Pulse-skipping: reasons, effects and relevance for automotive applications – Part 1
Switch-mode power supplies (SMPS) mostly operate at a fixed frequency, using pulse-width-modulation to regulate the output. Under certain circumstances, they may enter pulse-skipping modes of operation.
This article discusses under which circumstances pulse-skipping mode may occur, details what implications this involves and if it is of concern in normal operation. Furthermore, solutions to avoid pulse-skipping mode are introduced.
There are three main scenarios that could lead to pulse-skipping:
- Low power mode (LPM): if LPM is enabled, the device will not switch continuously at low load conditions, but skip pulses to reduce the switching losses.
- Another occasion is often referred to as “Drop-Out” mode. If the input-voltage is close to the desired output-voltage, the part attempts 100% duty-cycle on the high-side-FET. However, in case of boot-strap-supply, a recharge of the boot-strap-capacitor is required to drive it.
- A further reason for pulse-skipping, not further covered herein, is driven by minimum on-time requirements: Any switch-mode-power-supply has a minimum-on-time of the FETs or gate-drivers. In particular with high input and low output voltages at high switching frequencies, this could lead to pulse-skipping.
In the following, only scenarios 1 and 2 are discussed
There are various ways to implement pulse-skipping into an IC:
- Hysteretic-mode pulse-skipping leaves out pulses that are not required, and will initiate a single pulse once the output voltage has fallen below a threshold. This allows for a relatively low output ripple.
- Another implementation is Burst-mode, which, as the name says, send a burst of pulses if the output-voltage drops below a threshold. This tends to introduce a higher output ripple.
- Further options include constant on-time or constant off-time, both coming with the disadvantage that even during normal operation the switching frequency varies with load and is difficult to compensate and filter. Those variants will not be discussed in more detail in this report.
As an example for hysteretic mode let’s use the TPS43340-Q1 multi-rail supply.
Figure1 shows the output voltage (AC-coupled) of a Pulse-Skipping device, TPS43340-Q1, (Channel1, yellow trace), and the corresponding switch-node voltage (CH2, red trace). The part is supplied with 14 V, and set to switch at 400 kHz. The output voltage is 5 V:
600kHz, VIN=14V, Vout=5V, Iout=10mA, LPM enabled
Figure 1: Pulse-Skipping in Low-Power Mode CH1=Vout (AC), CH2=PhaseNode, bottom Graph shows a Zoom into the pulse
Hysteretic pulse-skipping will generate a single pulse (red trace), where the cycle-time depends on the load, while the pulse-duration depends on the selected switching frequency (at lower frequencies the pulse would be prolonged). For some combinations of switching frequency and load-conditions, two or more consecutive pulses may be required to reach the upper voltage threshold. In LPM, the draining of the output capacitors is slow resulting in a small output voltage ripple, which is not a concern in usual applications. For TPS43340-Q1, it’s about 30 mV.
Most parts, including TPS43340-Q1, do not maintain the preselected frequency (or a fraction of it) in case of low power mode, as here the duration between the pulses changes from micro-second-range to milliseconds or even seconds, depending on load (in the example above its about 150 us or 6.7 kHz). The unpredictable frequency (or repetition of the pulses in case of burst-mode) could be a concern but with the low load and appropriate layout, it generally is not.
In case it is of concern, most parts feature a selection if LPM is supported. Figure 2 shows the same scenario as above, however, the device is forced into continuous mode. It is switching at the selected 600 kHz, the ripple is minimized (>20 mVpp), however at the expense of higher switching losses.
600kHz, VIN=14V, Vout=5V, Iout=10mA, forced continuous mode (LPM prohibited)
Figure 2: Forced Continuous Mode at low output power, CH1=Vout (AC), CH2=PhaseNode, bottom Graph shows a Zoom into the pulse
Switch-mode converters or controllers supporting Drop-Out mode or Tracking will attempt a 100% duty-cycle in case the input voltage approaches the desired output voltage. To drive the high-side FET, an internal supply is required, and depending on the architecture, a recharge may be required: If a charge-pump or another external supply is used, a “true” 100% duty-cycle can be achieved. More commonly, a bootstrap-architecture is used and the boot-strap capacitor needs to be recharged by momentarily toggle the FETs, hence turning on the low-side briefly. Obviously, in doing so every few pulses, those in between are skipped. Here again, different approaches exist: e.g. for TPS43340-Q1, every 4th pulse is used to recharge the boot-strap, below example shows for an input-voltage of 5.1 V an output voltage of 4.96 V at 1 A load. The original switching frequency of 600 kHz is reduced to 150 kHz:
600kHz, VIN=5.10V, Vout=4.96V, Iout=1A
Figure 3: Pulse-Skipping during Drop-out-mode, CH3=VIN, CH1=Vout, CH2=PhaseNode
In general, here the reduction in output voltage, due to the RDSon of the FETs and the missing supply during the recharge is of higher concern than the induced sub-harmonics: The TPS43340-Q1 is a buck-controller with external FETs, here one with about 60 mOhm RDSon.
Below is a summary of parameters that effect LPM-Entry and their potential to help avoiding it:
– In most cases, devices have a mode-selection to allow for or prohibit LPM. If the device is forced into continuous mode, switching current losses increase, but it operates at a fixed frequency and avoids pulse-skipping. Devices driven by an external clock are predominantly operating in forced continuous mode as well.
– Output load: defined by to-be-supplied systems. The entry-threshold may be a pure current sensing or more sophisticated, like in TPS43340-Q1, detecting zero-inductor-current for a predefined period per cycle (here: 60%). In LPM, the ripple does not increase significantly or can be buffered with appropriate output capacitors, so no action might be required.
– Again, in drop-out-mode the lower output voltage is of higher concern than the change in switching frequency, in particular since most parts offer a predefined binary fraction of the set switching frequency, allowing for easy filtering.
LPM is either of no concern or can be disabled. Drop-out-mode helps to maintain regulation at insufficient supply-voltages, with limitations, but better than causing under-voltage-lock-out.
About the author
Frank Dehmelt is Application Engineer, Mixed Signal Automotive at Texas Instruments.