage-mode control for switching power supplies. In voltage mode control, a sawtooth ramp is compared to a voltage from an error amplifier to set the duty cycle of the power switch of the converter. In this way, the output voltage and inductor currentare indirectly controlled.
Current-mode control directly controls the peak of the switch (or inductor) current, and as soon as it was introduced, it became widely accepted as a preferred method of control. It offered the following advantages:
1. Peak current limiting on a cycle-by-cycle basis;
2. Damping of the LC-filter characteristic;
3. Ability to parallel multiple converters together to provide higher power, (critical for the multiphase power supplies used for microprocessors);
4. Increase of phase of the control-to-output transfer function;
5. At least a ten times reduction of input-to-output noise transmission;
6. Simplified design of the feedback compensation.
It took several years before Unitrode came out with the first current-mode control chip, but once this happened, there was no turning back for the industry. However, along with all of the advantages of current-mode control came some drawbacks:
1. The signal-to-noise ratio of the control modulator is not as good as voltage mode. Great care must be taken with the current sensing implementation;
2. The feedback current loop, although an intuitive control scheme, has an inherent instability as a 50% duty cycle is approached. This became known as subharmonic oscillation since it occurs at one-half the switching frequency;
3. Existing analysis methods were unable to properly explain the operation of the current-mode system.
First, let’s address the signal-to-noise issue. It is your job as a designer to make sure the current is sensed cleanly and accurately. For almost all power supplies, the best way to do this is with a current transformer for lowest dissipation, or with a sense resistor at lower power levels. While clean current sensing requires some work, the advantages of current mode greatly outweigh the downsides, and it should always be the first choice for your power supply.
Secondly, the subharmonic instability in the current loop was quickly discovered, and the fix was found to be adding a sawtooth ramp back into the system – or a little bit of voltage mode control. How to add the compensating ramp is another question. Most methods suggested by control chip application notes do not work well, and it is recommended that you read the paper at the end of [2] for the best way to implement this.
That leaves us with the modeling issue. For many years, from 1978 until 1989, there was no clear-cut method of modeling the current loop that explained all of the behaviors observed. Simple models that treated the inductor as a current source could not explain the instability. More complex models that recognized that the current loop gain was finite also failed to properly explain the instability, and they also varied greatly from one to another in terms of system parameters. Researchers could not even agree on a methodology for measuring and designing the system properly.
The most complex models of all, using discrete-time analysis instead of averaging techniques, gave accurate predictions for the current-mode instability, but failed to result in a usable model for the design engineer.
Finally, in 1990, all of the different approaches were compared and resolved into a model that accurately predicted the current loop instability, used average models for the part of the circuit where it applied, and sampled-data for the current loop phenomena. A simple approximation then revealed a new model that was very simple to use.
Since this work was done, there has been nothing published to improve on the model, and the results are used widely in the industry and in research. The new current model offered several insights that helped understand the system properly:
1. The best representation of a 2nd order converters is given by a third order transfer function. A dominant pole represents the current-source effect, and a double pole at half the switching frequency shows the subharmonic oscillation.
2. Current mode control can go unstable even at duty cycles below 50%, and a compensating ramp must be added even for some converters that are limited to 50% duty cycle.
3. The current feedback loop has two right-half-plane zeros in the transfer function that lead to instability.
4. The PWM switch model, developed by Vorpérian, works perfectly well in the current-mode model. This allows a single model for current-mode, voltage-mode, and the important case where significant compensating ramp is used and the resulting system is somewhere in between the two.
Full details of the model can be downloaded in the current-mode book which is now available from www.ridleyengineering.com. This 200 page book covers the history of current-mode control and modeling, detailed analysis, practical application examples, and PSpice model listings. An abbreviated summary at the end gives quick results for helping you with design when you don’t have the time to read the full story.
We hope that you find this book useful. Please feel free to email any comments to
[1] C.W. Deisch, “Switching Control method Changes Power Converter into a Current Source”, IEEE Power Electronics Specialists Conference, 1978 Record, pp.300-306.
[2] “A New Small-Signal Model for Current-Mode Control” Raymond B. Ridley, Virginia Polytechnic Institute and State University PhD dissertation, 1990, available free from www.ridleyengineering.com