Modular switching DC/DC converters (“regulators”) are all the rage. The integration of all the components required for a power supply into a single package saves space and simplifies circuit design (if at a little extra cost compared with a device built from scratch. See the TechZoneSM
article “The Inductor’s Role in Completing a Power Module-Based Solution
As the first TechZone article (“DC/DC Voltage Regulators: How to Choose Between Discrete and Modular Design
”) in this series discussed, inductors have proved tricky to squeeze into the package. It has been done by a few pioneering suppliers. However, if that was all there was to it, modules using a discrete inductor would be in rapid decline, yet sales remain healthy. That is because, in truth, integrating the inductor introduces a compromise.
This article builds on the previous submission with a description of this design trade-off, increased switching frequency allowing the use of a smaller inductor, but at a cost to efficiency, using example reference circuits from Texas Instruments
Upping the frequency
The inductor plays an important role in a switching regulator and its performance is largely determined by its size (see the previous article, “The Inductor’s Role in Completing a Power Module-Based Solution
,” for more detailed information). Unfortunately, inductors tend to be bulky, for example, of the order of 400 to 750 mm³ for a switching regulator used in a mobile device.
The trick the semiconductor vendors have used to shrink the inductor, such that it can be shoehorned into the module, is to increase the switching frequency. It turns out that the inductor value (inductance) required for a given switching regulator is inversely proportional to switching frequency for equal peak-to-peak ripple current.
Lower inductance means fewer loops and/or thinner wire for the coil, and a smaller core (area inside the coil), thus reducing the inductor’s volume¹. An additional bonus of a higher switching frequency is a reduction in size of the output capacitor.
Increasing the switching frequency compromises efficiency, and the trade-off between the conveniences of a module that includes an inductor and another that is more efficient but does not include an inductor, is one that requires careful consideration.
The effect on efficiency
The efficiency (in percent) of a switching regulator is determined as output power divided by input power x 100. Higher losses reduce the output power and, hence, compromise efficiency. Moreover, higher losses also result in greater thermal challenges. Table 1 shows where the power is dissipated in a modular regulator.
|FET driving loss
||Function of gate charge, drive voltage, frequency
|FET switching loss
||Function of VIN, IOUT, FET rise/fall time, frequency
||I² x RDS(on)
||Vf x IOUT x (1 – D)
||I² x DC resistance + AC core loss
||IRMS² x ESR
|IC loss [IQ]
||Datasheet specification for IQ for when the IC is operating
Table 1: Losses in a switching regulator. (Courtesy of Texas Instruments)
To demonstrate the effect on efficiency of increased switching frequency, consider a modular switching regulator reference design based on Texas Instruments’ TPS54160
. This device is a 2.5 MHz, 60 V, 1.5 A buck regulator. The chip uses an external inductor, but the example holds if the inductor is integrated into the module. The reference circuit is shown in Figure 1 (L1 is the inductor, C2 the output capacitor).
Figure 1: TI TPS54160 switching regulator reference schematic. (Courtesy of Texas Instruments)
Although there are losses due to the FET (which is integrated into the module) and the rest of the chip, they can be ignored here as the same chip is used in each example. The capacitor loss can also be ignored since ceramic capacitors with very low equivalent series resistance (ESR) have been selected for the reference design.
Therefore, the major influence on the losses, and hence efficiency, is the inductor. Table 2 shows the switching frequency, and capacitor and inductor values for the three test circuits.
Table 2: Switching frequency and component values for test circuits. (Courtesy of Texas Instruments)
|Switching frequency (kHz)
||L1 DC resistance (max) (mOhm)
The results of the efficiency tests are shown in Figure 2. This figure shows the efficiency decreasing as switching frequency is increased. The footprint of the inductor has decreased from 420 to 43.5 mm², but peak efficiency has also dropped, from 86.5 to 80 percent.
Figure 2: Efficiency of switching regulator shown in Figure 1 with different components and switching frequencies as detailed in Table 2. (Courtesy of Texas Instruments)
Incidentally, to improve efficiency at any frequency, look for a switching regulator with a low drain-to- source ON resistance (RDS(on)
), a gate charge, or a quiescent-current specification at full load; or search for capacitors and inductors with lower equivalent resistance².
Size and convenience vs. efficiency
Higher switching frequencies allow the use of smaller inductors, which can then be integrated into a module, but at a cost of efficiency.
, for example, offers its LTM4601
DC/DC µModule - integrating switching controller, MOSFETs, inductor, and all support components - in a 15 x 15 by 2.8 mm LGA package. Operating over an input voltage range of 4.5 to 20 V, the LTM4601 supports an output voltage range of 0.6 to 5 V.
The LTM4601 is designed to run at 850 kHz and has an efficiency of 93 percent with a 5 V input and a 2.5 V, 4 A output.
For comparison, consider the same company’s LTC3610
. This device is a modular buck regulator with a 4 to 24 V input, and a 0.6 to 24 V output. It requires an external inductor (and capacitor). Linear Technology allows the operating frequency to be set by the user, but typical operation is in the 550 kHz range.
At this frequency, the LTC3610 has an efficiency of 95 percent with a 5 V input and 2.5 V, 4 A output. That is not a huge difference compared to the LTM4601, but if efficiency is one of the critical features of the design, then that two percent could make all the difference.
It is a similar story with Intersil’s
switching regulator modules. The company’s ZL9117M
is a non-isolated point-of-load power module integrating a digital PWM controller, power MOSFETs, an inductor, and all the passive components required for a complete solution. It has an efficiency of 87.5 percent when operating at 571 kHz with a 5 V input and a 1.2 V, 4 A output.
In comparison, Intersil’s ISL95210
, a 5 V, 10 A buck regulator that requires an external inductor, offers 95 percent efficiency at 400 MHz with a 5 V input and a 1.2 V, 4 A output.
If the main influence on the switching regulator design is the space it takes up and the ease with which it can be added to the board, then a fully integrated module with the inductor included in the package is a good option.
But there is a price to pay, because the chip maker will likely have had to increase the switching frequency to shrink the inductor to fit the package, which increases the component’s power dissipation with a subsequent reduction in efficiency.
Through clever design and product optimization, the manufacturers have limited the impact of loss of efficiency to a few percent, but it may still be too much if high efficiency is critical to the design.
Fortunately suppliers offer a range of power modules, with or without an inductor, allowing the designer the flexibility of choosing either type, depending on the requirements of his or her product.
- “Effects of High Switching Frequency on Buck Regulators,” ON Semiconductor.
- “Challenges of designing high-frequency, high-input-voltage DC/DC converters,” Richard Nowakowski and Brian King, Texas Instruments Analog Applications Journal, 2Q 2011.