Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > Power/Alternative Energy
Power/Alternative Energy??

AnalogicTech stays 'fabless without foundries'

Posted: 21 May 2009 ?? ?Print Version ?Bookmark and Share

Keywords:analog market? power management? fabless?

Fabless strategy

What is your manufacturing strategy?
Our vision is to be fabless without foundries. What we did is we partnered with basically DRAM fabs. DRAM fabs have the highest volume, lowest cost and most advanced technology on earth. They only know make DRAM. They have to keep moving the curve and when they do, they have to vacate their last fab. Then, they have to backfill it with something. Normally, they have to backfill it with LCD line drivers and camera image sensors.

So we went to those fabs and took our own process technology. We say: 'Make that process for us and we'll show you how. And then you'll have all of our business or half of it as long as you use our technology and don't sell it to anyone else.' We have now gotten to the point where we can bring up a process without anybody knowing exactly what it does.

Who makes your product?
The two fabs we work most closely with are MagnaChip and Vanguard. We've also worked some with a China fab call CSMC. We have more fabs than we know what to do with it.

South Korea's MagnaChip is an IDM and foundry provider. Taiwan's Vanguard was in the DRAM business years ago, but it now focuses on the specialty foundry business. Are you interested in partnering with TSMC?
Absolutely not. Whatever I teach them, they will use against me.

In your fabs, you have developed a so-called ModularBCD process. Is that a generic BCD (bipolar, CMOS, DMOS) process?
It's a very unique way of doing BCD. BCD is preferred in high voltage and power. In analog, it gives you different combinations of transistors. Bipolar is good for precision analog. The CMOS is very good for digital, because you always have to have interface and CMOS is also good for low-power analog. The DMOS makes robust power devices.

The problem is that everybody's BCD process in the world evolved the B from the bipolar processes. And those old bipolar processes come from the 60s and 70s. And it's high-temperature with a layer of silicon called epitaxy grown on top of wafer. The thickness of the epi sets the voltage. If you change your mind and want to go from 40V to 60V, now you have to grow thicker epi. Once you do that, all the design rules change.

ST and TI have more BCD processes than anybody else. They also have like 25 flavors of different processes. If I had huge design teams, I can do that. Because we created a process that is modular, the 3-, 5-, 12-, 18-, 30- or 60V sections don't even know they are on the same silicon. There is no epitaxy. There no isolation diffusion. There is no high temperature processing. It's all done in low temperature. Because of that, the 5-, 12- or 30V is just a module. That means the circuits are also modules.

What node are you on?
Most of our competitors are at 0.5?m. And some of them are using 3?m. We're 0.35?m on the mainstream effort we're putting in now. The next node we will bring up is 0.18?m. We will skip the 0.25?m node. We don't think 0.25?m is worth it. 0.35?m has a huge installed capacity. 0.18?m has a huge installed capacity. The problem is 0.18?m is more expensive. If you look at the cost of the mask sets at 0.35?m, I don't have to consider them in my ROI. At 0.18?m, the glass cost is enough that I have to consider will I make my money back?

When will you roll out 0.18?m devices?
No earlier than next year. We're actually looking at our first design concepts right now. If it were to be in production at all next year, it would be a limited list of modules.

?First Page?Previous Page 1???2???3???4?Next Page?Last Page

Article Comments - AnalogicTech stays 'fabless without ...
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top