VDC Research: Embedded Microprocessor, Board & Systems Market Blog

Design West / ESC Show Highlights - Part 2

In this blog I will continue to explore some of the VDC Embedded Hardware team experience at the Design West ESC show. We saw a lot of great product demonstrations along with some excellent detailed briefings and meetings so it’s difficult to boil it all down to a reasonable size blog but here we go:

AMD: We saw a number of embedded computer products from multiple manufactures that featured AMD processors. Many of these would be great for scalable edge node applications. Heard a bit more about the latest Opteron 3200 series of processors which will likely find many cloud based applications. While at AMD we visited partner Xi3 they have some really nifty looking cube type computers that can be deployed in array like structures. The concept they were showing was a datacenter on wheels.

Atmel: Was showing some new products that seemed really great for embedded M2M type connectivity but, according to the press material I received, the details are embargoed for another week or two.

Digi-International: Digi was a company we covered in the Migrating to the Embedded Cloud report that published this week so we really wanted to stop by and see if there was anything new going on.  What we saw didn’t disappoint as there was a lot of evidence about the partnerships we talk about in the report. Digi and Wind River were announcing a collaboration to deliver M2M wireless connectivity solutions using Intel processors. This is on the heels of a similar partnership that Digi has with Freescale. We saw that Digi was using another company’s embedded computer hardware products as part of the cloud connectivity demonstration but, as that partnership is not announced; I can’t write more about that now.

Integrated Device Technology (IDT): In this booth there was a very impressive demonstration of  serial RapidIO technology being deployed in a number of different companies’ products. This is very important in cellular 3G and 4G deployments. Despite being handled by different protocols, hardware and connection methods the data travelled end-to-end efficiently and, most importantly without being corrupted.

Imagination Technologies:  We saw some really great examples of their IP used in mobile devices and applications. As people become more ingrained with mobile devices, high resolution videos, and larger screen sizes, it takes some pretty complex systems on chip to make it work. The difficult thing is getting the needed performance while not sucking the mobile equipments battery dry.

Inside Secure: As the market for M2M is growing there needs to be ways to ensure of the identity of the machines and people being connected. Inside Secure gave us a briefing on several of their security technologies that can be embedded into products to address these issues.

Lantronix: As an OEM is making design decisions on new products or looking to update older ones adding wired and/or wireless connectivity can be a problem. Lantronix briefed us on several of their products where the connective capability can be added to new designs or even old ones on an as needed basis. Almost as a proof of concept, Lantronix produced xPrintServer using technology they usually sell to OEMs to allow Apple devices to directly connect to existing legacy printers using a downloadable app.

Microchip: The VDC Embedded SW and HW teams had several meetings with Microchip and we were particularly happy to have an opportunity for a great discussion their President and CEO Steve Sanghi. As this blog looks to be running a little long, I will give the special focus to topics we covered with Mr. Sanghi in a blog next week. The hardware team learned a lot about some of the new Microchip MCUs that are adding analog circuitry such as ADCs, DACs, Op-Amps, and Comparators.  This puts more functionality into a single package while, at the same time often reduces device pin count.

Micron: I saw a detailed briefing on the latest about the Micron memory cube product. The through hole vias on the semiconductor dies that make this design possible are interesting in themselves.

National Instruments: This was another company that is covered in the Embedded Cloud report and, we saw that the Compact Rio product has some new, even more compact, product lines extensions. In the booth there was also a mock-up of a Siemens smart grid transmission line breaker module. The N/I Compact Rio was part of the design in that it could capture and transmit events that happened on the transmission lines. One of the neat things is Siemens/NI project is that the breaker can be reset remotely.

Netronome: If you ever want to see a place where powerful embedded processors are used in large quantities in high volume applications, a network flow processor is a good place to look. These impressive units we saw inspect packets and move internet traffic at extremely high rates.

Power.org: An interesting talk with one of the Directors at the IBM booth to learn more about this organization that unifies standards among its members around the Power Architecture technology with a goal of making sure that processors and communications products work efficiently as the scale of connectivity grow ever increasingly higher.

Silex: We saw some product briefings on their connectivity modules. With respect to M2M connectivity this is pretty interesting if for example you are a product designer supporting a legacy product that you want to add M2M services to or, in other cases, you are worried that a particular standard fall out of favor, and you want the product you are designing to be future proof.

SuperMicro: They have a very large line of products and the MicroCloud product was particularly interesting to us because of the embedded cloud report where we had profiled SuperMicro.  The MicroCloud product impressed us with its ability to scale up as a cloud service and/or the amount of machines being supported in an edge node application grows.

Texas Instruments: TI had a lot to show us with all types of embedded hardware products adding GPS and motion sensing as well as Wi-Fi and other connectivity. Anyone that has taken a portable device with GPS applications into a building, large city, or tunnel will realize that these types of products have a waiting market.  We also got briefings on some new process intensive DSP products that are becoming increasingly important to many markets. This is one of the topics I will expand on in the next installment of this blog series.

Next week, I’ll give a few last high level takeaways about things we saw and discussed at the show.

ADLINK TECHNOLOGY ACHIEVES NUMBER TWO PC/104 MARKET SHARE THROUGH ACQUISITION

Approximately six weeks ago, ADLINK Technology, Inc. (New Taipei City, Taiwan) announced the 100% acquisition of LiPPERT Embedded Computers GmbH (Mannheim, Germany), providing a major increase in market share for modular computing products. (http://www.lippertembedded.de/en/information/news/312-adlink-acquires-lippert.html )

 ADLINK’s first major push into the modular computing segment was the firm’s 2008 acquisition of AMPRO Computers, Inc. (San Jose, CA), now doing business as AMPRO by ADLINK. AMPRO, historically, was the inventor and a leading supplier of PC/104 family modules, as well as a pioneer in the field of Computers-on-Modules (COMs). Indeed, AMPRO’s Encore COMs had won several awards, including the EDN Innovation of the Year in 2001, and Electronique’s Electron d’Or in 2002.

Since the 2008 acquisition, the combined firm’s market shares have been quite healthy. VDC estimates that, in 2011, ADLINK held a 6.25% market share in PC/104 Family modules (the number 5 position), a  3.01% share in COMs (the number 8 position), and a 13.69% share in “other stackable” CPU modules (EBX, EPIC), yielding the number three position.

LiPPERT is also a major player in the modular field, particularly in the area of PC/104 Family modules. The firm had an estimated 2011 market share of 3.41%, occupying the number 13 slot. The combined shares of ADLINK and LiPPERT (9.67%) would have given ADLINK the number two slot, second only to the Eurotech Group.

Although LiPPERT has also been active in both the COMs and “other stackables” arenas, the firm’s market shares are substantially smaller; the addition of LiPPERT’s shares to those of ADLINK would not have been sufficient to move the firm higher in rank.

However, both firms bring considerable across-the-board expertise to the table. VDC expects this combination to be quite formidable, particularly in the PC/104 Family field.    

Connect with VDC Research Group at the Design West/Embedded Systems Conference 2012!

VDC Research Group will be joining the Design West/Embedded Systems Conference 2012 exhibition and conference.  During the conference, we will be presenting the coveted VDC Embeddy awards to a deserving product in each of the 2 software and hardware categories. To make sure your product is considered, please make sure that:

• The product is formally announced at the show or, has been announced as of January, 2012
• That the VDC Research team will be briefed on the details of the product by your show staff.

VDC’s Embedded Hardware Team will be arriving March 27th and will be at the conference through March 29th.  During that time, we welcome the opportunity to connect with attending vendors.  We look forward to explaining VDC’s research methodology, learning about your latest product releases, and discussing your market research and strategic needs.

If you would like to learn more about the show, please click here.

If you would like to schedule a meeting around Embedded Hardware, please contact:

David Laing, Senior Analyst, Embedded Hardware & Systems Practice, VDC Research Group at: dlaing@vdcresearch.com or 508.653.9000 x146.

Or

Chris Rommel, Vice President, Embedded Hardware & Systems Practice, VDC Research Group at: crommel@vdcresearch.com  or 508.653.9000 x123.

If you would like to schedule a meeting around Embedded Software, please contact:

Jared Weiner, Analyst, Embedded Software & Tools Practice, VDC Research Group at: jweiner@vdcresearch.com  or 508.653.9000 x143.

Quanta Case versus AMD Underscores Importance of Component/Assembly Testing - Part 2

In the last week’s blog we looked at the semiconductor manufacturing process and the various testing steps that happen as raw silicon is turned into finished devices. In this week's follow up we continue looking at the testing process on the product manufacturing side. Last week we noted that Quanta was manufacturing notebook computers. Because of this, I want to make two observations about the product designs and manufacturing process for notebook computers.

• Notebook computer suppliers are continually trying to make them as thin and lower cost as possible. This means that Quanta might not have been using sockets and could be mounting the AMD/ATI devices directly on the circuit board. This can be a potential problem in some cases where the device was exposed to humidity before being heated as part of the surface mounted device (SMD) process as it can cause device degradation that leads to future failures.

• SMD processes increases costs with respect to failures as repairing a SMD CPU or Graphic component on a computer motherboard with hundreds of densely packed conductors is time consuming, difficult, and the scrap rate of the entire unit can be high.

• As the issue cited in the lawsuit is thermal in nature, it is worth noting that a higher power device such as a CPU or graphics chip often require heat sinks/cooling features to avoid problems. This is another area where manufacturing problems could have been introduced as these devices need to have excellent heat transfer to the cooling feature. Thermal pastes and a process to ensure optimal surface contact between the device and heat sink are needed.

Now, we will look at a few key testing steps on the product manufacturing side.

1.) Incoming Test: This process is considered as being redundant to supplier testing before shipment. Product manufacturers used to commonly test incoming components but, due to cost reduction pressures, that practice is now very uncommon outside the Mil/Aero market. Although it is not likely relevant to the Quanta vs. AMD/ATI case, counterfeiting and other supply chain related cases where lesser specification devices are re-labeled make incoming test more relevant again even in consumer type product manufacture.

2.) Circuit Board/Module Test: The fully assembled circuit board or module is tested before it is embedded in the final product. Thermal transfer issues can be identified by the use of infrared, optical, and or sensors. Repairs are expensive for problems found here but, it is still far less costly than having the product failing downstream.

3.) Highly Accelerated Life Testing (HALT): This is one last type of testing process that might have mitigated AMD and Quanta’s issue at either the packaged device, module/circuit board or completely assembled product stage. The unit being tested is put through extreme levels of hot and cold cycles while also experiencing other stresses such as vibration and g-forces. In this way, a myriad of potential production issues can be detected before the product ends up in customer hands and/or is in a mission critical role that embedded computers are frequently placed in.

In closing, HALT testing is an almost de-facto step in Military products let alone ones that might be launched into space. In Quanta’s case, you would never do this with all of the units but certainly at least a sample of them. If Quanta did not do this, the brand owner should have.  It does seem like a mystery to me about why this case is happening. It should be interesting to find out where the process became broken or which steps were skipped. If something interesting does come to light in the future we will surely revisit this case.

Quanta Case versus AMD Underscores Importance of Component/Assembly Testing - Part 1

This week I read of the lawsuit filed by Quanta Computer Inc. against AMD and its ATI division. The lawsuit alleges that components that they sold Quanta turned out to have heat tolerance related issues that caused the laptops they were used in to fail. It seems a little strange to me that Quanta is the only company with the problem unless they are buying a unique product or batch of products from AMD/ATI. Even so, with a multi-layered testing process the type of problem claimed by Quanta should not happen. The facts of the case will no doubt be revealed if additional claimants come forward and/or the case goes to trial.

Quanta is a Taiwanese contract manufacturer of notebook computers. Since they are a contract manufacturer this means that most of the finished goods they produce are someone else’s brand. They compete on the basis of cost and reliability/quality to get business from the owner of that brand. The damages they are seeking would be from their production losses from needing to repair or scrap finished products and/or subsequent damage to the perception of their own corporate brand.

This case has great relevance to the embedded hardware markets we cover and underscores the importance of a multi-tiered testing process. Therefore, I thought I would share some insights from my 30 year experience in the automatic test industry. I can safely say that despite the perception that testing increases cost, the costs of failure go up significantly at each step of the process between wafer creation and when the finished consumer/industrial product is completed. As one might understand, the absolute worst case is when the manufacturing/design failure occurs in the finished product when it is in the hands of the end customer. How can this be avoided? In this blog, I will look at the manufacturing process for semiconductor devices and, in next week’s follow up, I will look at what happens on the product manufacturing side.

1.) Raw Silicon Wafer: Optical checks are used to look for impurities and surface imperfections before the wafer goes through the extensive chemical/photo process that creates the semiconductor product. This test is very fast and it can save you the cost of chemicals and lost production time.

2.) Wafer Test: The semiconductor devices are still on the round wafer. The wafer is tested by using a probe mechanism that makes temporary contact with all of its contact pads. The tester than makes fairly extensive tests to make sure that the device is worth packaging. Tests can be made at various temperatures as part of this test cycle. In some cases, higher temperatures are used to speed up testing.

3.) Package Device Tests:

• Quick continuity and resistance tests are made to make sure the wire bonding/connecting process between the individual semiconductor dies and the package were good and that the device does not have any major faults.
• More detailed tests are then made to ensure the device works perfectly. Several cycles may be involved with the devices being subjected to high or low heat and less than optimal input voltages. The ultimate goal is to subject to the device to similar conditions to what it will see when it is installed in the finished product.

If the device passes through all of these testing steps, it is then further packaged for safe transport and easy assembly into the circuit board by the company like Quanta.In next week's follow up, we will look at what happens on the product manufacturing side.

5 Keys to Winning and Holding Embedded Computer Share in the Semiconductor Test Market

With respect to Embedded Integrated Computer Systems (EICSs) the semiconductor test market has some unique attributes that may not be immediately obvious or logical to outsiders. The recent VDC report on EICSs used in the industrial automation market estimated 2010 revenues of ~$210 Million for semiconductor processing making it an attractive market to enter. Embedded computing suppliers that thrive here are likely to follow these 5 key rules.

Make it small: Floor space is at a premium in wafer fabrication/semiconductor test facilities. These facilities are often very carefully controlled for dust, static, electrical interference, vibration, temperature, and humidity and therefore represent some of the most expensive square footage in the industrial automation market with respect to operating costs. Computers that can be embedded inside or flexibly mounted to take advantage of available niches in test cells and or test equipment are well received.

Make it Fast: Reducing test times for a given device by even a few milliseconds or having the ability to test many devices in parallel are keys to winning the tester sale. EICSs in addition to deeply embedded Digital Signal Processing (DSP), Field Programmable Gate Arrays (FPGAs) and Application Specific Integrated Circuits (ASICs) are often used in high quantity to achieve this goal. It is important to remember that a semiconductor tester has to be faster than the state of the art devices it is testing. In this blog, I am focusing on EICSs but many of the 5 keys are applicable to deeply embedded computing components as well.

Make it easy/fast to service: Semiconductor testers are extremely expensive with it being quite easy for a well configured unit to cost several million dollars. Even so, the return on investment can be made in only a few weeks to the owner. Therefore, any downtime is very visible and Mean Time to Repair (MTTR) is expected to be in minutes, not hours. Suppliers should design EICSs to have very high reliability but also with easy to access mounts, enclosures, and internal components that allow them to be serviced while wearing a clean room suit and gloves.

Make a flexible configuration: The EICS that is required for a semiconductor tester varies depending on the role it is asked to perform. A production tester needs only a simple Human Machine Interface (HMI) but one that is used for test program development and debugging will need more memory depth and graphics capability to allow the engineer to see and manipulate test patterns as well as analyze the data that is captured while tests are run.

Make it exactly the same – for a long time: A semiconductor test platform will usually be actively sold for at least 5 years but often needs to be supported for at least 10 years and sometimes even longer. Once a tester platform is discontinued a market can develop for the used ones and, in some cases for them can be equal to or even exceeding their original factory price. This can happen when the demand for some legacy semiconductor devices becomes higher than expected. Once a test program has been written and specialized probe cards for wafers and/or interface boards for packaged device handlers have been designed it is extremely expensive process to move them to another tester platform.

Throughout the entire tester platform lifecycle, any changes in embedded computers can require that thousands of hours be spent to re-certify test programs and debug them if problems are seen. Faster computers will often be problematic if, for example, the programmer did not have enough settling time after an instrument was set up before making the measurement.
Changes to an EICS can also lead to increased inventory costs. Because of the MTTR concerns discussed earlier, caches of spare parts are stocked in globally dispersed warehouses and even right at customer sites to allow instant or very quick availability should a failure occur. Changes to an EICS can require multiple sets of slightly different inventory to be stocked.

In summary, a key to winning an embedded hardware product sale to a semiconductor tester company is being active in the design phase and then executing a commitment to provide a stable product through the entire tester product lifecycle. The surprise can be that a newer, faster, or cheaper EICS product will typically not unseat the incumbent unless the original supplier falters in one of the 5 key areas.

ESC Boston is Coming Soon - Schedule Time to Meet with VDC

The Embedded Systems Conference will be held this month (September 26^th-29^th) at the Hynes Convention Center in Boston.

VDC will be attending the conference once again this year and will be presenting our 7th annual Embeddy Awards for Best in Show live at the conference. The winners will be announced live ahead of Wednesday's morning keynote session.

So how can your company win the Embeddy award?

To be considered, you must schedule a meeting with VDC to discuss the announcement that you are making at the show. You can arrange a meeting time with VDC by doing one of the following:

Contact Stephen Balacco, Director, Embedded Software & Tools Practice, VDC Research Group by contacting Stephen at: sbalacco@vdcresearch.com or 508.653.9000 x 124.

Still need to register?

Online registration is still open and you can always register in person at the show as well.

We are looking forward to another great show.  See you all in Boston!

Freescale Files for IPO to Reduce Debt and Improve Strategic Flexibility

Freescale, the $4.4 billion global semiconductor company, filed a Form S-1 Registration Statement on February 11, 2011 indicating its intention to file for an initial public offering. The potential offering, which VDC predicted in a separate June 2010 blog, calls for the issuance of over 1 billion shares of common stock at an as yet undetermined share price in an effort to reduce a significant portion of its consolidated indebtedness, totaling approximately $7.6 billion at the end of 2010.  

VDC sees Freescale, which lost $1 billion for the year ending Dec. 31, 2010, also leveraging this IPO to improve its strategic flexibility.  Historically, technology companies that operate in competitive environments often struggle to move beyond large debt burdens which can limit strategic flexibility.

By having a significant cash-generating event, Freescale can reduce its debt and more easily fund new R&D efforts, new product development, expand product marketing campaigns, and compete with Intel to better target potential acquisitions. With significantly less debt, VDC believes Freescale can enhance its (1) flexibility for planning and exploiting opportunities and (2) adjusting more quickly to changes in the fast-moving semiconductor industry.

Channels of Distribution Expand as Computers-on-Modules Market Grows

Our Embedded Hardware and Systems research team recently published critical global findings of the rapidly expanding Computers-on-Modules (COMs) market.  Based on our analysis, COMs dollar volume shipments over the 2009 - 2014 forecast period are projected to grow at a rate faster than nearly all embedded hardware technologies which VDC analyzes.  In fact, based on extensive supply-side and demand-side primary research, we are forecasting the total global COMs market size will reach just over $910 million by 2014, and display a compound annual growth rate (CAGR) exceeding 23%.  

Among the clear drivers of COM shipment growth are (1) the system expansion and application-specific customization capabilities of the technology, and (2) the relatively low power consumption and small form-factor dimensions the technology possesses.

Another influential factor is time-to-market advantages OEMs, SIs, and independent hardware designers gain, as this audience continues to seek significant reduction in development time of embedded systems in industrial and commercial applications when utilizing COMs.                                    

From the channels of distribution perspective, our research suggests the marketing and sales of COMS will begin to fragment somewhat, as suppliers look to add more third-party distributors, VARs, and other resellers in an effort to capitalize on the expanding COMs market. 

In summary, VDC believes as the complexity of embedded hardware design and implementation accelerates, an ever increasing number of vendors - third-party distributors, and VARs, etc. - will continue migrating into the embedded space, with COMs and COMs-based solutions.  In fact, many large distributors have already begun offering development kits online, as well as offering other support services, that will help enable a more rapid solution development cycle.

Let us know your thoughts about global COMs market opportunities and how the channel will impact pricing, availability, and hardware and software integration.

Dear PICMG: When Will We Get a New, Smaller COM Express Standard?

Over the past 18 months I have asked a question multiple times that I know many customers have been asking as well and that remains unanswered, at least definitively: When will PICMG release a new smaller version of the COM Express standard?

This will allow for smaller, more compact, and lower power COM modules that also conform to an industry standard.  Something that customers have been demanding for some time and that vendors have been providing, albeit through smaller but also proprietary COM designs that have been marketed as closely tied to their true COM Express products, but that don’t actually met any official standard. Mostly because one does not exist. I also know that there has been a PICMG working group working on a COM Express extension that will allow for smaller COM Express modules, but nothing has been released to date.

I believe the time is well past due for this new standard. As every supplier continues to offer their own proprietary design(s) for a smaller COM Express form factor it will have a “fracturing” effect on the market that could cause customer confusion and hurt the entire market. Customers might begin to defer decisions because they do not want to bet on the wrong technology or design. The fear of selecting a form factor that may disappear in two or three years time because it is not an industry standard and therefore at the mercy and whim of the single supplier has a paralyzing effect on customers.  

The COMs market has been here before. When COMs were first offered by suppliers there was a large number of competing “standards” such as DIMM-PC, SOM-144, ETX, and COM Express. This delayed the rapid growth of this market as it caused a lot of customer confusion. It took about a good 30+ months for any of these to gain enough momentum over the others to become a de-facto industry standard that gave customers enough confidence to make a selection and allow growth to take off.  A lot of progress has been made since those early days and it has left us with two major COMs standards – ETX and COM Express. COM Express is clearly the path to the future in higher bandwidth applications, so it would be a shame to see this progress reversed as we wait for a smaller form factor version of the COM Express standard. VDC would love to see PICMG release such an extension to the COM Express standard in early 2010. This is a vital step for the market as a whole.