Bronco Webinar 800x100 1

How crowdfunding can help save Silicon Valley from its harebrained investors

How crowdfunding can help save Silicon Valley from its harebrained investors
by Vivek Wadhwa on 06-10-2016 at 7:00 am

There are fears that another Ice Age is about to hit Silicon Valley because of the implosion of its unicorns — start-ups valued at more than one billion dollars. By one estimate there were 229 such companies in January of this year. Their valuations are dropping precipitously because they were overpriced and overhyped. The fear is that venture capital will dry up and hurt the innovation ecosystem.

In previous eras, such a setback to venture capitalists would surely have had a chilling effect on the innovation ecosystem because startups were dependent on their funding. But in today’s era of exponential technologies, there will hardly be a blip.

To start with, the cost of building new technologies has dropped so significantly that inventors no longer need venture capital. The desktop computers, server farms, racks of hard disks, and enterprise software that were needed would cost hundreds of thousands, sometimes millions, of dollars. Today, there is on-demand computing and cloud storage — which can be purchased for almost nothing from companies such as Amazon, Google, and Microsoft. And tools such as sensors and 3D printers, which are needed for building sophisticated medical devices and robots, are inexpensive. What costs the most in Silicon Valley is rent and food. But you can share an apartment and live on pizza and ramen noodles.

And instead of begging venture capitalists, angel investors, or friends for the $50,000-100,000 that it typically costs to start a technology company, founders can go directly to the people they are building their products for. They can post a video of a heart-felt pitch and demonstrate a prototype of their ideas on sites such as Indiegogo, Kickstarter and Plum Alley. If they get funded they’ll know they have a good idea; otherwise is time to go back to the drawing board and come up with something better.

The crowd makes better decisions than venture capitalists do. With crowdfunding, there is direct feedback from the market and a strong connection between the inventor and the funder. The community of funders feels a sense of ownership for the product and helps spread the word. And there is no filter such as a venture capitalist who has his own race and gender biases and only invests in the same trendy technologies as other VC firms.

The failure rate of crowdfunded projects is remarkably low. Three quarters of venture capital investments fail to return investor capital. Yet only 9 percent of crowdfunded projects fail to deliver on what they promised, according to Ethan Mollick of University of Pennsylvania –who researched 47,188 Kickstarter projects.

When entrepreneurs take money from venture capitalists, they know that this is coming from deep pockets and is just a financial investment. When dealing directly with customers it is personal; so entrepreneurs put in extraordinary effort and spend their own money to fulfill their promises. This is what leads to better outcomes.

One of the best examples of a technology that would not have seen the light of day without crowdfunding is virtual reality. As Mollick explained, this was largely ignored by traditional funders after it failed to gain traction in the 1990s. In 2012, a 19-year old Palmer Luckey, who had built a prototype of a virtual reality headset in his parent’s garage, launched a Kickstarter campaign for a commercial product. His goal was to raise $250,000 but there was so much demand that he ended up getting $2.4 million in orders. The product he later developed, Oculus Rift, was acquired by Facebook in 2014 for $2 billion. This set off a frenzy of funding by venture capitalists and greatly accelerated the progress of a world-changing technology.

So far, there have been limits to what start-ups could offer the crowd. They could only pre-sell their product and offer perks such as T-shirts and badges. This is about to change.

Starting May 16, the Securities and Exchange Commission is rolling out a new program that will allow private companies to use crowdfunding to sell securities — up to $1 million over a 12-month period. This was a provision of the 2012 Jumpstart Our Business Startups Act (or JOBS Act) to assist small companies with capital formation.

Individual investors with less than $100,000 of net worth will be allowed to invest the lesser of $2,000 or five percent of their annual income or net worth. Wealthier individuals can invest up to 10 percent. The investment must, however, be through an authorized funding portal. These portals are required to vet the companies and let investors shop among offerings and discuss them online. They can’t offer investment advice, make recommendations, or solicit purchases.

One of the first funding portals, Crowdfunder didn’t waste time in taking advantage of the new rules. It recently announced a VC Index Fund which offers an investment in a portfolio of hundreds of venture-capital backed startups. Crowdfunder chief executive Chance Barrett said his goal was to “allow everyday people to invest online alongside the world’s leading venture capitalists, while targeting a fund 10x more diverse, in the number of investments, than a traditional VC.” In other words, the public can become “super VCs.”

It remains to be seen if equity crowdfunding achieves the same success as product crowdfunding. The stakes are now higher and the risks of fraud are much greater. But one thing is certain: the balance of power is rapidly shifting — from venture capitalists to entrepreneurs. This is a good thing because it will lead to a greater diversity of start-ups. And with a bit of luck, there will also be fewer over-priced unicorns and less wastage of investment capital — because the venture capitalists will follow the crowd.

For more, follow me on Twitter: @wadhwa and visit my website: www.wadhwa.com


Facebook and the Internet of Things

Facebook and the Internet of Things
by Sudeep Kanjilal on 06-09-2016 at 4:00 pm

Something very important happened recently at the annual developer conference (F8). Facebook firmly staked its claim on IOT. Facebook events (like the Google annual developer events) are always interesting, as they give a tantalizing view of what is coming next. Yes, it lacks the panache the Apple events have. However, just because Facebook and Google events are deeply technical, does not mean they are not momentous, or exciting.

Several interesting data points were shared, several important consumer features announcements were made. I will not repeat them here, expect perhaps point out an important point – F8 was perhaps the pivot where Facebook finally shed its web legacy and became a native mobile ecosystem player.

Russian Nested Dolls – Stack within Stack within Stack

What, however, caught my attention was something that was not consumer facing – something regarding Parse, a mobile app backend / platform-as-a-service firm that Facebook acquired 3 years ago and then seemed to have forgotten about it.

At the level of the consumer internet, it’s been clear for some time that Apple and Google won the platform war. That leaves other consumer-facing/consumer-service firms in an interesting predicament – how do they survive, and thrive, on someone else’s platform? How far can they capture attention and intent of the consumers, what other interaction models will emerge, and so on.

The smart answer is – build your own stack inside someone else stack/platform!

That’s what Facebook and Amazon have been up to, for the past couple of years – moving up and down the stack simultaneously.

Moving up, these firms are building a new run-time overlaying the runtime embedded in the iOS and Android. Zuckerberg pointed out at F8 that five years ago, most content on Facebook was text. Now it’s photos, soon it will be video, and eventually it will be immersive content like virtual reality and augmented reality. If the content sharing of the future requires a headset, Facebook needed one, so it acquired Oculus.

Moving down, Facebook acquired Parse

Facebook wants to manage all interactions – between people, between people and things, and between things
We’re rushing headfirst into this era of “Internet of Things” — a time of connected coffee makers, connected fridges, connected light switches. There’s been very little done, however, in the way of standardizing how these things work (and work with each other) behind the scenes.

And that’s where Parse related announcement comes into play. Parse launched SDKs that act as the backend brains for IoT projects. It’s compatible with Arduino first, with other platforms on the way.

So, Facebook, which was basically an app (the top consumer layer) on the iOS and Android based mobile ecosystem/stack, expanded that into a full 5-layer stack. Third party consumers apps as the top layer, Occulus as the runtime/second layer, facebook itself, along with Messenger as the service/third layer (which will include payments!), Parse as the infrastructure/fourth layer.

What it means in layman terms is that Facebook could very well become the central control for Smart Home, Smart Auto, Smart Health, etc. Basically, on the day it completed the pivot from web to mobile, it also took the first step towards the next ecosystem – IOT!


How to prevent Electrical Overstress failure in NFC interfaces

How to prevent Electrical Overstress failure in NFC interfaces
by bkeppens on 06-09-2016 at 12:00 pm

Last year, about 40% of new smartphones included Near Field Communication (NFC). Analysts predict that by 2017 there will be 1 billion NFC enabled phones. Clearly, the use of NFC is ramping up because it can simplify aspects as diverse as communication, secure payments, user authentication, and retail loyalty programs for instance.


A typical, simplified NFC approach: Near field (max. 10m) communication between 2 devices.

Adding NFC functionality to an integrated circuit involves connecting the wireless interface pins to an antenna/coil. The voltage on those pads strongly depend on the distance between and alignment of transmit/read devices and the power of the transmitting device. Simulations by one of our customers showed a worst case of almost 10V between pads, which is rather high for advanced CMOS technology.

There are basically 2 ways to cope with this excess voltage.

Use high voltage transistors
Designers can use transistors that tolerate voltage up to 10V. Fortunately, many NFC applications also require on-chip non-volatile memory (NVM). Such NVM circuits typically use a thicker gate oxide and foundries provide high voltage transistors for such embedded flash process flavors. Many companies use those NVM transistors for the first stage of the antenna circuit. The problem is that these high voltage transistors are easily damaged during electrostatic discharge (ESD) stress. A parallel on-chip ESD protection circuit (with ~10V tolerance) is the easiest solution.


Because high voltage (memory) transistors are typically rather weak during ESD stress it may be best to use a local ESD protection clamp in parallel.

Limit the voltage
It is possible to reduce the excess voltage with a so-called clipping or limiting circuit. The simple approach is to use a set of diodes.


Simple clipping variations based on diodes.

Other options include clipping or limiting circuits based on transistors. Check out this case study with a clipping circuit.


Semiconductor market going negative?

Semiconductor market going negative?
by Bill Jewell on 06-09-2016 at 7:00 am

Several recent forecasts for the 2016 semiconductor market point to a decline. The title of this post is the same as we used for in October 2015. In October, Semiconductor Intelligence projected the market would grow 1.0% in 2015 despite several predictions of the market going negative. 2015 finished with a slight 0.2% decline. The chart below shows four forecasts in the last two months calling for a negative 2016, ranging from a 0.6% decline from Gartner to a 2.3% decline from IDC. Databeans projects 3.0% growth. We at Semiconductor Intelligence are calling for 1.0% growth in 2016. For 2017 semiconductor market growth, Mike Cowan’s May forecast was 6.1%. We at Semiconductor Intelligence are projecting 7.5% growth.


What is behind the weak 2016 outlook? According to World Semiconductor Trade Statistics (WSTS), the 1[SUP]st[/SUP] quarter 2016 semiconductor market declined 5.5% from 4Q 2015, the weakest first quarter since the recession in 2009. However, there are signs of improvement in 2Q 2016. The table below shows 1Q 2016 revenue growth versus 4Q 2016 and guidance for 2Q 2016 for the largest semiconductor suppliers.

[TABLE] align=”center” style=”width: 500px”
|-
| colspan=”4″ style=”width: 200px; height: 34px; text-align: center” | Key Semiconductor Company Revenue
|-
| colspan=”4″ style=”width: 200px; height: 21px; text-align: center” | Change versus prior quarter in local currency
|-
| style=”width: 50px; height: 21px” |
| style=”width: 10px; height: 21px” | Reported
| style=”width: 10px; height: 21px” | Guidance
| style=”width: 100px; height: 21px” |
|-
| style=”width: 50px; height: 21px” | Company
| style=”width: 10px; height: 21px” | 1Q16
| style=”width: 10px; height: 21px” | 2Q16
| style=”width: 100px; height: 21px” | Comments on 2Q
|-
| style=”width: 50px; height: 21px” | Intel
| style=”width: 10px; height: 21px” | -8.1%
| style=”width: 10px; height: 21px” | -1.5%
| style=”width: 100px; height: 21px” | high end +2.2%
|-
| style=”width: 50px; height: 21px” | Samsung
| style=”width: 10px; height: 21px” | -15.6%
| style=”width: 10px; height: 21px” | n/a
| style=”width: 100px; height: 21px” | solid DRAM & SSD demand
|-
| style=”width: 50px; height: 21px” | Qualcomm
| style=”width: 10px; height: 21px” | -18.5%
| style=”width: 10px; height: 21px” | 0.9%
| style=”width: 100px; height: 21px” | high end +8.1%
|-
| style=”width: 50px; height: 21px” | SK Hynix
| style=”width: 10px; height: 21px” | -17.2%
| style=”width: 10px; height: 21px” | n/a
| style=”width: 100px; height: 21px” | solid DRAM & SSD demand
|-
| style=”width: 50px; height: 21px” | Micron
| style=”width: 10px; height: 21px” | -12.4%
| style=”width: 10px; height: 21px” | 0.5%
| style=”width: 100px; height: 21px” | high end +5.7%
|-
| style=”width: 50px; height: 21px” | TI
| style=”width: 10px; height: 21px” | -5.7%
| style=”width: 10px; height: 21px” | 5.4%
| style=”width: 100px; height: 21px” | high end +10.8%
|-
| style=”width: 50px; height: 21px” | Toshiba
| style=”width: 10px; height: 21px” | 0.3%
| style=”width: 10px; height: 21px” | n/a
| style=”width: 100px; height: 21px” |
|-
| style=”width: 50px; height: 21px” | NXP
| style=”width: 10px; height: 21px” | 38%
| style=”width: 10px; height: 21px” | 5.4%
| style=”width: 100px; height: 21px” | 1Q includes Freescale
|-
| style=”width: 50px; height: 21px” | Infineon
| style=”width: 10px; height: 21px” | 1.0%
| style=”width: 10px; height: 21px” | 2.0%
| style=”width: 100px; height: 21px” | high end +4%
|-
| style=”width: 50px; height: 21px” | MediaTek
| style=”width: 10px; height: 21px” | -9.4%
| style=”width: 10px; height: 21px” | 28.0%
| style=”width: 100px; height: 21px” | China handset demand
|-
| style=”width: 50px; height: 21px” | ST
| style=”width: 10px; height: 21px” | -6.5%
| style=”width: 10px; height: 21px” | 5.5%
| style=”width: 100px; height: 21px” | high end +9%
|-
| style=”width: 50px; height: 21px” | Renesas
| style=”width: 10px; height: 21px” | 1.8%
| style=”width: 10px; height: 21px” | n/a
| style=”width: 100px; height: 21px” | uncertain after earthquake
|-

Q1 2016 revenues declined from 4Q 2015 for most companies, including double-digit declines for the Memory companies (Samsung, SK Hynix and Micron) and Qualcomm. NXP’s 38% growth is a result of its acquisition of Freescale. The outlook is fairly strong for most companies in 2Q 2016. All but Intel are projecting growth, with Intel’s high end guidance for 2.2% growth. The high end guidance for Qualcomm, TI and ST is in the 8% to 10% range. MediaTek is guiding 28% growth in 2Q 2016 based on strong handset demand from China. Broadcom is not included since it is in the process of integrating its merger with Avago.

The outlook for key end equipment is major factor for the weak semiconductor market in 2016. Gartner’s March 2016 forecast for PC plus tablet units is a 2.5% decline in 2016, following a 10% decline in 2015. Growth is projected at a positive 2.5% in 2017. Mobile phone units are projected to show growth in only the 1% to 2% range. The International Monetary Fund’s (IMF) April 2016 forecast is 3.2% global GDP growth in 2016, only slightly higher than 3.1% in 2015. GDP is expected to pick up to 3.5% growth in 2017.

[TABLE] border=”1″
|-
| colspan=”5″ style=”width: 714px; text-align: center” | Forecast Annual Change
|-
| style=”width: 146px” |
| style=”width: 143px” | 2015
| style=”width: 143px” | 2016
| style=”width: 143px” | 2017
| style=”width: 137px” | Source
|-
| style=”width: 146px” | PC + Tablet Units
| style=”width: 143px” | -10%
| style=”width: 143px” | -2.5%
| style=”width: 143px” | 2.5%
| style=”width: 137px” | Gartner, March
|-
| style=”width: 146px” | Mobile Phone Units
| style=”width: 143px” | 2.0%
| style=”width: 143px” | 1.4%
| style=”width: 143px” | 2.1%
| style=”width: 137px” | Gartner, March
|-
| style=”width: 146px” | Global GDP
| style=”width: 143px” | 3.1%
| style=”width: 143px” | 3.2%
| style=”width: 143px” | 3.5%
| style=”width: 137px” | IMF, April
|-

Our Semiconductor Intelligence forecast of 1.0% semiconductor market growth in 2016 takes into account the weak end markets. However, the 2Q 2016 outlook for major suppliers is encouraging. The growth acceleration for end equipment and GDP in 2017 leads to our forecast of 7.5% growth in 2017.


NVIDIA GeForce GTX 1080 Poised To Claim The Gaming And VR Performance Crown

NVIDIA GeForce GTX 1080 Poised To Claim The Gaming And VR Performance Crown
by Patrick Moorhead on 06-08-2016 at 12:00 pm

NVIDIA has been teasing the promise of Pascal for years, but this year Pascal’s performance became much more real. With the announcement of the Tesla P100 based on Pascal, NVIDIA was able to show the neural-network world what Pascal was really capable of. However, that chip is very different from the gaming-focused Pascal chip that NVIDIA launched on May 6th in Austin at an event VR analyst Anshel Sag and I both attended. NVIDIA introduced a plethora of new features and improvements that enable their new Pascal-based GeForce GTX cards to really shine against the competition. NVIDIA’s own Founder and CEO, Jen-Hsun Huang, deemed the GTX “1080 is the new King.”


NVIDIA CEO Jen-Hsun Huang at the GeForce GTX 1080 launch event held in Austin, TX (Photo Credit: Patrick Moorhead)

GTX 1080 “firsts”
The Pascal GPU architecture is NVIDIA’s most ambitious ever. The biggest version of the Pascal family is the GP100 inside of the Tesla P100 high performance computing (HPC) GPU. That version of Pascal features 15.3 billion transistors, the most of any chip, ever created by anyone, ever. The newly announced GeForce GTX 1080 features the GP104 graphics chip with fewer transistors paired with 8GB of Micron GDDR5X, a brand new type of graphics memory. The improvements in memory bandwidth and latency mirror NVIDIA’s improvements in the Tesla P100 with the use of HBM2. However, while AMD didintroduce HBM in their Fiji family of GPUs, it was limited to 4GB of VRAM capacity and this new generation that NVIDIA is using does not have such a limitation. Additionally, AMD is not using GDDR5X anywhere in any of their cards, so NVIDIA is the first to introduce this technology to the market in a product.

16nm FinFET helping
The biggest reason why NVIDIA is able to cram so many transistors inside of a relatively small chip is thanks to the new 16nm FinFET process from TSMC. This process allows for NVIDIA to not only build a small chip but to pack it with plenty of performance and to do it at a very low wattage. In fact, NVIDIA claims to be cramming approximately 29% more performance out of the GeForce GTX 1080 than they do from their most powerful graphics card right now, the TITAN X. Not only is NVIDIA claiming to be about 29% faster than the previous generation, they are able to accomplish this with only 180 watts of power. This represents a power reduction of 28% while at the same time delivering a 29% boost in performance. Traditionally, GPU manufacturers have had to give up some power savings for additional performance or vice versa, but with the new Pascal architecture and 16nm FinFET from TSMC you get the best of both worlds.

VR-focused performance and features
The GeForce GTX 1080 delivers even more performance once you start looking at performance in Virtual Reality (VR). It appears the entire GTX 1080 and Pascal were specifically designed with VR in mind. This includes support for hardware asynchronous computing as well as some of their VR Works software features like simultaneous multi-projection (SMP). NVIDIA claims that with the added hardware and software features on the GeForce GTX 1080 actually bring its VR performance to approximately two times that of the TITAN X, NVIDIA’s next fastest graphics card.

However, in order to accomplish such performance, game developers have to implement NVIDIA’s Game Works proprietary SDK which includes their VR Works that enables the use of features like SMP. As many middleware programs go, the chance that it will be broadly adopted is to be determined, but there’s a good chance that there will be a handful of games that do and those that do could benefit from this performance boost. This is the promise of NVIDIA’s new technology inside of Pascal and the software that accompanies it. Plus, the VR industry is looking for anythingto boost the category, so I expect developers and game engines will support both SMP and Async. Already, VRWorks is integrated into game engines today with features like VR SLI supported.

Hardware Async compute
Prior to Pascal, NVIDIA’s Maxwell architecture supported asynchronous compute in hardware. Improvements to work scheduling in Pascal increases the overall performance in VR and improves the overall user experience allowing both compute and graphics functions to happen as they need to. This feature is extremely important in order to deliver a smooth VR experience and I suspect that NVIDIA knows this and has made sure their architecture supports it. NVIDIA prides themselves in the best possible gaming experience, so it only seems logical that they would want to deliver the best VR experience possible with the GTX 1080.

Exceptional pricing for high-end card
In addition to all of the features in the GeForce GTX 1080 that make it so VR friendly and power-friendly, NVIDIA’s new GPU is also extremely price friendly. The “Founder’s Edition” from NVIDIA comes in at a relatively low $699, which is a solid $400 less than their current top-end card the TITAN X, for more performance. NVIDIA also introduced the Founder’s Edition GTX 1070 that will be $379, which claims approximately TITAN X performance for about 1/3 the price. These new GPUs from NVIDIA deliver on performance, power and price, they are a complete win for virtually any gamer looking to spend more than $300 on a GPU. They will also deliver the best possible performance today, even if NVIDIA’s claims are overstated.

NVIDIA should continue to dominate over $300
NVIDIA is trying to capture the entire segment of the market above $300 that represents the “enthusiast segment”, which is about 20% of the overall GPU unit market volume and where the highest margins and revenue are. The biggest market segment is the “performance” segment which is below $300 and while it represents the most units sold this is where margins are thinner in this segment.
Unless something unforeseen happens, which I don’t think will happen, I believe NVIDIA should continue to dominate in this high ASP, lower units market above $300.

The real fight will be from $149- $299
GPU’s that are more than $150 and less than $300 are the sweet spot for unit volume and unit market share which is why NVIDIA will likely need to extend the GTX 1000 family down to something like a GTX “1060” soon. This will be especially important for NVIDIA because Advanced Micro Device’s upcoming Polaris GPU will likely be squarely aimed at this price segment in order to recapture unit market share and potentially profits. After all, this is Advanced Micro Device’s self-proclaimed “year of graphics”.
If NVIDIA can successfully launch and price a GTX “1060” within this price segment they can defend their market share from Advanced Micro Devices in this price point, even if they effectively own the high-end. NVIDIA’s market share position is strong right now, but they already have a pretty strong command of the high-end and that’s why it’s important that they can launch into the lower price segments as well.
While AMD hasn’t yet launched Polaris, I believe it’s going to be an absolute fight from $149 to $299, so get ready for it.

Wrapping up
The NVIDIA GeForce GTX 1080 and GTX 1070 are absolutely marvels of engineering from a company that knows what gamers want and who has been executing incredibly for the last few years. The company are delivering on the promise of Pascal and 16nm FinFET and they are even doing all of this while bringing down the price of their GPUs compared to the previous generation.

These new cards are important for the PC VR category as they radically improve the price-experience curve. While HTC is delivering and Oculus has been choking, at least we know that higher-end VR will look much better from a graphics perspective.

NVIDIA’s launch of the GTX 1080 has been extremely well received by gamers around the world and those who wish to see high performance and lower cost VR solutions. NVIDIA has without a doubt introduced what appears to be the fastest GPU for VR and it will be interesting to see how it actually stacks up in real world scenarios.

Thankfully, NVIDIA has provided us with a GTX 1080 in order for us to test their claims ourselves. NVIDIA has finally, fully committed themselves to VR and they don’t appear to be stopping any time soon, I expect to see more announcements in the VR space from NVIDIA as they continue to push Pascal through their different product lines.

More from Moor Insights and Strategy


Highlights of the 22nm FD-SOI San Jose Presentations

Highlights of the 22nm FD-SOI San Jose Presentations
by Adele Hars on 06-08-2016 at 7:00 am

VLSIResearch FDSOI whybiztech SanJose16 534x610

This is part 2 (of 2) of my coverage of the recent FD-SOI Symposium in San Jose (April 2016), this time looking at the 22nm presentations by GlobalFoundries, ARM (finally!!), VLSI Research Inc and Sigma Designs. (Part 1 looked at the 28nm presentations.) Most are now available on the SOI Consortium website – click here to see the full list.
Continue reading “Highlights of the 22nm FD-SOI San Jose Presentations”


IMEC Technology Forum (ITF) – Secrets of Semiconductor Scaling

IMEC Technology Forum (ITF) – Secrets of Semiconductor Scaling
by Scotten Jones on 06-07-2016 at 4:00 pm

IMEC is a technology research center located in Belgium that is one of the premier semiconductor research centers in the world today. The IMEC Technology Forum (ITF) is a two-day event attended by approximately 1,000 people to showcase the work done by IMEC and their partners.
Continue reading “IMEC Technology Forum (ITF) – Secrets of Semiconductor Scaling”


Is GaN Disruptive? Revisiting the Criteria

Is GaN Disruptive? Revisiting the Criteria
by Alex Lidow on 06-07-2016 at 12:00 pm

In March 2010 Efficient Power Conversion (EPC) proudly launched our GaN technology at the CIPS conference in Nuremberg, Germany. Parts and development kits were readily available off-the shelf and therefore designers could immediately get started with a new state-of-the-art semiconductor technology.
Continue reading “Is GaN Disruptive? Revisiting the Criteria”


The Guiding Light & Other Photonic Soaps

The Guiding Light & Other Photonic Soaps
by Mitch Heins on 06-07-2016 at 7:00 am

I’m a child of the sixties and seventies and on the occasion when I was sick and couldn’t go to school I got to experience the world of daytime TV soap-operas. Back then we only got 3 channels and it wasn’t until the late 60’s that we got color TV! I remember titles like “The Guiding Light”, “Secret Storm”, and “As The World Turns”. Forty plus years later, I am again re-living “The Guiding Light” except now it’s in the form of reading about “guiding-the-light” on silicon photonic integrated circuits (PICs). Like the daytime soaps, there seems to be a never ending cast of characters (photonic components) that are being presented. I thought it would be instructional to review one of the characters used to “guide-the-light” on a PIC.

Waveguides are the primary components used in photonic design to guide light. More than a conduit for light, they are building blocks from which other components are created to modulate, filter and switch light. From the book Silicon Photonics Design, section 1.4, silicon PIC waveguides are most commonly made from the active device layer of silicon-on-insulator (SOI) wafers (see fig 3.1– cross section of SOI wafer). Much research has gone into engineering the waveguide geometries. There are several types of waveguides, but the most commonly used are the strip (or rectangular) and rib waveguides (see fig 3.4). Strip waveguides are typically used for routing and low-loss tight curves while rib waveguides are often used to create electro-optic devices such as modulators as the rib allows for electrical connections to be made to the waveguide. And yes, you read that right. Photonics routing uses curves, not Manhattan style shapes and waveguides are typically single-layer designs as the silicon crystal of the waveguide core is grown, not deposited on the wafer.

One of the key metrics for photonics is signal loss. The signal intensity must be great enough to be sensed at the end of the signal’s journey and every piece of waveguide it traverses takes its toll. There are several contributors to waveguide loss including absorption due to metal in proximity, scattering and reflections due to sidewall roughness, material loss in active doped structures and surface-state absorption from improperly or un-passivated waveguides. One might think that the best waveguide would be one in which no loss is allowed at all. However, it is these loss mechanisms that actually enables the manipulation of light in the waveguide. Without them, waveguides would just be simple conduits or light wires. Fortunately for us this is not the case. Before going further, a couple of keys points should be noted about photonics and waveguides.

[LIST=1]

  • Light can be made of different wavelengths (colors) and we can encode different data or messages on these different wavelengths.
  • Light of different wavelengths (and therefore the messages encoded on those wavelengths) can simultaneously occupy the same space without interfering with each other.

    The implications of these two points are far reaching. In some cases, it is signal absorption in un-passivated waveguides that is used to make detectors. But that’s a topic for yet another article. For waveguide routing these points enable us to use what is referred to as wavelength division multiplexing (WDM). WDM lets us convert from the spatially parallel electric domain to a wavelength-parallel optical domain, significantly reducing the number of waveguides needed to transmit large amounts of data (see Fig 2.2 from the book Photonic Network-on-Chip Design). Imagine a 64-bit bus worth of data all traveling down 1 waveguide! Secondly, it is the radiation of light outside of the waveguide that lets us couple the light in one waveguide to other waveguides. By controlling the resonance points of these specialized resonant waveguides (micro-rings) we can switch one or multiple signals simultaneously. Moreover, we can also use this capability to filter one or more wavelengths from the main signal for additional processing or routing to specific sensors. This is the basis for building high performance, low power switching matrices suitable for switching wide parallel data such as required for CPU-to-memory applications.

    More recently another waveguide character has come onto the scene, that being Silicon Nitride (Si[SUB]3[/SUB]N[SUB]4[/SUB]). Si3N4 can also be used as a waveguide material and unlike crystalline Si, Si[SUB]3[/SUB]N[SUB]4[/SUB] can be grown onto the wafer which means it can be stacked, much like metal systems in electrical ICs. While photonics allows us to cross waveguides on the same layer without interference (see point 2 above), there is a loss penalty that is incurred at each crossing. Having the ability to stack waveguides with less lossy crossings enables switch-matrices that look like figure 5.53.

    Just like in the soaps of the 60’s, this new character, Si[SUB]3[/SUB]N[SUB]4,[/SUB]adds an entirely new twist to the photonics story line. It has already prompted an entirely new dialog around subjects such as photonic networks-on-chip, packet-switching vs circuit-switching networks and new terms like selective transmission, an interesting combination of electronic and photonic networking that uses some of both technologies. We are at the beginning of a truly interesting time for silicon photonics. The plot is starting to evolve and become richer as more new characters are added into the story line … and just like the daily viewers of those early TV soaps, I can hardly wait to see what will happen in the next episodes.