Pulse quietening at source for higher-frequency power and signal switching

Lead Research Organisation: University of Bristol
Department Name: Electrical and Electronic Engineering

Abstract

Today's portable microelectronic systems, such as mobile telephones, require high energy efficiencies to further battery life. They also require compact electronics. Combining these two requirements poses a problem with relation to their power supplies, since it implies greater miniaturisation, high conversion efficiencies and high power densities.

Using silicon-based DC/DC converters places limits on how far these improvements can go. We intend to make use of a new gallium-nitride transistor technology to develop smaller, more efficient power supplies. Specifically, we will produce a 10W power supply in this new technology, with a high voltage conversion factor, and integrate it inside a contemporary microelectronics package.

The only way that this will work is to operate the new power supply at incredibly high switching frequencies (>100 MHz), which is 10-100 times faster than today's power supplies. The project then revolves around solving the challenges of: 1) how to operate such a power supply at very high frequencies; 2) how to integrate it into a small, modern, microelectronics package. We expect key challenges to be the creation of unacceptable electromagnetic emissions from the high switching speeds, and the need to accurately control the impedances of the circuit, since circuit impedances become more significant the faster one switches.

We will solve the challenges by deploying an advanced version of a drive-pulse shaping technique that we call "pulse quietening", and by using modern integration techniques, including the creation of a custom chip to control the new power supply. Our method is to create several prototypes, running at increasingly high speeds, from 1MHz up to 100 MHz. We will create models and theories about the most efficient way to drive the power supply transistors and measure the outputs, as well as furthering our knowledge and application of pulse quietening.

Planned Impact

The project is structured in such a way as to impact both the academic sector and the commercial one. We will, for the first time, address the question of electromagnetic noise reduction in very fast power supplies, whilst simultaneously enabling their miniaturisation. The outputs from the project will include new theories and models of GaN FET transistors, waveform generation and measurement and electromagnetic noise reduction. Each of these represents a new contribution to the worldwide body of knowledge, and a key discovery in its own right.
Using the accepted methods of journal and conference publications as primary mechanisms, we will produce and disseminate a significant body of novel research contributions to the field. Specific outputs will include waveform shaping methods, measurement techniques, validated high-frequency models, reference layouts, reference switching aid-circuits, reference package equivalent electric circuit models, device design improvements, and design processes.

We have a detailed communication plan for engaging with the academic community and interested commercial parties. This will enable system designers in both realms to deploy our new high frequency switching techniques to reduce overall power losses in their power supplies and/or reduce the area, cost and packaging requirements of a microelectronic system. Existing techniques, such as DVFS, will be enhanced using our novel techniques, and we anticipate this would represent one initial direction for follow-on research.

The outputs from this project will address an immediate need in the UK's £62 billion commercial microelectronics sector and the projected US$8.4bn worldwide GaN FET sector.
The area is one where the UK remains a world-leader, and statements of support from UK-based companies, such as Imagination Technologies and multinational NXP can be found in the letters of support that accompany this proposal. Their enthusiasm for this project indicates its significance to industry.
The creation of an integrated, high-efficiency switched mode power supplies would help many UK companies to produce more energy efficient microelectronic devices than they currently do. This would result in smaller and cheaper devices and longer battery life, delivering a competitive advantage.
The proposal includes close collaboration with NXP, and direction and support from Imagination technologies, who represent two key beneficiary sectors of the UK economy. This level of interaction will maximise the research's applicability to the UK and worldwide commercial sector, whilst still allowing us the freedom to explore the intellectually interesting problems and deliver novel models and concepts to the academic community.

The research groups involved have track records in engaging the wider community, and will continue to do so during this project. Schemes such as media interaction and outreach will enable us to communicate our main discoveries to a diverse audience, disseminating formal and informal community knowledge, and ensuring that our discoveries reach as wide a potential user set as possible.

The project will employ two post-doctoral researchers and two members of academic staff. Over the course of the project, all will develop their knowledge and research, analytical and engineering skills. Combined with the university's formal staff training programmes, and the ability to inspire other individuals who encounter our research, we will make a contribution to the development of "people capital" in the high-tech sector. Finally, one of the post-doctoral researchers will spend time at the end of the project embedded in a partner company. They will use this time to identify contributions of commercial interest from the project deliverables, and organise a UK-wide workshop to disseminate this knowledge and spark wider interest from the commercial sector. We expect to gain new future partners and accelerate UK R&D efforts by this mechanism.
 
Description Most generation and usage of electricity uses power electronics to control the flow of power. Some of the controlled power is however lost, and this loss needs to be minimised. The power electronics switches currents on and off 100'000 times a second, and it is during the switching that the majority of the loss occurs. The next generation of "wide bandgap" power electronics will switch at least 10 times more often and 10x faster, and have 10x lower power losses, see for example the winners of the Google Littlebox Challenge at www.littleboxchallenge.com. In addition, the flow of power will be controllable with a 10x faster response, which will increase energy efficiency of usage, e.g. in microprocessors, by allowing power availability to continuously match the power requirements of a complex task.
This 10x increase in power efficiency and controllability relies on being able to switch current from on to off extremely fast, that is in around 1 ns. This turn-off transient, and its opposite turn-on transient, is completely uncontrolled, and therefore currents and voltages do not follow a desired path or waveform, but they overshoot and oscillate. This leads to an increase in power loss, in electromagnetic emissions, and in device failures.
It has, until now, been impossible to control these fast transitions and reduce these problems. In fact, the current transient cannot even be observed because inserting even the most advanced current sensor into circuits with 1 ns transients, slows the transients down, thereby increasing loss and undoing the benefit of using wide bandgap devices.
1) This project has for the first time made these transients controllable:
We have developed a chip with an integrated function generator that is able to control wide bandgap devices during the 1 ns transient. This has required developing electronics that changes its output 8x faster than its own clock frequency (in our case once every 150ps), and that is powerful enough to drive the resulting complex waveforms through the circuit connections to obtain significant controllability of the switching devices.
2) The project has also, for the first time, made these current transients visible in a high-speed, high voltage circuit:
We have developed a concept to accurately sense 1 ns current transitions in a way that does not impact the behaviour of a high-speed power switching circuit.
3) By being able to control the transients, and see what is going on, we can at last begin to explore exactly how to control power electronic circuits in order to reduce power loss, electromagnetic emissions, and device failures. This research is currently ongoing.
Exploitation Route Control of switching transients is referred to as 'active gate driving'. Its benefits have been demonstrated in current generation MOS-gated silicon devices such as current balancing in parallel devices, voltage balancing in series devices, reduction of current and voltage overshoot, suppression of EMI generation without significantly affecting power loss, and optimisation of power efficiency under constantly changing load conditions. With the move to a new generation of faster wide bandgap devices, the need for shaping the switching waveforms further increases, however none of these methods have been applicable so far as no control or sensing devices were available. Now that these are available through this project, the transfer of established high efficiency techniques can be transferred.
We are disseminating our knowledge and data to low and high voltage design communities in the first instance. We believe that our devices form the required research tools that companies will require in order to develop low-cost control techniques for wide bandgap power electronics.
Sectors Aerospace, Defence and Marine,Digital/Communication/Information Technologies (including Software),Electronics,Energy,Healthcare,Transport

URL http://tinyurl.com/PulseQuietening
 
Description SUMMARY Recently emerged SiC and GaN power transistor technology is predicted disrupt power electronics as silicon did when it took over from valves: power lost in converters could drop 10-100× and therefore this electrical equipment could shrink 10×. This is, however, only possible if SiC and GaN transistors can be switched 10-100× faster than existing silicon (e.g. at rates of 100 V/ns), and there is currently no complete solution for industry to achieve this speed. TECHNICAL OUTCOMES From 2013-2018, the EPSRC Pulse Quietening project developed a number of new high-speed solutions to this problem: 1) 10 GHz driver: The world's fastest arbitrary waveform gate driver chip, whose output that can be changed between 10 A and -10 A every 100 ps. This is very fast: 400× faster than the next best active driver; 100 ps is the time it takes light to travel 3 cm! 2) Active GaN driving: The first demonstration on GaN of fast active gate driving (where gate signal transitions are complex profiles instead of a single step), to combat problems that fast switching causes, such as ringing at 100s of MHz, overshoots, and EMI. 3) Infinity sensor: The world's lowest-impedance current sensor (0.2 nH, 10× lower than best competition), and the only one, to our knowledge, to directly visualise the high-speed current problems in a GaN power circuit, without affecting switching waveforms. The high-speed layouts, sensors, and drivers have given industry options of creating intelligent power modules where drivers and power devices are integrated into one Power Electronic Building Block. These will switch faster than is currently possible with both GaN and SiC. IMPACT ON INDUSTRY AND NEW INDUSTRY COLLABORATION We have changed how industry thinks about GaN and SiC. Before the Pulse Quietening project started, industry was developing GaN devices with integrated drivers on the actual GaN power device (very expensive). We proved however that full-speed active driving from a separate silicon die is possible, despite interconnect impedances. Around 100 infinity sensors have been requested and supplied to over 20 global companies and 20 research organisations. We have had requests to put sensors into prototype products, and several repeat requests for more sensors. Semiconductor companies have requested meetings about the driver technology, and we have secured funded collaborative work from one large semiconductor manufacturer to use the knowledge developed during the Pulse Quietening project. We were invited to present the results at the invite-only European Centre for Power Electronics SiC and GaN user forum which had 80% industry attendance. We have had numerous unsolicited requests for drivers from companies and university research groups. A selection of these are being pursued. The 10 GHz driver is a development tool for engineers. Rather than a commercial driver, it is the world's fastest and most powerful arbitrary waveform generator on a chip. One UK and one mainland European university have adopted it into their research and have created publishable results, a further two UK universities are working on the transfer. Several companies have requested consultancy or joint development projects to achieve fast switching, to apply the outputs of our research. SPAWNING OF FURTHER RESEARCH PROJECTS These research successes have led to two further projects funded by the EPSRC Centre for Power Electronics: First, we formed a consortium (led by Bristol, including Imperial, Strathclyde, and Edinburgh, and with 8 active industrial partners) who successfully bid to EPSRC for £2M to develop fast-switching GaN and SiC technology, that is adaptive to changes in real-world conditions such as temperature and load current, and immune to the variability of power devices. This is giving the 8 sponsors and an increasing number of industry partners direct access to the technology, and new means of overcoming the hurdles to adopting SiC and GaN. Second, we were invited to form a consortium led by Warwick, to develop methods of using active gate driving and high-bandwidth sensing to determine GaN junction temperature and ultimately the health of GaN devices.
First Year Of Impact 2017
Sector Aerospace, Defence and Marine,Electronics,Energy,Transport
Impact Types Economic

 
Description Imperial power electronics 
Organisation Imperial College London
Country United Kingdom 
Sector Academic/University 
PI Contribution Over the years we have had several joint research efforts. At present we are working towards a transfer of new technology from Bristol to Imperial in order to facilitate the broadening of our research.
Collaborator Contribution Imperial will apply our research outcomes in new areas, such as RF-based power device health monitoring methods.
Impact 1) Joint funding. 2) Fruitful exchange of ideas.
 
Description Low EMI GaN converter 
Organisation The Otto-von-Guericke University Magdeburg
Country Germany 
Sector Academic/University 
PI Contribution We have tranfered our high-speed gate driver electronics to Magdeburg Uni, helped them build it into their research, and helped with the paper writing.
Collaborator Contribution Magdeburg are building our electronics into their test facilities to research how to reduce electromagnetic noise in power converters. They have shown for the first time in a commercial electromagnetic emissions test facility, that the methods developed at the University of Bristol indeed reduce emissions without reducing efficiency.
Impact IEEE Transactions on Power Electronics publication of new EMI quietening method.
Start Year 2017
 
Description Warwick power electronics 
Organisation University of Warwick
Department School of Engineering
Country United Kingdom 
Sector Academic/University 
PI Contribution We are in the process of discussing the joining up of research outcomes in order to permit instantaneous temperature sensing on wide bandgap power devices. This has never been achieved before, and could be commercially important.
Collaborator Contribution We are in the process of discussing the joining up of research outcomes in order to permit instantaneous temperature sensing on wide bandgap power devices. This has never been achieved before, and could be commercially important.
Impact This is multidisciplinary, between device physicists and power electronics engineers.
Start Year 2014
 
Title AWG driver 
Description 8 GHz arbitrary waveform driver. The nearest commercial device is a £6000 device that has a 60x lower bandwidth, and 10x lower driving capability. It has a significant software element. 
Type Of Technology Physical Model/Kit 
Year Produced 2016 
Impact This is still in its infancy and transfer out of the research group is currently being discussed. 
URL http://tinyurl.com/PulseQuietening
 
Description Invited talk at ECPE workshop on SiC and GaN 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Invited presentation at the European Centre of Power Electronics Workshop on use of GaN and SiC. 200 industry attendees, 50 from academia. My presentation provoked over an hour of discussion, and two companies and a university proposing collaborations.
Year(s) Of Engagement Activity 2017
URL http://tinyurl.com/gatedriving