Reducing the weight of unmanned aerial vehicles (UAVs) pays large dividends in their ability to carry more fuel, support more advanced payloads (radar, imaging, sensors, navigation and guidance, uplinks and downlinks), achieve longer flight times, and operate from shorter runways. Reducing the weight of the electronics, and especially the power-supply subsystem, is a major area for potential improvement.
A typical UAV has a turbine-based generator supplying primary ac power, which is converted to 270 V dc, 48 V dc, or 28 V dc and then distributed and down-converted as needed via intermediate dc-dc converters. The 270 V dc can be dropped down to 45 V, for example, by a fixed-ratio, efficient bus converter.
While lower voltages are easier to manage in terms of insulation and safety-required spacing, it’s much more efficient to route power at higher voltages with correspondingly lower currents to minimize I2R losses and then provide the many dc rails needed as close to their loads as physically possible. At the same time, the distribution loss must be weighed against conversion losses at each stage.
Within the UAV, a wide range of voltages is required, from 1 to 28 V, at levels ranging from 50 to 75 W for the nose-wheel actuator to 130 A/1 V for the computing-engine core. In a representative design, a single-slot 28-V Versa Module Eurocard (VME) backplane power supply provides up to four independent voltages at more than 500 W total power.
Choose Your Topology
To improve the power/weight ratio (kW/kg), designers have a choice of topologies to balance both conversion loss and distribution loss:
- Centralized: A single converter takes the high voltage, then provides and distributes the lower-voltage rails. This approach is rarely used because it is inefficient and physically restrictive.
- Distributed bus: The high voltage is transformed to a mid-range distribution value such as 48 V dc, which, in turn, goes to modular “brick” converters (often isolated) for the individual subsystems (Fig. 1a).
- Intermediate bus: The distribution bus voltage is stepped down to an intermediate, lower voltage such as 12 V dc and routed to be closer to the loads, where point-of-load (PoL) converters (which can be non-isolated) provide rail voltages (Fig. 1b).
- To achieve even better performance, a partitioned or “factorized” topology lets designers decouple the voltage-transformation and regulation functions (Fig. 1c). Doing so allows optimization of each block without impacting the others, while minimizing duplication of functions.
The power chain can be more efficient, lighter, and less expensive, since the constraints and requirements on the individual functions are reduced. For example, the switching regulators can run at 1.5 MHz, compared to the 250 kHz of an isolated brick (for smaller, lighter filters), while any isolation and transformation can be at a higher, more space-efficient frequency.
While reducing the power consumption of the payload is critical, there’s a limit to being able to do so with lower-power designs and still meet operational requirements. Therefore, reducing the weight of the power-supply subsystem is critical and can yield large benefits.
To do this, start with the obvious: efficiency. Going to a more efficient supply chain, such as from 85% to 90%, may seem to be a costly exercise that provides only a minor improvement of a few percentage points. However, that perspective is like looking through the wrong end of a telescope.
Instead, look at it in terms of inefficiency. You’ve dropped the loss from 15% to 10%, an improvement of one third, which is a significant gain. As a result, heatsinks can be far lighter and smaller, and the cooling subsystem can be simplified. (Remember, convection cooling can be a challenge in the thinner air of higher altitudes.)
Designing for pulsed versus constant-power load requirements is another area for improvement. A reliable design must be sized to handle the pulse-load peaks, of course, but doing so can mandate extra capacity (weight, cost, quiescent dissipation) that is rarely used.
One solution is to use a converter topology such as Vicor’s Sine Amplitude Converter (SAC), which uses a high-frequency resonant tank to move energy from input to output and typically has 50% overpower capacity if the pulsed load is less than 1 ms in duration. That may be enough headroom for some low-duty-cycle pulse loads in the system.
Finally, reducing the bulk capacitance needed by the supplies will dramatically reduce weight and size. How can you do this without compromising performance? A low series impedance in the switching transformation stage is key. By using a high-frequency SAC between a regulated point and the final load, the series impedance is scaled down and energy-storing bulk capacitance scaled down by K2, where K is the SAC’s VIn/VOut ratio.
As with most engineering challenges, there’s no single magic solution, but a quartet of techniques can improve the power/weight ratio significantly, by as much as four times. Also, significant bill of material (BOM) cost may be possible, since smaller magnetic and capacitive components generally cost less, while reducing cooling requirements cuts material costs as well.