Design Considerations for High-power AC/DC Power Supplies
2019. 2. 5
Learn the key considerations for power solutions in heavy automation for next-generation robotics and e-mobility systems.
1. Introduction
The purpose of this whitepaper is to highlight key considerations for power solutions in heavy automation for next-generation robotics and e-mobility systems. These application spaces are being driven by the convergence of advanced manufacturing (a.k.a. – Smart Manufacturing, Industry 4.0) and telecommunications (e.g. – 5G & Beyond, machine-to-machine or M2M communications, vehicle-to-everything or V2x communications, Internet of Things (IoT), Industrial IoT (IIoT), etc.).
Another prime objective is to facilitate in mapping these considerations to an informed decision-making process for evaluating power solutions. In general, this whitepaper focuses on the aspects of higher-power solutions, though the linkage to high numbers of tiny-power devices and ubiquitous sensing will also be touched upon.
Another prime objective is to facilitate in mapping these considerations to an informed decision-making process for evaluating power solutions. In general, this whitepaper focuses on the aspects of higher-power solutions, though the linkage to high numbers of tiny-power devices and ubiquitous sensing will also be touched upon.
What does high power mean to you?
When discussing high vs. low power, it is prudent to first stake a step back and define what these terms mean since they can certainly be relative and do not carry formal definitions in the industry (as opposed to voltage levels, for instance). If one is working in the power engineering space on utility-scale converters, then a low-power system might be something on the order of a megawatt. Conversely, someone working in the IoT/IIoT space may consider a high-power system as one with a budget of a single watt or even less.
Please also keep in mind “high power” does not automatically imply both high voltage and high current. When discussing power levels in applications with others, the first thing one should always do is ask what high power means to the person they are talking to as well as give some example voltage and current levels of the application at hand. This will help to level set and prevent a whole lot of miscommunication down the line.
Please also keep in mind “high power” does not automatically imply both high voltage and high current. When discussing power levels in applications with others, the first thing one should always do is ask what high power means to the person they are talking to as well as give some example voltage and current levels of the application at hand. This will help to level set and prevent a whole lot of miscommunication down the line.
Industry 4.0 & e-Mobility are Redefining the Power Industry
Industry 4.0 and e-mobility are driving paradigm shifts in the way systems (and therefore power solutions) are designed and implemented. As a matter of fact, these new opportunities to integrate technology and power with system solutions are having big impacts on the supply chains themselves, since formerly disaggregated solutions must now all converge and work cohesively in automated systems.
An autonomous vehicle (AV) is the quintessential example because it is essentially an advanced electromechanical system that combines integrated motor drives for the electric drive train with energy storage/management (e-mobility) with a ubiquitous sensor network that communicates internally/externally (IIoT/M2M) and increasing amount of personal entertainment functionality (infotainment). The AV must essentially become a data center/networking base station/multimedia hub on wheels. The economic viability of such solutions also requires the best in robotics and factory automation to marry new designs with the changing markets. Given the mobility requirements of all the aforementioned solutions, size, weight, and power (SWaP) characteristics are critical enablers.
There are some general trends that are fairly universal, such as the need to increase overall system power and power density. The lights may be switched off in the factory because the machines do not need them (so-called ‘dark warehousing’) and power converters will get ever more intelligent and efficient, but the overall power bill will still increase because power consumption tends to always increase as a result of packing in more loads. This applies from tiny-power systems for ubiquitous IoT sensors all the way up to the biggest ones for robot arms, even though the size of the power supplies themselves may shrink by increasing switching frequencies to decrease the size of bulkier filter components.
As new application spaces are constantly being invented, overall power budgets do not tend to shrink (or even stay flat), even with improving power technology that increases power density to reduce power solution size and decrease thermal losses.
In untethered applications, fuel cost is typically a top metric for application viability, so whether it be a fossil-fuel-powered jet or an electrically-driven train, a cost can be associated with each kilometer traveled and therefore translated directly into the impact of the on-board energy consumption. Since power solutions are all about energy conversion, commutation, and regulation, the associated SWaP factors are critically linked to both capital expenditures (CAPEX) and operational expenditures (OPEX) in total cost of ownership (TCO) calculations.
An autonomous vehicle (AV) is the quintessential example because it is essentially an advanced electromechanical system that combines integrated motor drives for the electric drive train with energy storage/management (e-mobility) with a ubiquitous sensor network that communicates internally/externally (IIoT/M2M) and increasing amount of personal entertainment functionality (infotainment). The AV must essentially become a data center/networking base station/multimedia hub on wheels. The economic viability of such solutions also requires the best in robotics and factory automation to marry new designs with the changing markets. Given the mobility requirements of all the aforementioned solutions, size, weight, and power (SWaP) characteristics are critical enablers.
There are some general trends that are fairly universal, such as the need to increase overall system power and power density. The lights may be switched off in the factory because the machines do not need them (so-called ‘dark warehousing’) and power converters will get ever more intelligent and efficient, but the overall power bill will still increase because power consumption tends to always increase as a result of packing in more loads. This applies from tiny-power systems for ubiquitous IoT sensors all the way up to the biggest ones for robot arms, even though the size of the power supplies themselves may shrink by increasing switching frequencies to decrease the size of bulkier filter components.
As new application spaces are constantly being invented, overall power budgets do not tend to shrink (or even stay flat), even with improving power technology that increases power density to reduce power solution size and decrease thermal losses.
In untethered applications, fuel cost is typically a top metric for application viability, so whether it be a fossil-fuel-powered jet or an electrically-driven train, a cost can be associated with each kilometer traveled and therefore translated directly into the impact of the on-board energy consumption. Since power solutions are all about energy conversion, commutation, and regulation, the associated SWaP factors are critically linked to both capital expenditures (CAPEX) and operational expenditures (OPEX) in total cost of ownership (TCO) calculations.
For Mission Critical, Large-System, & Transportation Applications: Safety is the Name of the Game
From enhanced isolation levels to much more stringent electromagnetic compatibility (EMC) and shock & vibration standards, most high-reliability (especially higher power) applications are regulated by a rigorous and confusing array of standards that may be required by a customer and/or be dictated by regional regulations. Grounding can be quite intricate and just as critical to operation as it is to safety.
Your application may be considered critical, but so is avoiding “negative, localized, thermal events” (a.k.a. – fires). If you are not familiar with the typical safety and qualification standards of an industry, then that is a good place to look before you start with the power supply design.
An example would be a power supply designed to be built into a medical-grade robotic device, maybe a robot arm to help a surgeon operate or a powered table than can lift or turn a patient to improve access for a medical procedure. Medical safety requirements are significantly stricter than those for industrial applications, require double Means of Operator Protection (2MOOP) and double Means of Patient Protection (2MOPP), as well as higher insulation withstand voltages, wider creepage and clearance separations, and more severely-limited leakage current limits.
Your application may be considered critical, but so is avoiding “negative, localized, thermal events” (a.k.a. – fires). If you are not familiar with the typical safety and qualification standards of an industry, then that is a good place to look before you start with the power supply design.
An example would be a power supply designed to be built into a medical-grade robotic device, maybe a robot arm to help a surgeon operate or a powered table than can lift or turn a patient to improve access for a medical procedure. Medical safety requirements are significantly stricter than those for industrial applications, require double Means of Operator Protection (2MOOP) and double Means of Patient Protection (2MOPP), as well as higher insulation withstand voltages, wider creepage and clearance separations, and more severely-limited leakage current limits.
The medical EMC limits are also more robust to reduce the possibility of adjacent equipment interfering with each other and impacting safety. This all means that an industrial-grade power supply that is perfectly capable of delivering enough power for the medical robot application would not be allowed to be used unless it was extensively modified to meet the additional, medical-grade safety safeguards, meaning a complete redesign of the PCB layout, transformer construction, and EMC filter components. Failing to understand the required standards of a particular industry can be very expensive.
Moving around a lot of energy, while still trying to maximize SWaP factors, typically means going to higher voltages because resistive power loss depends on the square of current flowing through the connections and conductors (I2R loss), although power is directly proportional to the current (P = I x V). Higher bus voltages may be used to reduce resistive losses by minimizing the distribution current for the same power delivery. This can result in a couple of challenges for the power supply design. One is that the higher the voltage, the greater the spacing required (e.g. – physically larger power supply) and/or the greater the insulation required (e.g. – heavier, more expensive power supply).
At very high voltages, corona effects, which can cause an electrical discharge due to the ionization of air, or tracking across non-conductive surfaces can occur, therefore potentially shocking a person/item that is not in direct contact with the energized conductor.
Safeguards against such hazards must be built into the design. Another challenge is that components must be rated for higher voltages, which tend to make them bulkier, more expensive, and perhaps more difficult to source than their lower-voltage counterparts. Thus a balance is required between low resistive losses and increased costs/complexity.
As mentioned before, safety is absolutely paramount in the types of high-power applications discussed here. A “reliability failure” of a high-power supply can mean serious injury or catastrophic damage to very, very expensive equipment (still to a lesser degree). Particularly when it comes to high-voltage applications, please be sure to do your homework on appropriate safety standards and do not attempt to mess with high-voltage designs on your own bench unless you are confident you have the proper training and equipment.
The very first place to start is in the voltage isolation, grounding, and leakage requirements for your application as these requirements and specifications can give you a real appreciation for the dangers associated with inexperience on these topics.
Moving around a lot of energy, while still trying to maximize SWaP factors, typically means going to higher voltages because resistive power loss depends on the square of current flowing through the connections and conductors (I2R loss), although power is directly proportional to the current (P = I x V). Higher bus voltages may be used to reduce resistive losses by minimizing the distribution current for the same power delivery. This can result in a couple of challenges for the power supply design. One is that the higher the voltage, the greater the spacing required (e.g. – physically larger power supply) and/or the greater the insulation required (e.g. – heavier, more expensive power supply).
At very high voltages, corona effects, which can cause an electrical discharge due to the ionization of air, or tracking across non-conductive surfaces can occur, therefore potentially shocking a person/item that is not in direct contact with the energized conductor.
Safeguards against such hazards must be built into the design. Another challenge is that components must be rated for higher voltages, which tend to make them bulkier, more expensive, and perhaps more difficult to source than their lower-voltage counterparts. Thus a balance is required between low resistive losses and increased costs/complexity.
As mentioned before, safety is absolutely paramount in the types of high-power applications discussed here. A “reliability failure” of a high-power supply can mean serious injury or catastrophic damage to very, very expensive equipment (still to a lesser degree). Particularly when it comes to high-voltage applications, please be sure to do your homework on appropriate safety standards and do not attempt to mess with high-voltage designs on your own bench unless you are confident you have the proper training and equipment.
The very first place to start is in the voltage isolation, grounding, and leakage requirements for your application as these requirements and specifications can give you a real appreciation for the dangers associated with inexperience on these topics.
Modern Applications Bring Together the Challenges of Multiple Industries in Single Applications
This convergence of power, computing, and wireless communications require different thinking and appreciation for each and every milliwatt in what is called the Power Value Chain (PVC), which is simply a representation of the energy flow across all the distribution/conversion steps between source and load.
An example of a PVC is shown in the figure below, which shows the energy flow from a power plant to a networked device at the edge of the network. It characterizes a component’s power consumption by a unitless number to assess the overall cost of energy utilization at any given point within the PVC, which is known as Power Cost Factor (PCF). In this case, the end user equipment could be a high-end manufacturing robot or a smartphone, but in any case shows the PCF (e.g. – the cost to power) the wireless link can be upwards of 1,000,000+x if the power for this link is sourced from an upstream power plant, where it can be 1x if generated locally (i.e. – energy harvesting or scavenging). That is a huge difference when each milliwatt worth of received data requires 20-60++ W to actually be generated by a power plant, when only a couple milliwatts may need to be generated locally.
When you scale this by 10,000s or even 100,000s of sense points or wireless sensor networks (WSN) in a single factory, it can quickly be seen how all those tiny sensors can have a disproportionate and unexpected impact on the upstream constituents of a PVC.
Fig. 1: Diagram of PVC for a Connected System with Associated PCF Metric [1]
This concept is important to consider in large power system applications because it is helps to visualize power requirements at the point of consumption in the context of the PVC that provides the power, which can really help to internalize impacts to both CAPEX and OPEX. In today’s connected world, this representation also facilitates understanding how the deployment of very large numbers of wireless devices (even if relatively small power) can have a disproportionate impact on the upstream energy sourcing needs.
The PVC concept transcends far beyond considerations for increasing power/density and must encompass the architectural landscape upon which so many new devices are deployed. For instance, one might need to design intermediate energy storage solutions to ensure a high-reliability application’s needs are always met, while still wanting to minimize the need for more costly (in CAPEX and OPEX) overhead solutions such as redundant, backup systems. One should also note this PVC concept (and associated PCF metric) can be used to analyze any string of sources and loads, whether it be across a cellular network or fully-contained within a system (or even IC).
An example of a PVC is shown in the figure below, which shows the energy flow from a power plant to a networked device at the edge of the network. It characterizes a component’s power consumption by a unitless number to assess the overall cost of energy utilization at any given point within the PVC, which is known as Power Cost Factor (PCF). In this case, the end user equipment could be a high-end manufacturing robot or a smartphone, but in any case shows the PCF (e.g. – the cost to power) the wireless link can be upwards of 1,000,000+x if the power for this link is sourced from an upstream power plant, where it can be 1x if generated locally (i.e. – energy harvesting or scavenging). That is a huge difference when each milliwatt worth of received data requires 20-60++ W to actually be generated by a power plant, when only a couple milliwatts may need to be generated locally.
When you scale this by 10,000s or even 100,000s of sense points or wireless sensor networks (WSN) in a single factory, it can quickly be seen how all those tiny sensors can have a disproportionate and unexpected impact on the upstream constituents of a PVC.
Fig. 1: Diagram of PVC for a Connected System with Associated PCF Metric [1]
This concept is important to consider in large power system applications because it is helps to visualize power requirements at the point of consumption in the context of the PVC that provides the power, which can really help to internalize impacts to both CAPEX and OPEX. In today’s connected world, this representation also facilitates understanding how the deployment of very large numbers of wireless devices (even if relatively small power) can have a disproportionate impact on the upstream energy sourcing needs.
The PVC concept transcends far beyond considerations for increasing power/density and must encompass the architectural landscape upon which so many new devices are deployed. For instance, one might need to design intermediate energy storage solutions to ensure a high-reliability application’s needs are always met, while still wanting to minimize the need for more costly (in CAPEX and OPEX) overhead solutions such as redundant, backup systems. One should also note this PVC concept (and associated PCF metric) can be used to analyze any string of sources and loads, whether it be across a cellular network or fully-contained within a system (or even IC).
2. Specialized design requirements for hight-power applications
Wide bandgap (WBG) power components such as those made from gallium nitride or silicon carbide, GaN or SiC, respectively, have gained fame in recent years to address the perpetual need to improve power supply SWaP factors, primarily driving an increase in switching speeds, which decreases the size of larger filter components.
Improved efficiency has the added bonus of mitigating thermal losses and therefore shrinking thermal management solutions (if not completely eliminating in some cases). Large systems stand much to gain from the unique properties of these materials over their Si counterparts.
GaN’s value proposition currently tends to be in applications <800V (competing with Si in terms of efficiency and size), where SiC’s value proposition tends to be in applications up to 3 kV (competing with IGBTs in terms of thermals and size).
SiC has a thermal conductivity figure of merit 3x better than Si and can therefore either run with less thermal margin, be more reliable in a common application, or operate in a much harsher environment for longer.
Improved efficiency has the added bonus of mitigating thermal losses and therefore shrinking thermal management solutions (if not completely eliminating in some cases). Large systems stand much to gain from the unique properties of these materials over their Si counterparts.
GaN’s value proposition currently tends to be in applications <800V (competing with Si in terms of efficiency and size), where SiC’s value proposition tends to be in applications up to 3 kV (competing with IGBTs in terms of thermals and size).
SiC has a thermal conductivity figure of merit 3x better than Si and can therefore either run with less thermal margin, be more reliable in a common application, or operate in a much harsher environment for longer.
Fig. 2: Radar Chart Benchmarking Figures of Merit for WBG Materials Vs. Si [2]
While all the merits of WBG devices sound great, their extremely fast transitions come with a handful of design challenges. The very high di/dt that results from nanosecond gate transitions must be carefully controlled for a variety of reasons, which can include generating excessive voltages from parasitic inductances that can cause an electrical overstress (EOS) situation and hard failure of your power supply or introducing too much energy to stay within conducted and radiated emissions EMC requirement thresholds. WBG devices also tend to have lower gate threshold voltages than their Si counterparts and may therefore be more susceptible to false turn on due to noise on the gate drive or a spike induced by the gate-drain capacitance (a.k.a. – Miller capacitance) of the switch itself.
The key takeaway message here is that WBG components are NOT drop-in replacements for their Si contemporaries. There are specialized design requirements far beyond the scope of this whitepaper, but can be worth a lot of the pain and effort when SWaP factors are tantamount.
That being said, the learning curve and design nuances should not be underestimated, which is why power vendors experienced in this space even offer gate-drive isolated power solutions with asymmetric output voltages specifically tailored for WBG designs so designers can focus on optimizing their system instead of just the drive train of their power supplies.
That being said, the learning curve and design nuances should not be underestimated, which is why power vendors experienced in this space even offer gate-drive isolated power solutions with asymmetric output voltages specifically tailored for WBG designs so designers can focus on optimizing their system instead of just the drive train of their power supplies.
Fig. 3: Picture Showing Si-based vs. SiC-based Home Inverter Solution [2]
3. Specialized resource requirements for hight-power applications
Beyond the design considerations for higher-power systems, the sheer amount of power and size of your system may necessitate some very unique needs to effectively test and qualify your total solution, maybe requiring a custom solution.
High-reliability applications must derive ...
High-reliability applications must derive ...
Want to read the whole Whitepaper?
용도
