by 
Based on some of the replies to my blogs and conversations with customers, I have noticed that some individuals are confused over single and sole sourcing. I will start with definitions and then share some observations and thoughts on the subject. (See: Single Source Premium: Unnecessary Cost and Purchasing: Risks in ‘Virtual’ Second Sources.)
Sole sourced: A component where there is only one manufacturer capable of making it.
Single sourced: A component that has more than one manufacturer able to make it but for which you have qualified only one source for your application.
I use the term manufacturer in my definitions because there are situations where companies believe they have multiple sources of supply but are really only single or sole sourced. For security of supply reasons, you want to ensure that a single point of failure cannot disrupt your ability to make and sell products. A fire in a factory, major earthquake, or strike could cause significant supply disruptions, if any of your components come from only one factory, in one region, with a common labor pool.
Many companies feel they are multi-sourced when their approved list of vendors for a component includes the manufacturer along with one or more distributors as the source of supply. By my definition, this is risky because they remain single or sole sourced, and, with the low levels of inventory held by distributors today, they are vulnerable. For single sourced components, you want to establish true alternate manufacturing sources, and for sole sourced components, you want to understand the security of supply measures taken by the sole source manufacturer to ensure no single points of failure exist.
Sole sourced components are usually devices that contain a significant amount of intellectual property and, once selected in a design application, cannot be replaced or substituted without a significant amount of engineering and redevelopment. FPGAs and telecom chips fall into this category. With this class of components, pricing is often decided during the design cycle, and once the design award has been made, it is difficult to achieve cost reduction.
Often associated with these parts is a process called registration. Registration sets a minimum price for a component through a specific channel to a customer, which means that the part will be more expensive through any other channel than the registered one. The registered channel is the representative or distributor that provided engineering services to the customer, and registration is the means of tracking the commissions to that representative or ensuring that distributor gets the business. Registered pricing is often misunderstood by the customer. Registration establishes the minimum price to customers, not the minimum price to the market.
Single-sourcing is an unnecessary disaster waiting to happen. Along with true single sourced devices, there are what I refer to as weak single sources. Both need to, and can, be eliminated. Here is my shortlist of what to look for on your approved manufacturers or vendors list. Seek approved vendors list (AVL) or approved materials list (AML) alternatives for any component that:
- Shows only one specified part.
- Is from only one manufacturer. (Sometimes a high-risk and expensive practice of specifying tighter tolerance components as alternatives is used.)
- Lists distributors as second sources.
- Shows an end-of-life component as an alternative.
Make sure that you have price quotations from all alternative suppliers to confirm that the alternate source is commercially acceptable. An alternative at three times the price or more is not a true alternative.
Ken–thanks for the definitions, to start. I can see where people may be confused. So really, no company has to be single-sourced unless they choose to be, correct?
Correct, no company needs to be single sourced; however as my blog points out, all companies will likely be sole sourced on some components.
Sometimes being single-sourced may be a viable option as long as the risks are recognized and hedged/managed. In cases like this, a company may choose a partner that has multiple manufacturing locations and which is able to guarantee supply. Apple, I believe, has done this in some cases and poured money into the supplier to help mitigate risks.
Agreed. Its important that everyone involved in the supply chain and related to the product development must stand against counterfeiting. The worst that can happen to a project schedule that designers get the counterfeited part and program suffers and at the end the whole organization suffers. I remember way back back in 2000 when we were designing around 8 bit controllers, once out 50 microcontroller we ordered 20 were not working. The production gets severly affected and so much time gets wasted. But sometimes the project delivery pressures are so high that buyer just have to give in just to get the parts in. But one can imagine the seriousness of the problem if these counterfeited parts find their place in medical or space application. Definitely it cannot be ignored. But how do one ensure that they are not getting these parts?
Dan, Exhaustive testing by LMI, Idaho National Labs, the Department of Defense, Altera, SMT, Micron, Raytheon and a host of other world class players support the Defense Logistic Agency recent mandate to use DNA Authentication solutions for counterfeit protection. I do not understand your objection, as expressed in another recent blog, to this technology and would appreciate further explaination. Thank you
Owen, I'd like to add to your message by asking if anyone has done any studies on the financial case for DNA authentication. This industry is happy to consider anything that would further its cause as long as it makes financial sense. Is the case strong enough financially for DNA authentification and if it is not, what would make it more compelling?
Bolaji, The cost of DNA Authentication, from what I have gathered from Douglas Alexander in his June EBN article titled: “Calm Down, Counterfeiters can be stopped “, appears to me to make financial sense, here's an excerpt:
“My next question concerned the cost of applying the marker to the materials and the cost of authentication. The cost for applying the DNA impregnated ink, adhesive, or spray is less than a penny per application. A hand-held scanner can quickly detect the presence of the DNA, but if an OEM wishes to authenticate the sequence, the part has to go back to Applied DNA Sciences' Labs for a full lab workup. “Aha,” I said. Here is where things get costly.
In my own research, I discovered the cost for sequencing to be about $1,000. However, because Applied DNA has its own labs and labor force and are interested in high volume deployment, they are able to control costs to make it very affordable”.
My guess is the DLA has done their own cost/benefit analysis but unfortunately I haven't been able to get info from them. Given the mandate is scheduled to be effective by September 30, 2012, I'm sure more details will be available soon.
Good to learn the difference between these terminologies. Many times in the BOM we face these issues and we also want to work on these but I guess management doesnt consider this to be a revenue gernerating work unless the problem sit on our head. I remember while working for a product company, I found many issues like this, duplicate part number, wrong parts getting bought due to part number problem. This itself is a big job and must be given equal priority by management.
The primary issues with DNA authentication are logistics and financial issues, but there are many more. The SIA has thoroughly documented all concerns directly back to DLA.
From a financial perspective, any cost added must be justified and fully understood. To date, the financial costs are neither justified nor fully understood. Adding $.01 to a part that costs less than $.05 or less makes no financial sense….so now what do you do? Exceptions for low-cost parts? Who makes this call? From a market perspective, long-term system markets contribute less than 1% of the total semiconductor revenue in the world. Long-term markets (DLA purchased parts and all of DoD) do not significantly influence the semiconductor OCM's. There's no way 1% of the market should be increasing the price for the other 99% of the market – it just makes no financial sense.
From a logistics perspective, this is where it gets really interesting. The 1% market is relegated to buy almost exclusively through distribution channels and not on a direct basis. Their revenue is simply not high enough to warrant direct sales most of the time. This means that the DLA mandate is mostly a logistics burden place on distribution and not the semiconductor OCM's. To add to the logistics issue, take into account most of the components that DLA will buy for the next 10 years have already been manufactured and are sitting in authorized (or non-authorized) storage. We have over 5 billion finished devices from over 50 different manufacturers…..the source for many DLA purchases……sitting at Rochester Electronics. Which DNA marker is used? Should the parts go through testing again once they are handled for marking? What does that DNA marker really do to combat counterfeit when we already know the parts at Rochester are FULLY AUTHORIZED to begin with?
In essence, the DLA mandate turns into a penalty (both financially and logistically) for a fully authorized source like Rochester Electronics. We know our inventory is fully authorized and direct from the manufacturer. We go to great lengths to ensure this and it is fundamental to our business model. The SIA and Rochester Electronics remain committed to anti-counterfeit, but this mandate is not the way to go.
Dan: I think that is the same argument that applies to RFID tags. It is one thing to put tags on a lot or a pallet; quite another to ID each and every component. Yet, if you can't track each component, how can you prove it is authentic, particularly if lots are broken up or kitted? In terms of the value of the component, a failure in a .05-cent part or a $100 part is still a failure, especially if the failure is in the end-product.
Barbara,
If an authorized source doesn't go out and mix customer excess purchased product with original product, authorized sources will have original and authorized product ONLY. When you buy from these types of authorized sources, you never get counterfeit…..no DNA marking necessary. The value of the component matters when the 1% market (DLA and DoD) tries to increase the price to the other 99% market.
Dr. Das at CALCE has brought up the fact that original product has a built-in signature just by how it has been constructed. Each and every single part has a unique signature consisting of die attach material, package material, die size, die material, etc…. This is but one example where industry could be involved to come up with a solution instead of a mandate. There are other ways too.
You bring up an interesting point about tracking and it relates to my logistics argument. Who is really doing the tracking here? What I see is somebody selling DNA and nobody tracking. Applied DNA is selling to anyone, not just authorized sources (SMT for example). ANYONE can put DNA on a product and Applied DNA gets to decide who that is. DNA marking says nothing about product authenticity or long-term reliability (handling), just who marked the product.
Dan
Dan: I think we may be beating a dead horse here, however, with regard to the esteemed Dr. Das of CALCE, as I've pointed out before, he has in the past supported marking of parts and material at its manufacturing point as a means of authentication. No doubt the Doctor has said many things which can't be ignored. For me, I'm a firm supporter of DNA verification, I believe it is a disruptive technology that will have a profound impact, not only in electronics, but throughout the counterfeit world we live in. In the meantime, I look forward to your Webinar. Best of luck.
http://www.calce.umd.edu/seminars/cws20120124.htm
I agree. I can't think of anyone in this industry whose entire passion is tracking counterfeits and making money from doing this. Tracking fake parts cannot be a part of a company's business at least not until this problem because so gargantuan someone sees an opportunity to make money from it.
It does seem silly when put that way. Not having to track stuff seems to be the better option. But I don't think OEMs, EMS companies are anyone, for that matter, won't mix products. Moto tried that when it founds its chips were being sold in the gray market for 2x their price: it declared anyone selling to the gray market will be 'cut off.' It never happened. So, given the reality of the market, what are the options? I know some of that will be covered tomorrow on the Webinar.
When we talk about counterfeits in brokers… are they buying those parts at very discounted prices? The buyer must know that there's something weird going on when a part costs 50% less than the original.
Shouldn't buyers be trained as well?