Don’t get tied up in knots with your cables!

Posted on September 28, 2011 by bburckle
Embedded PC and cables

Photo by: Jonathan Arehart

It always seems like it’s the little things that often cause the most problems with your embedded system.  The more I work with electronic systems, the more respect and care I have developed for cables and interconnect systems.  This goes for projects at work, my home office, and simple repairs around the house.

As a new engineer, I was given the assignment of installing the first generation of computer terminals into police cars.  I thought it was a pretty crummy job that was given to the “new guy.”  Running power wires from the battery to computer system, data cables to the radio, and coax to the antenna seemed trivial.  Why should someone with a master’s degree in EE be assigned this dirty and mundane job of crawling around and under a car loaded with electronics to hook up a few simple cables?

Guess what?  If it is not done correctly, then nothing works!  And if the installation is not done well, it won’t last long before strange and intermittent problems appear which can be traced back to poor connections.  After my first installation of a small fleet of cars, I realized that this mundane job mattered greatly.  It’s a life lesson every engineer needs to learn outside the classroom.

Cables are the lifelines of an electrical system.  Cable reliability is based on both durability and signal integrity, and a cable system needs to withstand any environment in which it will be used, regardless of how challenging it may be.  With embedded systems applications can include automotive, security, pipeline, military/aero, heavy industry, outdoor digital signage, and much more.

There is an interesting article in Electronic Products Magazine titled, The right cable system for your environment  written by the WL Gore company.  The article describes the advantages and disadvantages of typical materials used in cable construction, including an explanation of the testing and data analysis a manufacturer should perform in order to verify that its cable will perform in a specific application.  The author also discusses potential factors can that affect the electrical performance of the cable system in an application.  These factors include electrical performance, mechanical stress, environmental stress, and application-specific stress.  These factors are interwoven and each has an effect on the other.  Finally, he challenges the reader to consider all the elements of the total cost of ownership before selecting any cable system.

The bottom line is that there is a lot of technology involved with selection and specification of wire and cable for your system.  Don’t just gloss over it.  Consider both the impact of performance and the consequences of failure.

Posted in Cables, Embedded PC | Tagged , | 1 Comment

I am not a fan of embedded computer fans

Posted on September 20, 2011 by bburckle

My blog last week pointed out that fans bolted on heat sinks are typically used in desktop and commercial applications.  However, the fans can eventually fail, which is problematic for remote and critical applications.  The best solution for an embedded pc is to go fanless. Embedded PC Fan Fanless operation requires heat sinks with adequate thermal conductivity to dissipate heat with natural convection cooling or conduction cooling to the chassis or other metal that can serve as a heat sink.  However, sometimes fans are absolutely necessary due to the heat generated by the fast and densely packed electronics on their circuit boards.

While researching data on fans and their failure mechanisms, I found an informative article titled Anatomy of Computer Fans, by Gabriel Torres, at a site called Hardware Secrets.  In this tutorial the author discusses the theory and operation of the brush and brushless fans and the typical source of failure, which is the bearing.  He focuses on a discussion of sleeve bearings vs. ball bearings plus a few other proprietary solutions as well.  He concludes the article by asserting that the advantages of brushless motors is that they produce less noise, don’t produce sparks, and are more reliable.

Another factor to consider is the bearing.  The bearing is the component that allows the rotor shaft to spin, the point of contact between the rotor and the stator.  There are two basic kinds of bearings available: sleeve bearings and ball bearings.  Sleeve-bearing fans have the lowest lifespan among computer fans and are the most inexpensive.  Ball bearings are more expensive but offer higher reliability.

If you need to use a fan in your embedded application, then you should select a two ball bearing fan.  These have the highest MTBF ratings, plus they allow the fan to be mounted horizontally or vertically since the lubricating material is locked inside the bearing.  They will endure hotter temperatures, have a greater life span and ultimately provide a greater long term investment.

The bottom line is that you need to do more with less.  Heat is the enemy.  It is best to use the lowest power processor to run your embedded solution  Fortunately, Intel with their Atom™, AMD with Fusion™, and DMP with their Vortex have all introduced new processor families with much lower power requirements for x86-based applications.  Now appearing in current and next generation single board computers, these sub-10 Watt processor families offer a great balance of computation power, I/O and x86 compatibility.

Posted in Extreme Temperatures, Uncategorized | Tagged , , , | Leave a comment

It’s Hot Out There!

Posted on September 8, 2011 by admin

by Bob Burckle, VP

Embedded Devices Must Be Able To Withstand Extreme TemperaturesIn North Texas, this summer has been brutal.  So far we have experienced 68 days of afternoon temperatures that met or exceeded 100° Fahrenheit.  And yet we still have the possibility in the next few weeks of exceeding 100 again.  Even at 10 p.m., the temperature could still be 97° outside.  For man or machine, running in this environment is difficult.

I like to jog starting at 6:30 a.m. before going to work.  It is cooler at that time (that is, if you think that 84°F is cool).  If you are not used to the heat and humidity, it is really difficult to run four miles.  As I pass by a large traffic control box on my usual route, I can hear a fan running to cool the electronics inside.  I am sure that there is an embedded computer inside, but I am puzzled as to why they would design a system that would need a fan.  We all know that rotational parts will eventually wear out, which will lead to overheating and system failure.

There are other options rather than using a fan for cooling.  For example, there are a number of vendors that offer fanless single-board computers (SBCs) and I/O that can withstand both extreme heat and cold.

What qualifies as extended temperature?   The standard commercial temperature range is nominally 0° to +50° Celsius (or 32° to 122°Fahrenheit).  That is what you would expect from a desktop PC motherboard designed to work in office environment.  However, industrial temperature operation is extended to a range of -40°C to +85°C.

So, how do you get an embedded PC to work in hot environments?  One key way is to keep the overall power consumption of the product as low as possible.  Lower power consumption generates less heat.  This means the components in these systems are under less stress, which results in longer life and higher reliability.

For simple devices such as resistors, it is important not to exceed the package’s temperature rating.  For semiconductors, you must stay below the junction temperature.  Using the lowest power consuming devices reduces internal self-heating.  Also, careful layout of the devices on the board plus heat flow analysis is important.

Fans are used in commercial applications but cannot be relied upon for remote and critical applications.  Fanless operation requires heat sinks with adequate thermal conductivity to dissipate heat with natural convection cooling or conduction cooling to the chassis or other metal that can serve as a heat sink.

Extended temperature capability is required for those systems that must operate in harsh environments.  This includes smart grid, Mil/COTS, security, pipeline, transportation, and other remote and rugged applications.  As the system designer, you know whether you need extended temperature operation for hot, cold or both temperature extremes.  Be sure to carefully review how and what has been done by your in-house designers or outsourced SBC vendor to meet the temperature specifications.

Posted in Extreme Temperatures | Tagged , | Leave a comment

Its Embedded Conference Time once again!

Posted on September 1, 2011 by bburckle
ESC Boston

ESC Boston

by Bob Burckle, VP

Summer is nearly over and a new school year has begun.  That also means that the fall conference and trade show season is back.  Two popular events for designers of embedded PC-based systems take place in September on opposite coasts of the United States.  One event is the Embedded Systems Conference (ESC/Boston) and the other is the Intel Developer’s Forum (IDF).

The IDF is scheduled for September 13 – 15, 2011, in San Francisco.  As its name implies, this is an Intel-centric show.  IDF is where people from every part of the technology world gather to hear about Intel’s latest advances and witness its vision for the future firsthand.  With hundreds of sessions, keynote presentations from top Intel leaders, and a strong set of sponsoring organizations from throughout the industry, IDF provides a rare chance to engage and learn across the entire compute continuum.  Over 150 Technical Sessions will be presented by top Intel and industry experts.  The sessions include tracks on Embedded Solutions, Intel® Atom™ processors, high-speed I/O technologies, solid state drives, ECO-technologies, and more.

A little over a week later is the Embedded Systems Conference Boston.  It is scheduled for September 26 – 29, 2011, in the Hynes Convention Center.  This conference supports multiple companies in the embedded space and a designer will see a wider range of processor technology, both x86-based and others.  This is the East Coast’s leading embedded systems event.  Engineers, designers, system architects, vendors, analysts and media converge for four days of hands-on training, educational sessions, and an interactive exhibit hall.  In conjunction with ESC Boston, three additional shows: DesignCon East, DesignMED and Designing with LEDs, will be in the same location as well.  One pass will allow an attendee to attend all four shows, plus they can customize their agenda to attend sessions from the different conferences.

Both conferences charge a fee for attending their technical sessions; however, the exhibits are free.   Register now for either the IDF or ESC conference by clicking on their respective links.

In this internet age, one question that continues to be asked is, “Are these shows worthwhile and cost-effective for me?”  With the Internet, finding information is relatively easy, but finding relevant information that is applicable to your design problem or specific issue is more difficult.  Simply searching the Internet is often like trying to get a drink of water from a fire hose.  The sheer volume of data is overwhelming and can knock you over, plus the accuracy of some of the information can be questionable.

Trade conferences can offer a great learning experience if the course content is applicable to your project and if the presenter offers more than just a warmed-over sales pitch.  In my opinion, both the IDF and ESC shows meet these criteria and are worth attending if your time and budget permit.

Posted in Embedded PC, Embedded Systems Conference | Tagged , , , | Leave a comment

SBC vs. COM: It Depends . . .

Posted on August 24, 2011 by bburckle

by Bob Burckle, VP

You know that you need the power of a computer in your project.  You also know that an embedded PC meets the computing power requirements and has an abundance of software resources to speed development.  So the question becomes, “What architecture and form factor should I choose as the core computer for my project?”

The answer depends upon a host of variables that include size, cost, performance, application software development and debugging tools, wired or wireless networking, power budget, time-to-market, number of units to be produced, operational temperature range, expansion flexibility, legacy support of existing products, RoHS compliance, ruggedization, specialized interfacing to other peripherals, packaging, documentation, quality, cabling, etc.

Fortunately for the embedded computing element, there is a variety of options for off-the-shelf, standard single-board computers (SBCs) or Computer On Modules (COMs) available in all shapes and sizes that will serve as a system component.  Each has its own technical advantages and limitations.  The good news is that an engineering team does not have the challenge of creating a custom computer, but rather they can start with a fully-tested design and focus on their core competency of developing a product for their company.

As you research various vendor pages, they will tell you why their SBC or COM solution is better than any other, yet the truth is often somewhere in between.  Much has been written by both camps in a variety of different publications.  I have found two interesting articles posted on RTC Magazine’s website that discusses these issues.  The first is titled “Interconnect Wars? Let Peace Prevail with Interconnect Standards.”  It discusses SBC vs. COM issues, such as power management, I/O, and expansion options.

The other article is titled “COM Express versus SBC: Deciding which to Use and Where.”  The author discusses how COM Express physically partitions the I/O section from the processor and memory into two modules whereas an SBC combines them in one.  Then he examines if these two are rivals or are there clear reasons for choosing one over another for specific applications.

So what is the right fit for my application ─ SBC or COM?  I can’t give a blanket answer to that question.  But I do recommend that you list and then rank in priority all the technical, business, and support issues that are important to the project’s success and then proceed accordingly with analysis, qualification, and selection.  Don’t underestimate non-technical issues, such as long-term support and availability.  Often these and others can end up causing major problems and/or disruptions during the ongoing production stage of your product.

Posted in SBC vs. COM | Tagged , , | Leave a comment

The Embedded PC is a System Component

Posted on August 17, 2011 by bburckle

by Bob Burckle, VP

Starting in the mid 1980s through today, I have seen a pronounced shift in how embedded systems are designed.  No longer is it necessary to design your own computer but rather it is faster, easier, and more cost-effective to buy most of it off-the-shelf from a variety of vendors worldwide.  The embedded computer is a “known-good” device which is then customized with specialty I/O and application software for your project.

The shift has been driven by (1) the ubiquitous PC, (2) time-to-market pressure, and (3) corporate downsizing.  The bottom line is that companies are demanding that engineers produce more complex product, at lower cost, and in less time.  Management and engineers have adapted to this challenge by “outsourcing” — purchasing some portions of their projects from an outside vendor.  Outsourcing means the embedded computer is now viewed as a “system component” for use in their product design.

One of the earliest architectures to realize the system component concept was the PC/104 bus.  It provides a small 90x96mm, modular, standards-based approach for embedding a PC into a product either as a standalone SBC or as a stackable mezzanine for computer I/O modules.  PC/104 modules are small form factor boards that are cost-effective, scalable, and available worldwide with a wide choice of products.  In a white paper titled PC/104 Embedded Modules, the New System Components, the author gives a historical perspective

SUMIT stack shot on EBX board

SUMIT stack shot on EBX board

on system design and how PC/104 system components grew to be a powerful building block for a variety of industrial applications.

Initially PC/104 started out with self-stacking modules using ISA for I/O.  As the PC market evolved, it added PCI and now it even supports PCIe for high-speed data transfers while still supporting legacy products.  Later a new small form factor product class called Computer-on-Modules (COM) was developed and has also been widely embraced as a popular system component as well.

Regardless of whether it is a SBC, PC/104 stackable module, or COM form factor, the unprecedented success of the x86 architecture in both office and mobile PCs has resulted in dramatic improvements in price/performance, power dissipation and size which have migrated into embedded PCs.  The key to the embedded PC’s success as a system component is that it is a much easier alternative to a proprietary design.  It has become a technology transfer tool that allows a system designer to open up seemingly endless applications that previously were unattainable.  Plus PC technology allows a designer to leverage widely available hardware and the vast software infrastructure which supports personal computers.

Posted in Uncategorized | Leave a comment

Does your Application need an Embedded PC?

Posted on August 10, 2011 by bburckle

by G.T. Hilliard, Applications Engineer

The PC has forever changed the face of the computing landscape.  It has spawned a generation of trained, computer-literate engineers and technicians across many different disciplines.  The result is that for new projects a question often asked is, “Why don’t we use an embedded PC in this application?”

The reason for this turn to PC technology is the designer’s ability to leverage widely available hardware and the vast software infrastructure supporting personal computers.  Plus embedded PCs have been hardened to run over extended temperatures from -40° to +85°C and without keyboards, displays, or rotational disks.  This opens up a vast array of industrial, medical, security, transportation, and military/aerospace applications.

With the hundreds of millions of desktop and mobile PCs installed worldwide, the PC architecture has become a de facto standard.  The PC architecture may not necessarily be the most efficient or the most sophisticated, yet it has become a popular embedded standard.  The reason is that it is the software, not the computer hardware that drives the selection process for a system.

The project design cycle has typically evolved into a software project focused on the specific application at hand.  The reason for this design approach methodology is that it focuses on a company’s core competency emphasizing areas where they can add value.  Another major reason is the lack of internal human resources within a company to design and build the system plus develop, test, and debug the software.  Finally the complexity of the new CPUs coupled with the difficulty of mounting BGAs on a board using RoHS-compliant manufacturing processes makes finding a reliable, on-going electronics assembly operation challenging as well.  The bottom line is that these and other factors contribute to hurdles and delays to complete a project in a timely manner.

If your project does not require tens of thousands of units or unique and/or exotic packaging requirements, then a small form factor computer is often the best solution.  In a recent article in the EE Catalog titled “Viva la Evolution”, the editor writes about challenges and choices in embedded small form factor boards.  She probes how embedded developers can make the best choices and also asks about some strategies that engineers can use to address the challenges inherent in using off-the-shelf embedded PCs.

Designing complete systems from scratch is becoming a luxury.  Since other options for system design are now available, the number of people planning systems with embedded PCs from proprietary hardware is falling.  As development times are decreasing and systems are becoming more complex, the predominance of the proprietary design has come to an end.

By moving to outsource options, embedded systems developers are finding a way to become more efficient.  Buying and integrating an embedded PC works because it is cost-effective by saving both time and money.  It also speeds up product design, development and delivery, which help with long-term competitiveness.

Posted in Embedded PC | Tagged , | Leave a comment

Semiconductor Wiki’s discussion of Intel® vs. ARM CPUs

Posted on August 3, 2011 by bburckle

by Bob Burckle, VP

I recently discovered a site called the Semiconductor Wiki project.  Its purpose is to be a knowledge sharing site rather than just a sales portal for a variety of issues that impact the semiconductor design and manufacturing ecosystem.  It has a strict “no shills” policy.  There are different forums and wikis on a variety of issues moderated by SemiWiki bloggers.  This site has a global reach and is designed to build relationships and network professionals throughout the industry.

The main article that drew my attention is titled “Intel’s Mobile Deja Vu All Over Again Moment” by Ed Mckernan about ARM vs. Intel® Atom™ processors for current and next generation mobile devices.  He offers an interesting discussion and insights, plus a comparison of the current debate of these two architectures based upon his experience with Transmeta’s processor back in the late 1990s.  This provides a historical perspective and parallels of a previous battle.  Obviously the market, players, and technology are different today, yet Intel is an even larger and more successful company with a huge R&D budget.  It will be interesting to watch what develops as they roll out new products.

Designers, manufacturers, and users of electronics used in embedded systems are all dependent upon the semiconductor manufacturers.  The most influential are the processor and memory companies because without their continuous evolution of new and improved components, the onward march of smaller, lower power, full-featured products could not continue.  Embedded devices are now everywhere and seemingly connected to everything, with their markets spanning the range from the consumer to the medical, industrial, transportation, and military/aerospace industries.  With the Semiconductor Wiki site, it will be interesting to go beyond the technical specifications to some of the back stories and strategies of the different firms.

Posted in Intel vs. ARM CPUs | Tagged , | Leave a comment

Embedded Systems Conference Silicon Valley

Posted on April 20, 2011 by bburckle

by G.T. Hilliard, Applications Engineer

The industry’s largest and most comprehensive technical conference and exhibition will be held from May 2nd through 5th at the McEnery Convention Center in San Jose, California.  ESC/SV brings together the largest community of designers, technologists, business leaders, and suppliers all in one place to provide practical design information you can use today and skills you can use for a lifetime.

The Embedded Systems Conference offers a variety of topics and training for engineers involved with embedded systems.  This year, the ESC Advisory Board consolidated the topics for the embedded community education and training under five broad categories.  They are (1) HMI and Multimedia, (2) Reliability, Security, and Performance, (3) Remote Monitoring and Wireless Networking, (4) Systems Architecture, and (5) Linux/Android/Open Source.

There are over 130+ technical classes, hands-on labs, and full day tutorials.  To gain more insight on the track descriptions, click on the links below.

There will be a keynote speaker each of the four days of the conference.  Keynote address speakers include Steve Wozniak, co-founder of Apple, Dan O’Dowd, President and CEO of Green Hills Software, Meg Selfe, Vice President of IBM Rational Software - Complex & Embedded Systems, and Jeri Ellsworth, American Entrepreneur & Self-Taught Computer Chip Designer.

The expo hall will feature more than 200 exhibitors, offering embedded systems designers an opportunity to explore a comprehensive range of hardware and software products and services.  Conference and Expo Registration can be done in advance online.

Try to be there, it is always an interesting learning experience.

Posted in Embedded Systems Conference | Leave a comment

Power over Ethernet

Posted on April 14, 2011 by bburckle

by Bob Burckle, VP

Power over Ethernet

Power over Ethernet

Power Over Ethernet (PoE) as its name implies allows devices such as security cameras, card scanners, etc. to draw power through standard Ethernet cabling.

A PoE system is comprised of a Power Supplying Equipment (PSE) on one end of a cable and Powered Device (PD) on the other end. The PoE specification outlines the different requirements for both PSE and PD devices. Each has uniquely different requirements. A single board computer (SBC) or embedded pc with PoE capabilities is considered as the PSE device because it supports the Ethernet interface which is required to supply power to the target device. But challenges can arise because of the number of possible loads and if the system is unable to deliver all the power required.

Providing DC power through an indeterminate array of switches, routers, crossover cables, and 100s of meters of 24 gauge wire to power an unknown device at the other end has some unique problems. Initially you could get about 15W, but the updated IEEE 802.3at-2009 PoE standard (also known as PoE+ or PoE plus) has allowed that to be raised up to about 25W. However some system and switch vendors have even raised that to nearly 50W by using all 4 pairs in the CAT5 cable. Yet PoE discussions are often over simplified when described as simply injecting a DC voltage on a CAT5 ENET cable. In addition to the IR losses which result in widely fluctuating voltages seen at the PD end of the cable, there are also some obvious safety issues involved in applying power to such a system.

Compliance with the PoE specification requires a PSE device to implement a smart, hot-swappable power interface. Such an interface must be capable of detecting line fault conditions and faulty PD devices before applying power to the interface. Additionally, it must continuously monitoring the PoE power supply status and quickly detect over current conditions and disconnects so that power can be promptly removed and reapplied as needed. Also, long runs of wire have significant inductances which can cause large voltage transients when devices are connected and disconnected. These short duration voltage transients must be filtered or ignored by the hot swap controller; otherwise, the PSE device may be unable to reliably determine the line status, and fail to apply power to the interface. There is also the issue of a lightning strike that could induce damaging voltages into the cables. There are other issues as well, but these are just a few of the key elements that a robust PoE interface must resolve.

If you are designing an embedded pc, SBC, or system that supports Power over Ethernet, then you need to make sure that you consider the various engineering issues involved with this technology. For more information there is a Power over Ethernet website devoted to the IEEE802.3af and IEEE802.3at standards. It features news, articles, and whitepapers. There is also a PoE video training presentation by Digikey and Linear Technology. The first 9 pages contain the technical data and the last 16 pages are product specific.

Posted in Uncategorized | Leave a comment