Tweet

Posted on 01 May 2019

Hysteretic Control Performance with no Output Ripple? Smooth!

Free Bodo's Power Magazines!

 

 

 

First implemented in buck regulators of the PowerWise product family

Whenever you start a new switch mode power design, you might ask yourself: What switch mode control scheme should I use? What are the advantages and disadvantages? Which one is the best for my needs?

By Werner Berns, Technical Support Manager and Michele Sclocchi, Principal Application Engineer, National Semiconductor Europe

 

Well, there is not just a simple answer like: Use control scheme 'A' and you are fine. The decision depends on your needs and your experience. The following article will give an overview about some of the most common control schemes such as PWM (Pulse Width Modulation) and Hysteretic Control as well as some of their different 'flavours'. At the end the article describes a new addition to the family of hysteretic controls that allows easier designs by maintaining excellent performance characteristics. But let’s have first a look at the PWM control.

PWM control

The most common control mode is the classic PWM control scheme. An internal clock leads the beginning of each duty cycle, which corresponds to the ON transition of the main MOSFET. The on-time is timed by the control voltage (Vc) compared with a saw-tooth ramp (Vp).

PWM buck regulator, basic architecture, e.g. LM3743

The saw-tooth ramp can be generated in three different methods, leading to voltage mode, voltage mode feed-forward, and current mode control scheme.

Voltage Mode Control

A constant saw-tooth ramp is internally generated with a fixed amplitude Vp. The loop gain and the loop bandwidth increases with the increase of the input voltage. A type 3 compensation is usually used to compensate a double pole given by the inductor and output capacitor, and its zero resulting from the output capacitor and its ESR. This makes the loop compensation a bit more complicated. Voltage Mode (VM) is widely used for applications where the input voltage is relatively stable.

Voltage Mode with Feed-Forward Control

With this methodology the slope of the sawtooth ramp changes with the input voltage removing the variability of the loop gain and bandwidth caused by a changed input voltage. Line transient respond is improved because the regulator changes the duty cycle before an error occurs at the output voltage. The other advantage associated with voltage mode feed forward scheme is that it allows the loop gain to be optimized over the entire input voltage range. An example for this implementation can be found in National’s LM5115.

Current Mode Control

Rather than using a constant sawtooth ramp to control the duty cycle, the current mode control uses the sawtooth ramp generated by the output inductor current (Figure 2).

Current mode buck regulator, basic architecture, e.g. LM201xx

A current sense amplifier detects the inductor current by measuring the current of the main MOSFET. A corrective ramp is added to avoid the problem of sub-harmonic oscillation for duty cycles larger than 50%.

In the current mode control scheme, the modulator, output switch and inductor operate like a transconductance amplifier, supplying a regulated current to the output. As a result, the gain in this stage is not affected by varying Vin, instead, the gain changes with the load resistance.

Current mode control offers several advantages, such as easy current sharing between power converters connected in parallel, better compensation due to the single pole of the system, precise cycle by cycle current limit, and immunity to input disturbance.

One of the main disadvantages of current mode control is the difficulty to measure the current at small duty cycles. This measurement can be quite susceptible to noise and the modulation can be erratic.

Emulated Current Mode (ECM) Control:

The emulated current mode control technique, patented by National Semiconductor, overcomes noise issues at low duty cycles and therefore makes it especially ideal for high Vin to Vout conversion ratios and/or at high switching frequencies. The current ramp is not directly measured as in the classical CM control, but built from two different portions. The first part is created by measuring the current in the low side MOSFET (or Schottky diode) and 'frozen' with a sampleand- hold element just before starting the next cycle. It basically represents the starting current of the main switch. The second part is an artificially created ramp where the slope varies with the difference of Vin and Vout. The resulting signal represents the inductor current during the ON phase but without the noise issues and blanking time issues. The ECM technology has been implemented for example in National’s latest Simple Switcher regulators (LM557x, LM2557x) and in the 100V buck controller LM5116. All are members of the company’s energy-efficient PowerWise® family.

Hysteretic control

Another possible solution is the hysteretic control scheme (figure 3). The modulator is simply a comparator with a few mV of input hysteresis that compares the feedback voltage with a reference voltage. If the feedback voltage exceeds the upper threshold, the comparator turns the switch off. The switch turns on again, once the feedback voltage falls below the lower threshold of the comparator.

Hysteretic buck regulator, basic architecture, e.g. LM3489

This topology reacts extremely fast to load and line transient, it is very simple and it does not require loop compensation.

The main issue associated with this control scheme is that the switching frequency is not set by an oscillator; it is not constant and dependent on many variables. The frequency varies very much on the variation of components’ parameters and operational conditions. Input voltage, load current, inductor value, output capacitor and especially its equivalent series resistor (ESR) can all have a huge impact on the switching frequency.

Overall this has some advantages and disadvantages. Positive arguments are the easy control loop. It is very easy to get such a controller stable.

The control loop is extremely fast with a delay response of less than 100ns. This results in extremely fast transient response. It is superior to any competing PWM regulator architecture.

The disadvantage of such a control scheme is mainly the very large range of operating switching frequency. The selection of the right components and corrective actions with respect to EMI might make it more difficult too.

Hysteretic Constant On Time (COT)

As mentioned above, the hysteretic scheme has some interesting advantages, with the only major problem of a 'more or less' unpredictive switching frequency. The introduction of a one-shot generator triggered by a comparator and the on-time being inversely proportional to the input voltage, the switching frequency remains now nearly constant.

In practice, load variations will vary the switching frequency around the centre point. This actually has a positive side effect with regard to EMI results, because the switching frequency will not create a single peak, so it becomes easier to fulfill EMI limits. Actually it behaves similar to a spread spectrum system.

There is one remaining thing that requires the designer’s attention: The comparator input (FB) does require a minimum ripple for stable operation. Thus, a minimum ESR is usually required at the output capacitor to generate a sufficient ripple voltage. Because of this, ceramic capacitors cannot be used without other measures such as integrating the switch-node voltage and then adding the resulting triangle waveform to the feedback input voltage (see National’s LM5010A datasheet, figure 11, page 15 for more details).

Besides this ripple issue, such a concept is easy to use and the transient response remains very fast. It actually combines many of the advantages of a PWM fixed frequency concept and a hysteretic mode concept into one solution. In discontinuous operation, the switching frequency decreases together with the load current. This keeps the conversion efficiency high at light load.

COT with Emulated Ripple Mode (ERM)

The emulated ripple mode is a new methodology, patented by National Semiconductor, which overcomes the need for an appropriate ripple at the feedback input. This allows the use of any capacitor type, including those with a low ESR, e.g. ceramics. Figure 4 shows the internal creation of the ripple (emulated ripple) which is of course still required for proper operation of the comparator, but because that creation is internally, you don’t need to worry about it anymore.

Internal creation of the ripple

This new technique combines ease-of-use, fast transient response and nearly constant switching frequency in one. It has been first implemented in the new LM310x synchronous buck regulators that are also members of National’s PowerWise product family. More products of this kind are planned to be announced within this year.

Conclusion

Table 1

The importance to understand functionality inside any controller is essential to benefit from its advantages associated with a particular application. Table 1 shows a summary of the discussed topics and control techniques and compares them without claiming for completeness. There are many other derivatives and combinations of techniques possible. ERM adds another 'flavour' to the picture and has its specific advantages as discussed above. National Semiconductor offers a wide range of solutions utilizing those mentioned and others for any application and power size, together with software simulation, technical support and training material that facilitate the selection and the design of your power supply.

 

 

VN:F [1.9.17_1161]
Rating: 0.0/6 (0 votes cast)

This post was written by:

- who has written 791 posts on PowerGuru - Power Electronics Information Portal.


Contact the author

Leave a Response

You must be logged in to post a comment.