Shrinking footprints in the landscape of switching regulators

Apr 6, 2020
Shrinking footprints in the landscape of switching regulators Blog Post Image
Switching regulator circuits have been around for many years and users have always had a choice between designing their own using discrete components or buying a finished module. Today, the technology required to meet the latest efficiency, EMI and power density expectations tips the balance in favor of the modular approach.

Table of contents

Non-isolated switching regulators have long been the workhorse of efficient conversion of DC power rails to lower or higher voltages, either directly for the load or as part of a distributed power architecture. The first designs from the 1950s used vacuum tubes and showed dramatic improvements in conversion efficiency compared with the alternative ‘linear regulator’ approach and also opened up the possibility of boosting DC voltages, only practical previously with unwieldy mechanical ‘vibrators’. It was only in the 1970s that the first switched-mode power supply IC controller appeared, the Silicon General SG1524, using ‘voltage mode’ control. The success of this device opened the floodgates to alternatives using different control and conversion techniques. As the decades rolled by, bipolar transistor versions came and went, replaced almost universally by MOSFETs; diodes also gave way to synchronous rectifiers, again using MOSFETs, and now even Si-FETs are threatened by wide band-gap materials such as SiC and GaN.

A measure of the development of the switching regulator is its conversion efficiency - over the years, the figures have been steadily climbing through 80%+ to 97% and higher in latest designs. Higher efficiency allows a higher power density, measured in watts/cm3, representing how much power can be delivered from a given volume of components in the design. Claiming ever higher power density has shown some ‘creativity’ in data sheets: for example, figures for some IC regulators have been advertised without all of the necessary external components, especially the bulky inductor and capacitors. Cooling is also often an issue, with spectacular power density only achievable with unrealistic air flow rates or overly-complicated water cooling. Also important is the ambient operating temperature range, not just the heatsink temperature - if the part has to derate heavily above a certain room temperature, that directly reduces the useful power.

Evolution towards state-of-the-art of switching regulators

Fig. 1: Product introduction delay means lost revenue
The history of the development of non-isolated switching regulators is one of continuous component integration with increased efficiency and functionality. This is with a background of decreasing output voltage as load requirements have dropped from 5V to 3.3V to sub-1V levels seen now. Input voltages have increased as well as system power levels increase, necessitating higher bus voltages with consequent lower current draw. Discrete component designs were simplified with IC controllers that integrated the switching transistors and more recently, the magnetics as well. Peripheral functions such as fault monitoring, current sharing, synchronisation and sequencing have been increasingly swept into the IC design.

Even from the early days of switching regulator designs, complete packaged converter modules have been available from vendors, offering proven solutions and saving customers the effort and risk of designing their own. This was sometimes a difficult sell, though, with more experienced engineers unwilling to pay a premium for something that they could design themselves. Even the development time and risk incurred making an in-house design could be tolerated for a product that sells for years that would pay for the R&D costs several times over. There was also an element of pride in being able to design a switching power supply from scratch.

Today’s landscape is different. Power design expertise is being lost to the OEMs and the techniques required to get the optimum performance can be very specialised, involving processes that even an OEM might not have available such as moulding of ferrite materials. Product life cycles are also much shorter meaning that development costs and delays due to design optimisation or repeated EMC testing have a greater impact on the ROI (Figure 1).
Fig. 2: Buck converter load steps cause voltage transients
Of course, control IC manufacturers do provide extensive application information making designs seem easy but these simplified design tools cannot predict actual circuit requirements. For example, suggested output capacitance is often far too low for real-life dynamic loads which may swing by a factor of one million between active and sleep states producing unacceptable voltage jumps (Figure 2).

Inductors are also often ‘glossed over’ in application notes with suggested parts chosen for best performance rather than price and practicality. In reality, the selection of an optimum inductor can take many weeks of evaluation of performance over temperature, frequency and load current, static and dynamic. Other parameters such as inductor saturation characteristics and leakage field can be critical in designs; EMC performance of a completed design is a ‘great unknown’ until the circuit is laid out on the end-PCB and final component choices have been made, when changes are costly to make. The story is similar with capacitors, with the optimum parts for performance and cost complex to evaluate with vital information such as self-inductance often not adequately documented in the datasheets.

State-of-the-art switching regulator designs achieve their high power density with control ICs that are often BGA packages that might be just 2mm x 2mm with a matrix of pads on a tiny 0.4mm pitch. This may not be compatible with a user’s PCB assembly process, requiring precision solder paste application with expensive X-ray imaging for inspection to check for shorts or bad joints. Similarly, a converter control IC may need a complex multilayer PCB with filled and buried vias leading to ground planes to efficiently spread the heat out of the package into the circuit board. If the user does not need this board complexity for other circuitry, a penalty is paid in the PCB manufacturing cost.

Latest switching regulators are versatile

Fig. 3: A 6A switching regulator in a 12.19mm x 12.19mm x 3.75mm package (RECOM RPM series)
Perhaps it could also be argued that power modules have to be general purpose products and may not be the best solution for just one application, but the latest generations achieve high performance over a wide range of operating conditions. The RPMH series from RECOM for example, is a 0.5A part with an input range up to 65V and an output trimmable between 2.5V and 28V. All this in a 12.19mm x 12.19mm x 3.75mm EMI-shielded package that can operate up to 105°C without forced air cooling (Figure 3). At higher currents, up to 3A. The company’s RPMB series in the same package operates up to 36V input with its output trimmable from 1V to 24V. A 6A part is also available, the RPM series, with a lower maximum input, again in the same package size.

Because of the extreme integration in these types of modules, there is little manufacturing cost increment to add a comprehensive range of control and monitoring features such as on/off, remote sensing, remote trimming, soft start, power-good signalling and power sequencing. The modules will routinely feature fault protection against under input under-voltage, short circuits, over-currents and over-temperature. It is understood that applications may operate from almost zero current draw to rated maximum under different sleep or fully loaded conditions so modules will often incorporate features to minimise light load power consumption while maximising efficiency, with techniques such as phase shedding in multi-phase converter topologies.
Fig. 4: A 2.5A switching regulator in a 4.5 x 4 x 2mm QFN package (RECOM RPX series)
For the highest power density, modules will use advanced production techniques such as flip-chip on leadframe technology with over-moulding. The RECOM RPX product (Figure 4) is an example, achieving a 2.5A rating in a 4.5mm x 4mm x 2mm QFN package with some parts in the range rated up to 95°C ambient at full load without forced air. The RPX series is an example of a part that requires external capacitors to function to specification. This can actually aid overall power density as, for example, input capacitors may already be present on the supply and output capacitors can be chosen for the exact voltage rating required. If these were internal, they would need to be oversized, rated at the maximum possible trimmed-up voltage.

On-board regulators often need to be able to boost as well as drop voltage. A typical application is in battery-operated equipment where power needs to be maintained for as long as possible as the battery discharges. For a positive output voltage from a positive input, the traditional solution has been a SEPIC, ZETA or Cuk converter, all needing two magnetic elements and with complex control loops. With their degree of integration, modular converters can implement a different topology at low cost such as the four-switch buck-boost which is effectively an array of MOSFETs which can be configured ‘on-the-fly’ as switches or diodes to seamlessly switch between buck and boost modes. The RECOM RBB series is an example featuring up to 3A rating in an LGA package and even a 3kW part in a half-brick format with 9V to 60V input and 0V to 60V output. Typical applications for this part would be 48V to 24V or 12V to 24V battery power conversion, electric vehicles, battery voltage stabilisers or high-power laboratory DC power supplies.

The module versus ‘DIY’ decision is now easier

Sourcing a switching regulator module which incorporates all of the processes and design techniques for optimum performance can therefore save a user time and money while de-risking product development. There is also the saving in supplier and inventory management – purchasing, stocking and handling perhaps just one component compared with many from disparate suppliers, maybe even including a custom magnetic part if best performance is required. Savings also accrue in SMD placement time, testing and agency approvals if necessary. Concerns about second sourcing are also being addressed with modules increasingly meeting industry standards for functionality and pinning such as DOSA. When total cost of ownership is calculated, a modular solution from companies such as RECOM can be the clear winner - we have done the work, so you don’t have to!

RECOM: We Power your Products
Applications