Dark Horse Enters MEMS Market

A dark horse in the optical components industry, Megasense Inc., yesterday announced that it had completed the construction of its MEMS (micro-electro-mechanical system) wafer fabrication facility (see MegaSense Makes Some).

Megasense was founded in late 1999 and has kept a low profile since then, starting off with plans to mass-manufacture MEMS-based sensors (hence its name) and then moving on to the idea of making telecom components instead.

A case of a solution looking for a problem? It's tough to tell, because Megasense won't say what type of components it's making, but there are least a couple of reasons why the company deserves to be taken seriously.

First, the folk running Megasense have some impressive credentials. They include Vladimir Vaganov, CEO, who was chief scientist at IC Sensors, the first company to mass produce MEMS based accelerometers. They also include Jim Northington, COO, a former VP of manufacturing operations at E-TEK Dynamics, now part of JDS Uniphase Inc. (Nasdaq: JDSU; Toronto: JDU); and Derek T. Harrar, VP of business development, formerly a VP with Morgan Stanley Dean Witter & Co.

Vaganov says the secret of success with MEMS components is a "systems approach" to manufacturing. This means designing components so they can be made in large batches using automated equipment, in a way that minimizes any work that has to be carried out on individual components.

Using this approach enabled IC Sensors to churn out millions of MEMS-based accelerometers at a cost of between $5 and $8 each, according to Vaganov, who's brought a bunch of his IC Sensors colleagues with him to Megasense. If anything, the telecom components being made by Megasense are less complex and thus easier to make, Vaganov adds.

Vaganov isn't expecting to make such massive numbers of telecom components but says that he can achieve "more than a tenfold reduction" in prices. Components similar to the ones that Megasense will make currently cost about $5,000, Vaganov says. He expects to bring the price down immediately to $3,000 with low volumes, then get it down to $1,000 with "not very high volumes" and eventually get the price down to $100.

The other reason why Megasense should be taken seriously is that it's got some smart money behind it. It raised $10 million last February from Bay Partners and Texas Pacific Group, and Gordon Stitt, chairman and president of Extreme Networks Inc. (Nasdaq: EXTR), sits on the board.

"We'll probably start looking for a second round later this year," says Vaganov. He says Megasense will break even after that, and won't require a third round.

— Peter Heywood, Founding Editor, Light Reading
Page 1 / 2   >   >>
iamnoone 12/4/2012 | 11:02:48 PM
re: Dark Horse Enters MEMS Market I'm always curious why newcomers to MEMS think that they can reduce the cost of optical MEMS components and systems by a lot. They assume that most of the cost is in the MEMS chip itself, which shows a lack of understanding of where the costs really add up -- in the optical alignment and packaging....
redface 12/4/2012 | 11:02:44 PM
re: Dark Horse Enters MEMS Market I believe this company is different from other MEMS companies who claim drastic cost reductions. MegaSense does have stuff to potentially make it happen.
Didham 12/4/2012 | 11:02:43 PM
re: Dark Horse Enters MEMS Market Agree with your comments!

A second issue - MEMS works with automotive products because the volumes run into millions. Reduce those volumes by 10 or 100 and your fab costs (dominated by capital and customised to meet the specific needs of opto-MEMS) will increase in near-exact proportion.
Peter Heywood 12/4/2012 | 11:02:37 PM
re: Dark Horse Enters MEMS Market I should have explained things better in the story.

Megasense has a pie chart that shows exactly what you say. It lists the five technologies used in making any MEMS-based photonic component and shows how much each would contribute to the overall cost:

1. MEMS (maybe 5%)
2. ASIC (maybe 5%)
3. Optics (maybe30%)
4. Testing (maybe 30%)
5. Packaging and assembling (maybe 30%)

The percentages are my own guesses - no figures on the pie chart.

Anyhow, the whole point about Vladimir's "systems approach" is that he's developed a way of designing the component and the manufacturing process at the same time, that a lot of the costs of the optics, testing and packaging are "redistributed" into the MEMS and ASIC portions of this pie chart.

For example, rather than making a MEMS mirror structure and attaching a single microlens to it, Megasence makes an array of mirrors, an array of microlenses and an array of pigtails and then bonds them altogether in an automated plant. The Megasense slices and dices the whole thing to get lots of little ready made subassemblies.

Ditto with testing. He tests a whole bunch of partially made components that are still in a wafer.

Ditto with packaging.

Obviously, these folk are far from being "new" to MEMS. They were some of the original pioneers in this field.

Hope this helps

iamnoone 12/4/2012 | 11:02:35 PM
re: Dark Horse Enters MEMS Market >>>
Hope this helps


It certainly does! In the present environment when moving mirrors alone don't count for anything, working on ways to minimize the total cost seems like a good pitch. It'll be interesting to follow their progress.

Obviously, these folk are far from being "new" to MEMS. They were some of the original pioneers in this field.

I made a mistake in saying they were newcomers to MEMS. What I actually meant was that they were newcomers to optical MEMS. Almost everyone in optical MEMS has a long history in MEMS sensors and other MEMS products.

lame_duck 12/4/2012 | 11:02:32 PM
re: Dark Horse Enters MEMS Market ICSensors, now that takes one back doesn't it. Sold to the now defunct EG&G in 1994 for ~$20M with the promise of going bigtime in the accelerometer for airbag biz. Went on to be highlighted in EG&G annual reports initially for more hype in the silicon MEMS market and then a year or so later as the big hitter for EG&G poor profit performance. Eventually EG&G sells out to Perkin Elmer and ICSensors scrambles for some traction in telecom components and biotech applications. Now the erstwhile MEMS heavyweight (bought from Perkin Elmer by some low profile outfit) is making pressure sensors for bathroom scales. Makes me wonder whether LR's talking up of the MEMS pedigree of MegaSense is of any value. Sounds to me like this silicon accelerometer business crashed and burned. If they really had the product and manufacturing expertise to pull this off for automotive (let alone telecom) why did it evaporate?
optobozo 12/4/2012 | 11:02:32 PM
re: Dark Horse Enters MEMS Market I remember reading about these folks in WIRED some time back. I hope they avoid the hype curse of some of the other WIRED feature stories (like Caspian) and actually deliver. We need some bona fide success stories!

On a side note, you can trace the success and failure of the .com era by looking at the thickness of WIRED issues over time. I guess the ad revenue isn't what it used to be.
Ames 12/4/2012 | 11:02:31 PM
re: Dark Horse Enters MEMS Market Peter
Let me pose a representative question (to give everybody the feeling of what is involved here).
At what stage are they final-testing the micromirrors (what we Fab guys call "Sort").
For example, if they do not do it at the wafer level, then it will translate to the cost since then they will have to test and sort chip by chip. Currently, everybody tests and sorts the MEMS chips after they are diced (singulated into single chips while still unreleased, under the silica glass) since the micromirrors can not whitstand the saw operation (harsh vibrations, water spray - will induce stiction). Then they must be released by (BHF oxide etch) and dried one by one. Then they are tested and sorted on by one. This affects the yield (60% die yield is considered high for MEMS micromirrors) and moreover thc test&sort economy.
Please educate us regarding this simple question/example.
And last but not least, if they follow the 3D MEMS path then the major cost factors are as mentioned in the optics, packaging, and feedback electronics (Lucent's LambdaRouter is several K$ per port)
Ames 12/4/2012 | 11:02:31 PM
re: Dark Horse Enters MEMS Market This coverage of LR says nothing. That makes me wonder what was the purpose of it. In any case, to get into MEMS now is pretty late unless you have some unique concept. There are about 60 start-ups (3 and 4 years old) trying to get MEMS micro-mirrors to work and we do not see it. Some companies like XROS/Nortel (does anybody remember XROS boasting for 1152x1152 ports at OFC2000?), OMM, Onix and other do not really make it big time.
Apart from that, batch processing is really no new theme. Any company that manufactures MEMS uses batch processing. LR, you should know better, this was supposed to be of the advantages of MEMS with respect to other technologies (e.g. Lithium Niobate etc.). Really, this is over used.
It sounds to me that MegaSense is thinking of raising money and this is the reason for this PR coverage.
In any case, those of you who are looking for MEMS as the solution for Photonic Switching will get disappointed.
I know, I've been there.
Ames 12/4/2012 | 11:02:19 PM
re: Dark Horse Enters MEMS Market Another late entrant. There are about 60 MEMS start-ups.
Surely you haven't heard about all of them .
BTW, 3dB insertion loss for a 4x4 switch is bad.
In any case, all the MEMS start-ups that plan to survive by selling 16x16 ports or less won't make it.
Lithium Niobate (electro-optic) and Silica (thermo-optic) switches will kill them very soon. They are more reliable and will match the micromirrors performance (especially the silica glass switches).
Page 1 / 2   >   >>
Sign In