Posts

Taiyo Yuden and GE Ventures announced today that they entered into a license agreement for an intellectual property (IP) to fabricate substrates embedded with electronic circuits in late 2014. With this technology transfer, Taiyo Yuden and GE will construct a joint development framework toward the commercialization of next-generation wirebondless, embedded electronics circuits. Taiyo Yuden will develop Si-, SiC- and GaN-based wirebondless embedded electronics circuits with the technology and the patent licenses provided by GE Ventures Licensing. Embedded electronic packaging technologies provide significantly improved electrical performance (for example, reduced parasitics), can increase functional density of the electronics circuits by more than a third, and can increase efficiency by over 10%.

GE General Electric Shinko Power overlay POL power module packaging

Extract from “Packaging Challenges and Solutions
for Silicon Carbide Power Electronics” – ECTC 2012 – Ljubisa Stevanovic

“We are extremely pleased to have Taiyo Yuden as an embedded electronics partner,” said Pat Patnode, President of Licensing at GE Ventures.

“The strong demand for high performance electronics circuits continues to drive advance research, and GE is excited to partner with Taiyo Yuden to bring wirebondless embedded packaging solutions to the next generation of electronics.” These embedded electronics circuits can be built into higher level power assemblies and systems for a wide variety of applications.

“Taiyo Yuden will target power devices, Internet of Things (IoT), and wearable applications with this GE Power Overlay (POL) technology. “

This technology will allow additional new application opportunities leveraging the strength of Taiyo Yuden’s current packaging and assembly technology capability and experience.” Said, Hiroshi Kishi, Operating Officer, Research and Development Laboratory. GE Ventures accelerates innovation and growth for partners by providing access to GE technologies and inventions through licensing and joint development partnerships.

Source

I recently had the chance to test drive a Tesla. Not like any Tesla, the super-powered one: S85PD.

The one that has an “Insane” mode and can beat a Ferrari at traffic lights. Even though there was no Ferrari at the traffic light we stopped at, I can trust the fact that it can compete with many supercars.

And as a power electronics market analyst, I had to ask the sale guy about inverter and power modules. He did not know much about that. It’s not really his job, and he is not a design engineer. He knew a lot about the car and its features (which is enough to sell it) but not about the inside of the inverter, the power train, or the car…

So I asked Siri Google, and I had answers. I ended up on a Tesla car owners forum. I looked into it, and I found pictures and description of the inverter. Like the the picture down here.

Fig.1: 2×14 IGBTs in parallel is one leg of inverter—All packaged in discrete TO247 – From Tesla Roadster and Model S inverter

All about Tesla and power semiconductor packaging

That’s the truth. Don’t go for a reverse engineering on a Tesla inverter. Engineers just made an awesome car with very simple off-the-shelf products. Do not look for any highly special package for power modules International Rectifier (now-Infineon) would have made specially for Elon Musk (just because he is Elon Musk). There is no such thing. The inverter is made of TO-247 packages (Figure 1), derived from TO-220 in the 1990’s in order to handle more power and heat.

They used 20-year-old power module packaging technology to build the fastest electric luxury car on earth. It’s that simple.

I was shocked again! (The first shock was experiencing the acceleration of “Insane” mode.) I wanted to share that with someone. So I went on the internet and asked a very dumb question on the Power electronics group of LinkedIn.

A question that sounded like that:

Linkedinquestion

And many comments, questions and reactions (and “likes”. There are “likes” in LinkedIn too… But you cannot “poke” anyone. I wish we had that feature).

And for those of us who are not hanging out in this virtual professional world of LinkedIn, I wanted to summarize discussions here.

What you get from these discussions

1. “Module or discrete?” is still a 2015 question.

The first power module packaging design question is to know if you are going to use a power module or not. And that is a question that Tesla engineers had to ask themselves and that JB Straubel, CTO of Tesla, asked himself designing the model S. He talked about it during 2010 APEC conference, and I wish I was there 🙁

But according to the attendees, the choice of going for TO-series package rather than well-marketed EV series of power modules were:

  1. Simplicity of assembly
  2. Heat dissipation (and thus easy cooling)

Which is in total accordance to any engineer mind designing a product to:

  1. Work
  2. Be efficient

As the rest of the design is not engineering work … most of the time.

2. Power modules & IGBT packaging preconception

Most comments agree on the fact that a power module is a great thing, with a lot of room for innovation and many great features.

But they also agree on the fact that there are other factors to think about, when going for a design in power electronics, depending on the applications.

You still need to think about:

  • Volumes of production
  • Total cost of the solutions
  • Heat spreading and cooling requirements
  • Peak currents, ripples and other EMC stuffs
  • Size

Still you have interesting feedback putting down some preconception on power modules:

  • “The TO series are indeed very tough little beasts and I have been told by a senior semiconductor reliability engineer in some ways more reliable than a module, with respect to bond wires. Not something they usually advertise as the same company also makes modules targeted at EV/HEV”
  • “The TO-series package are really quite outstanding and I’ve used them a lot”
  • “That design allows Tesla to ‘insanely’ run 1500A through the inverter IGBTs and then the motor”

3. Theories on why Tesla made unexpected IGBT packaging choices

For most of the participants in the discussion, the choice of TO-247 totally made sense, but not always for the same reasons:

It could be because:

  • “The design was done when they were not sure about the volumes”
  • “There was no suitable module available at that time” (then what about today?)
  • “They wanted to reduce the risk of failure for this new product”

The point is not actually to try to find the right one, but to think about it.

We are in 2015, and we still find so many reasons why discrete TO-247 packaging could be a very good choice.

Tesla Model S inverter

Tesla Model S inverter – 3 phase represented by the triangle, having each two switches of 14 IGBTS in parallel

4. Power module packaging for EV/HEV—What’s next?

The state of the art and what would be the best choice in the future has also been discussed.

An interesting comment points out that:

the state of the art in power device packaging […] will most probably be in the higher volume, mass manufactured electric vehicles—from Ford, Toyota, GM, etc., There are definitely higher performance packages—with better characteristics in terms of thermal, inductance, tolerance to vibration, etc.

or that:

Tesla might not have the volume to drive custom power electronics packaging.

Which totally makes sense and proves that Toyota has a huge knowledge of how to build a power converter and hybrid car. They started Prius back in 1994 and have all the feedback and experience since then. On the other side, and that is Point The Gap comment: Is power electronics for Hybrid cars and Power electronics for full electric cars the same thing?

You get rid of a lot of constraints, starting with space and cooling restrictions when you design for a full electric car. (e.g.: Tesla’s cars have a rear and front trunk, with batteries under the car and inverters/motor group not taking much space).

4.1 .XT, SKiN and other IGBT packaging innovations

There is also a reminiscence of .XT Technology (Infineon) and SkiN (Semikron). These two technologies have been in one of our articles about PCIM 2015.

From Point The Gap point of view, these are all very interesting innovations. It’s just that they have been presented 2 to 4 years ago, and we are still waiting for available products. Remember the marketing campaign from Semikron in 2013 about SKiN technology. According to them, it may be ready in 2016. Why so much communication 3 years before availability…

And finishing with something that sounds very true to us:

“According to Tesla’s website the Power Electronics Module uses 84 IGBTs to power the 3-phase induction motor, or 14 IGBTs in parallel per switch. Is it the best solution? I seriously doubt it, although it may be the ‘lowest cost’ solution based solely on the bill of materials.”

4.2 Is best power semiconductor packaging “no-packaging”?

A parallel discussion also emerged on discrete packaging innovations. You must know EPC’s WLP (Wafer Level Package). The innovation here is that there is no package. The die has a protection coating on one side and bumps on the other. No wires, no wire bonding issues.

Of course this reduces the risks of failures. But as pointed in a comment: “Chipscale packages/no-packages are exciting at lower voltage and power for converters, power supplies, etc., but PCB interconnects bring back resistance and inductance.” An affirmation that is disputed by no-package manufacturers.

Conclusion: Join the conversation

If you have comments on this, join us in the group. We would be happy to have your feedback. It’s an interesting discussion.

Sources: http://insideevs.com/; https://teslamotorsclub.com

“Taking into account the increasing challenges IGBT technology is facing, we are very pleased to present a package that answers the needs of our industry both for today and for the foreseeable future.”

Infineon Technologies AG today announced the launch of two new power module platforms designed to improve the performance of high-voltage IGBTs in voltage classes from 1200V up to 6.5kV. To make the benefits of the new module broadly available, Infineon is offering a royalty-free license of the design to all providers of IGBT power modules. First products using the platform concept will include the high voltage classes 3.3kV (450A%

Mitsubishi Electric corporation announced today it will begin work on the development of standardized-package high-power semiconductor modules for use in heavy industry, including traction and electric-power applications, aiming to offer an optimal design for energy savings and high efficiency in high-power electronics systems.

Design Concept of High-Power Module with New Package

  • Common package design for modules of up to 6.5kV rating
  • Simple, easy parallel connection realizes various current ratings
  • Package compatibility with products of Infineon Technologies AG (Germany)

 

The first products to incorporate the new design platform will be for the high-voltage classes 3.3kV (450A), 4.5kV (400A) and 6.5kV (275 A). The standardized package will measure 100mm x 140mm x 40mm.

High-power modules are key devices used in power systems of between several kW and several MW. High-current modules with maximum ratings of 6.5 kV exist already. The industrial power market requires diverse modules suited to various current and voltage ratings according to each system’s power-conversion capacity. Products with compatible package dimensions from multiple manufacturing sources are also in demand.

Mitsubishi Electric intends to satisfy these market demands with its new high-power modules. Further details will be introduced in power electronics-related exhibitions, such as TECHNO-FRONTIER in Japan and Power Conversion Intelligent Motion Europe in Germany, both of which will be held in May 2015.