Part 3:  Accelerometer Development - The Analog Devices Story

Analog Devices was started by Ray Stata in 1965.  From 1965 to 1990, the company has been built into a strong business of semiconductor design and manufacturing. The company enjoyed double digit growth. The company was listed in the US stock exchange (Symbol=ADI). However, in 1990, the growth starts to slow, dropping to 10%. ADI, with its specialty in analog devices, was not in the computer chip segment.

In 1987, Analog Devices began to develop a plan for accelerometers. Richie Payne, then working for Analog Devices, starts to develop new business in new areas. They experimented in MEMS technology and began to see its business potentials. Initially, they focus on acceleration sensors for airbag triggering. The then commercial standard is ball-in-a-tube design. A MEMS option could promise reduce the size, reduce the cost, and increase reliability.

The company developed the MEMS technology with a lot of internal resources. Even though the MEMS technology and business model is similar to electronics circuits, the fabrication process is not entirely compatible. Both doing research and fabricating MEMS devices would take up resources the company has for making semiconductor circuits. Besides, there are a lot of uncertainties, including:

-         the automotive market is unfamiliar customer for ADI;

-         the business model require ADI to mass produce a lot of sensors with small variety;

-         manufacturing and quality control present significant challenges.

The automotive market is also very cost and safety conscious. Products that get sold into the automotive market must perform extraordinarily well for very low price. The MEMS fabrication is, on the other hand, capital intensive and costly. Therefore, only a market with large volume can satisfy the business model. However, it is difficult to navigate the growth period. It was difficult to borrow funds and facilities. It was difficult to convince auto makers to bet on a brand new technology. It was not sure how cheaply the company can eventually make it.  Therefore, the picture of profitability is not clear at the beginning.

ADI banked on the surface micromachining technology invented at UC Berkeley (by Roger Howe and Ph.D. advisor Richard Muller). The device is based on surface micromachining process using LPCVD polycrystal silicon as a structural layer and LPCVD oxide as a sacrificial layer. Both materials are commonly used in the semiconductor industry. The polysilicon surface micromachining process allows MEMS parts to be built on top of circuitry (although with a lot of detailed recipe adjustment).  

Between 1991-1994, the company sold first batch of sensors to Saab. Richie Payne and team has to propose to the auto makers an aggressively low per-unit price to start. ADI does not make money on the deal. This kind of deal would naturally cause concerns internally in the company, as MEMS division has to compete for resources from other larger, more profitable divisions. In the end, ADI did not make profits until at least 1998.

Despite all the difficulties, ADI should be thankful of its involvement with MEMS technology.  Between 2004-2007, ADI posted a 1% compounded annual growth rate in revenues, because its core business (handset baseband, modems inside phone) was becoming a commodity product.  In 2008, ADI sold the PC power-management business to ON Semiconductor (nasdaq: ONNN - news - people ) for $185 million, and the handset baseband business to MediaTek for $350 million. After this reshuffling, ADI is 90% analog circuitry and 10% MEMS. Now ADI looks set to deliver double-digit growth of 13% five-year CAGR.

The company had to solve many tough technical issues, including film stress, damping, and control algorithms.

Page<< 1 2 3 4 5 >>