Power trends: TDK’s Advanced Technology Centre

Interviews | June 21, 2017

By Nick Flaherty

TDK’s Advanced Technology Centre (ATC) emerged from the purchase of power supply maker Coutant Lambda in Ilfracombe in 2005. “We are 50% funded by Ilfracombe R&D and Israeli R&D but we are global so we share what we are doing with other groups globally Japan, China, Singapore, and the R&D managers from across the company get together each year,” said Andy Skinner, centre director and chief technology officer.

“There are six in the ATC team and four in design team and they work on a number of projects,” said Skinner. “The things that we are most interested in are switching devices so we have been paying a lot of attention to gallium nitride in the past three years. DC to DC converters have got smaller and smaller, and AC-DC conversion has lagged behind, so improving the methods of construction and thermal management is an area we are quite interested in.”

“If you are selling a power supply into a large piece of equipment size is not an issue but if you want it in an oscilloscope then bench space has a value. So if you can make it smaller or easier to cool and better RFI then those are drivers for the converter,” he said.

“We have good relationships with a number of suppliers and try to keep our roadmaps aligned with the them,” he said. “For GaN we’ve had samples from a number of suppliers and we are able to evaluate the technologies and the differences in the options that are available. The attraction is that they know it’s testing out early pre-production parts.”

The ATC evaluates new technologies separately from the design teams.

“One of the challenges of developing new products is that if you don’t have a separate mechanism for evaluating new technology you end up taking higher risks with new technologies in a product development programme,” said Skinner. “The result of that is often project slippage, and time to market matters to us as for any engineering company,

“Because of the high mix of end user requirements we have a history of developing modular and/or configurable products to make all sorts of end products,” he said. “So we have a history of working with platform designs so we can evaluate technologies in a platform type applications and understand what it can do and also what it can’t do, what the limitations are.”

The findings are then shared with development groups around the world. “We have a steering committee that oversees what we do and involves the R&D groups in what we do – we don’t want to develop something to an end product to evaluate it but develop it enough to understand it. That depends on the technology – for example with a new semiconductor technology where reliability might be a concern you need to take through the standard reliability testing in a fairly finished process. A new MOSFET may just need a tweak with the same processes,” he said.

“For new technologies such as ceramic inductors, the interturn capacitances are very different to existing devices so we need to understand the limitations,” said Skinner. “There are some circuits where that make no difference and others with fast edges where it makes a huge difference, and so we recommend that it is only used in a subset of designs. We produce a report and share that with the R&D groups.”

“Its quite important that we have a dialogue with the engineers and we have design procedures and guidelines – we have a technology release process from scoping the project, early definition, milestones and the roll out, with seminars and white papers within the company to get the maximum awareness.”

The centre works on the technologies from the silicon to packaging and the wider the system. “Surface mount packaging is a big, big issue,” said Skinner. “The nice thing with leaded packages is you can bolt them in and add headsinks but you add 5 to 10nH lead induction and when switching at high speed that causes a lot of issues on driving and voltage overshoot. You can get much, much better performance from a SMT layout but then it is how you manage the heat – technologies like IMS (insulated metal substrates) give you a circuit board that can handle the heat but add a lot of capacitance especially with fast switching.”

Lessons from the substrates used for LEDs help. “One of the fortunate things is the LED industry has been facing thermal management issues on PCB for longer than we have and they can use IMS in a lot of the DC applications,” said Skinner. “There are a lot of companies now making FR4-like materials so we work with a lot of those – they are ceramic loaded but we have control over the thickness so we can control the capacitance.

Recruitment of engineers is also a key concern, especially with Brexit negotiations starting.

“A power engineer in Shanghai costs the same as the UK,” said Skinner. “The number of people educated in the UK [in the centre] is a relatively small percentage of the total, probably less than 20%. So it is important then UK universities produce more engineers. Power electronics has been given govt focus Virtual Power Electronics Centre, we hope that’s going to raise the profile of power electronics and produce more graduates from UK universities.”

One reason for this is that the centre combines both hardware and software expertise for digital controllers.

“We all the algorithms here,” said Skinner. “We tend to work with mixed C and assembler as we work in real time where 5ms is a long time. We find that we can better realise the higher speeds with assembler but the background functons are better implemented in C so they can move between controllers.”

“We try to minimise how much assembler there is – the choice is mostly driven by peripheral capability – such as TI’s C2000 and there are a number of ARM suppliers, but we are not tied.

“60% of the world’s electricity turns motors so for motor drives with digital control routines there’s a huge potential market and a number of suppliers have developed PWM modules to make the waveforms easier but these don’t have the resolution for us as they switch at 20KHz or less. The problem that most people haven’t got a good solution for the peripheral elements that go around the core eg how you start the microcontroller and do the gate drive at the lower end – that is the area that is not clear on the direction that the industry want to take.”

One of the biggest challenge is refining the algorithms to fit into cheaper controllers.

“The things that we are most interest in are some algorithm work on a control strategy,” said Skinner. “One of the things that drives the decision on using an ARM-M0 or M4 controller is how much maths you do, so if you have an algorithm with less maths you can move down to the 50 cent M0 parts. On one of the topologies we are looking at new control methods to simplify and reduce the maths – it will reduce the number of manipulations that are necessary. That’s the benefit of not being tied to a product design schedule, we can develop these kinds of things.”

www.tdk.com

Related stories: