This is indeed a huge issue for our customers at VPT as well. Thank you for posting this article; we will share it with our customer base.
We spend a lot of time educating customers about 'commercially available' vs. 'commercial grade.' On EE Times, we've posted a whitepaper that addresses these same issues when selecting DC-DC power converters for hi-rel systems. Tin whiskers, MIL compliances, and many other issues are discussed. Here is the link for anyone who may be interested: http://www.eetimes.com/electrical-engineers/education-training/tech-papers/4372316/Selecting-the-Correct-Level-of-DC-DC-Converter-for-Your-Application
Our most important advice to customers is to read the datasheet extremely carefully to ensure they are receiving the quality level that is needed and expected for a particular application. We also suggest that customers review a manufacturer's qualifications, counterfeit parts policies, obsolescence policies, and materials policies as part of the purchase process.
Thank you for highlighting this important industry issue.
She seems to have it right to me. Life cycle, obsolescence, quality, these are some of the key issues. Even when you look at COTS electronic devices specifically built for mil-aero markets, like an avionics computer, for example, device manufacturers and buyers of those systems tend to have the same issues. For example, the semiconductors used in the units (like Intel, ARM, or AMD processors) are often not available with industrial ratings or ruggedization, so the mil-aero board manufacturers will find themselves temperature-screening their boards and throwing away anything that doesn`t survive, driving up the price of the COTS mil-aero device astronomically. Even after this process, the customer or system integrator still has to answer the question of component obsolescence management, often by making LOT purchases of, say 20-30,000 chips or other critical components to accommodate maintenance over a 20+ year mil-aero program life.
Having said that, I'd be very interested to hear the "basic concept of COTS" on which which you are basing your disagreement.
There's also an industry dedicated to converting COTs devices and IP for military and life-criticle requirements. You can get lead-free components (e.g. BGAs) converted to leaded solder. You can also find companies that take pretty standard PC designs and create ruggedized versions.
Perhaps it's not so much that COTs is "the solution", nor that COTS is going away. I can see a hybrid approach. Components and designs that don't have a commercial equivalent will be custom designed. If longevity and reliability above commercial requirements is not necessary, than pure COTs can be used. If there are existing components or designs available, but not with the high-reliability requirements, then the components can be reworked and the designs can modified.
this is what I saved years ago:
The case for COTS: follow the money
By John Keller, chief editor Military & Aerospace Electronics
Commercial off-the-shelf (COTS) components and subsystems have brought many things to the table in military and aerospace electronics design over the past decade, not the least of which are enhanced capability and reduced costs.
Despites its benefits, however, I wonder whether we are on the verge of losing COTS and moving back to the bad old days of closed systems and proprietary architectures. It's not the technological issues that worry me; it's the business issues — or more to the point, the prospect that too many people are confusing the two.
When COTS first started back in 1994, let's remember, it was all about business. Then-Defense Secretary William Perry gave birth to the COTS movement not out of concerns for technology, but because he was sick and tired seeing the government spending exorbitant sums of money developing technology for the military that already existed in the commercial sector.
Many of these technologies that the Pentagon under Perry was paying to develop were already there, ripe for the picking. For Perry, it was about the money, not about the technology. COTS was Perry's attempt — and his legacy — to save the taxpayers money by refusing to reinvent the wheel again and again. ...
And this is a reference in a paper about COTS:
Perry, W., “Specifications & Standards - A New Way of Doing Business,”
Internal Memorandum, U.S. Department of Defense, June 29, 1994.
I have seen similar problems with non-mil/aero use of COTS. In particular, when the next year's model is NOT an exact drop in replacement there can be a number of unintended consequences. Things like: my system worked last year but when I replaced XYZ box every now and then it locks up or dies. This can be most frustrating and difficult (if not impossible) to address. Oftentimes, the manufacturer of the new COTS item IS NOT INTERESTED in your problems and you are left holding the bad. Sometimes, it just makes sense to make it yourself, sometimes not.
If obsolescence in 5 years is considered reasonable with IDS-based black-box replacements regularly procured as technology progresses, there are still a lot of options out there. I used to work for a large Peoria manufacturer that expects a 5-year life out of in-house and customized designs used on heavy equipment, and makes plans for obsolescence of parts and assemblies. The need here is for the military to make up its mind, "Do we require gold-plated hermetic everything electronics that we are willing to pay manufacturers to continue to produce well after the technology lifetimes of several of the parts involved, or do we want relatively inexpensive and slightly less reliable equipment that we can afford?"
Let's review the bidding that got us to where we are today... During the Reagan era, it was hang the expense, makes lots of military stuff to bury the Soviets economically. COTS happened starting about the 1st Bush era, and was to try to control the extreme costs of meeting a bunch of very over-blown and over-grown requirements that were often based on issues with old technology being placed upon any new design. Then and now "COTS" systems were generally not pure "commercial", but ruggedized and carefully-parts-chosen designs to keep the military skeptics at bay (yes, I was also a submarine officer in a prior life). To expect a true "commercial" design to perform for a 20-year product life in a military application at the rate of technology change today, while not funding and supporting cutting edge R&D and product releases, is a total pipe dream.
As regards RoHS, some of the first lead-free solder was developed for military and space applications, based on Indium alloys. These alloys, albeit painfully expensive, are still available and in use today. Several manufacturers can readily adapt to Indium based paste for board assembly, but you definitely won't get cheap made-in-China real-commercial board assemblies made that way. I was also rather surprised that the discussion of tin whiskers did not include a side-note about gel-potting and conformal coating use to limit this effect, nor a discussion about clearances and circuit lowering of voltages that can be used to reduce this effect. I do understand that QFN and BGA packages make both of these techniques a little limited. However, a modern SMT package that is designed specifically for aqueous cleaning beats the socks off a metal-ceramic-frit "hermetic" package in cost and availability.
Thanks for your response. You are correct, I should have added more detail on the Reduction of Hazardous Substances Directive (RoHS), which was put in place in the European Union in 2006. (I'm still trying to catch up with how much detail I need to provide in these blogs. (Note to self: that would be more))
So, a quick explanation of RoHS. Designed to reduce the amount of toxic substances released into the environment, RoHS sets maximum levels for six controlled substances: : lead, mercury, cadmium, hexavalent chromium, poly-brominated biphenyls (PBB) or polybrominated diphenyl ethers (PBDE). In order to sell into the EU, manufacturers need to demonstrate that their products comply. In the case of electronics, that has led to manufacturers converting from lead-based solders to tin, which carries the aforementioned risk of tin whiskers.
As you pointed out, RoHS is a European standard, which means COTS components bought and sold outside of that sound do not need to comply. That said, today's manufacturers play in a global marketplace. If they want to sell into the European Union, they have to comply with RoHS. That gives them the choice of either developing and supporting two product lines/process lines – one leaded, one lead free – or a single lead-free line that produces devices to sell into both marketplaces. It's understandable, if frustrating, that many manufacturers elected to go the easier route. Yes, it has ramifications, but mostly for a small subset(mil/aero)that is not the primary market for most of these vendors.
And that brings us back to the unfortunate issue of how market dynamics can create issues with the use of COTS components in mil/aero systems.
Blog Doing Math in FPGAs Tom Burke 16 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...