Switch-mode regulators for space applications
Today's spacecraft subsystems require an increasing number of power rails and supply distributions with loads ranging from milliamps to tens of amps. It is important to choose the appropriate solution to meet the performance and reliability requirements for the target mission.
Switched-mode power supplies (SMPS) use energy storage elements such as inductors, capacitors or transformers to transfer energy from the input to the output at periodic intervals. In a SMPS, transistors are operated in their low-dissipative switching states instead of active mode as used by a linear regulator. When a transistor is on and conducting, the voltage drop across its power path is minimal, and when it is off, there is almost no current through the power path.
Inductors, capacitors and transformers are reactive components and given ideal parts, a SMPS could achieve power conversion and regulation with 100% efficiency. All power loss is due to non-ideal components and deficiencies in the control circuitry.
Inductors are fundamental to the design of switch-mode regulators described in this post, particularly their tendency to initially resist a change in current by establishing an induced voltage whose polarity opposes this change. An a.c. current through a coil induces a voltage across it as a result of the changing magnetic field, whose direction counteracts the change, i.e., it opposes the source for an increasing current and aids for a decreasing current.
The basic circuit topology for a Buck (step-down) converter together with its input and output currents and voltages are shown below. The direction of the inductor/load current is highlighted in green.
When the FET switch is closed, a voltage equal to VIN-VOUT is applied across the inductor causing a current to flow in a clockwise direction increasing at a rate equal to (VIN-VOUT)/L. The inductor stores energy by creating a magnetic field with the polarity of its left side being positive, opposing the source, in response to the changing current.
When the FET switch is opened, the source is removed but current must continue to flow through the inductor and load in the same direction with the free-wheeling diode now providing the return path. The voltage across the magnetic becomes VOUT, but its polarity reverses to aid the source, with the inductor current decreasing with a slope equal to -VOUT/L. The stored energy in the magnetic field is discharged through the load.
For a Buck converter, the output, load current is continuous while the input current pulses, which has implications for input and output filtering.
The basic circuit topology for a Boost (step-up) converter is shown below:
When the FET switch is closed, the unregulated input voltage is applied across the inductor causing a current to flow in a clockwise direction increasing at a rate equal to VIN/L. The inductor stores energy by creating a magnetic field with the polarity of its left side being positive, opposing the source, in response to the changing current.
When the FET switch is opened, the energy previously stored in the magnetic field will be transferred to the load to maintain current flow. The only path available for the inductor current is through the flyback diode, the capacitor and the load. A voltage equal to VOUT-VIN is applied across the inductor with its current decreasing at a rate equal to (VOUT-VIN)/L. The voltage polarity will be reversed, i.e., the left side of the inductor will now be negative, with two sources in series generating a higher voltage to charge the capacitor through the diode.
When the FET switch is closed again, the capacitor is able to provide the voltage and energy for the load. For a boost converter, the input current is continuous while the output current pulses.
The above descriptions for the Buck and Boost converters have assumed ideal components, e.g., the input source has zero impedance, the FET switch does not have any on-resistance with instant turn-on/off times, the diode has no forward-voltage drop, the inductor has zero winding resistance and does not saturate.
In reality, power dissipation is associated with every component and d.c. conduction losses and a.c. switching effects impact overall efficiency. The FET has a finite, drain-to-source on-resistance, the diode a forward-voltage drop, the inductor has winding resistance, and the FET has parasitic gate capacitance which has to be charged and discharged affecting the time required to switch the transistor.
For both the Buck and Boost topologies, continuous mode of operation is defined if the inductor current never reaches zero. To a first order, the magnitude of the output voltage is determined solely by the duty cycle of the switching rate, independent of the inductor value, the actual frequency and the load current.
A SMPS can be divided into two sections: the power-conversion stage and the small-signal control circuit. The inductor current and/or output voltage can be sensed and then regulated using a negative-feedback closed loop.
For simplicity, Figure 3 shows a voltage-controlled Buck converter: if the output voltage rises due to an increase in the unregulated input or a change in loading, the feedback voltage, VFB, rises. The output from the error amplifier decreases, reducing the duty cycle of the PWM controlling the switch which lowers (regulates) the output voltage, equalising the feedback and reference inputs at the comparator.
A well-designed SMPS is quiet, both electrically and acoustically, with the performance of the negative feedback regulation determined by the loop bandwidth and stability margins.
The Buck topology is an example of a forward converter, i.e., energy goes from the input, through the inductor and to the load simultaneously. The boost circuit has a flyback architecture as energy is periodically stored into and retrieved from a magnetic field.
Variants to the basic Buck and Boost topologies exist that offer both step-up and step-down regulation, the ability to invert the polarity of the output voltage, as well as the conversion of negative rails, e.g., Buck-Boost, Cuk, Zeta and SEPIC.
Both the basic Buck and Boost topologies are non-isolating converters and transformers can replace the inductors to provide full galvanic isolation between the input and output.
Qualified LDMOS-based regulators are available that offer higher efficiency and GaN FETs offer the space industry both higher efficiency and improved radiation tolerance. Any transient or long-term change in the output voltage of a SMPS caused by radiation may affect the circuits being supplied, with failure occurring if this alteration is sufficiently large!
The efficiency and functionality of SMPS degrade with increasing exposure to total-dose radiation leading to eventual functional failure, e.g., a change in the FET's threshold voltage can prevent it from switching, shorting the input to the output. Hardening-by-layout, e.g., enclosed transistors are being used to protect against total-dose effects. Annular designs increase chip area, capacitance and can be restrictive, and hardening-by-process, e.g., trench isolation between wells and local oxidation of silicon isolation between devices are also being used to alleviate total-dose effects.
Single-event effects can result in missing or extra-long pulses at the output of the PWM which can produce output transients of sufficient duration and amplitude to damage the parts being supplied. The various SMPS topologies exhibit different radiation responses and orbit sensitivity is one factor that may affect the choice of converter architecture.
The increasing influence of digital control within space-grade SMPS, e.g., within the PWM controller, has increased the sensitivity to single-event upsets, necessitating the use of mitigation techniques such as triple-majority voting, reducing the drain-to-source on-resistance of susceptible transistors, increasing the capacitance of sensitive nodes, and increasing the bias currents of analogue functions.
Cosmic radiation can trigger hard-failure modes such as single-event burnout (SEB) and single-event gate rupture in power MOSFETs. SEB is caused when high-energy particles pass through a reverse-biased, PN junction, leaving an avalanche trail in their wake. Whether this event causes permanent damage depends on the energy of the particle, the semiconductor technology, the type and physical design of the device, and the voltage stress around the junction. Reducing the applied voltage can dramatically lower the risk of component failure and improve reliability.
This blog article has just 'scratched the surface' of SMPS and their use in space applications, and many improvements can be made to the basic components and circuit topologies to achieve the most efficient, reliable and appropriate power solution for a mission. How switching regulators are integrated with the signal-chain hardware is equally important to meet system-level requirements, and careful PCB floor-planning, layout, and the correct allocation of signal, power and ground layers within the stack are paramount to realise the intended design goals - I will discuss some enhancements and best practices in a future post!
Switching regulators were used on the Sentinel-1, Earth-Observation spacecraft which launched on the 3rd of April. This mission will provide all-weather, day and night radar images for land and ocean services and further details about the Sentinel mission, the Copernicus programme, as well as the first images received can be viewed at: http://www.esa.int/Our_Activities/Observing_the_Earth/Copernicus/Sentinel-1. Many congratulations to all our partners who contributed to Sentinel-1 and hope you enjoy the following video!
Sentinel-1 lift-off, spacecraft separation and mission overview.
p.s. I'd like to credit Arianespace and ESA for the use of the video clips.