Printed Circuit Board Assembly & PCB Design SMT Electronics Assembly Manufacturing Forum

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

un-tested assembled circuit reject rates



un-tested assembled circuit reject rates | 7 December, 1999

The company I represent provides electronic manufacturing services. Some of my customers ask that we ship their boards "un-tested". By un-tested I mean we build them, visually inspect them and out they go. I find that on un-tested boards, it's not uncommon to find an occasional solder bridge, or a defective component ( inoperative relay for example ). It seems to me that the cost involved in building a test fixture and performing a functional test are definately worth it, especially if we're talking about a unit that has a hundred dollars worth of components on it. Anyway, back to my question... Has anyone ever done an statistical analysis on assembled circuits using only visual inspection and their failure rates? If we find defects in 5% of our pre-test units ( I don't know what our rates are, I'm just a sales guy starting some research...) is that high? low? about average? Before I address my concerns with my principle and customers, I would like to see if I can't gather some unbiased industry information. Any feedback you could offer would be greatly appreciated.


reply »

Brian W.


Re: un-tested assembled circuit reject rates | 7 December, 1999

In my experience, the most commonly accepted theory is that 100% inspection is only 80% effective AT BEST. That means that if your inspoectors are happy, noy outside concerns, well-rested, etc, they will catch 80% of the defects. This number goes down very quickly when outside factors start to influence them. You can improve this number short of testing by incporporating good, effective process controls (including SPC)and preventive maintenance. Automated Inspection equipment can also catch many of these defects, including solder bridges, polarity, some wrong parts, missing, misaligned, and other solder defects. Incircuit test will cost about between $5K and $10K per fixture and program. Cost depends on complexity. This does not include the tester itself. ICT effectiveness depends on the test node access and board design. ICT can test some IC's functionally, catch many (not all) wrong/missing components or polarity issues, and some solder issues (unsoldered, bridging, etc), and components defects(defective components, out-of-tolerance, etc). I have seen boards with less than 50% test access, meaning that only half the components on the board were actually tested. What functional test will catch will depend on the level of testing. Some components may be set in redundant circuits, and unless each circuit is isolated and checked, one circuit could be bad and the board still pass test. Other defects may not show up unless you are doing ESS testing where the board is subjected to various environmental conditions. Keep in mind that such testing adds labor cost in addition to the NRE and maintenance costs. I do consider these value added though. If you have set up your processes and controls correctly, control your suppliers, and have an aggressive process control procedure inplace, you can eliminate many of your defects via a combination of your supplier controls, process controls, visual, and automated inspection. I say this because I have been part of a team effort that did it. We reduced our SMT and Wave Solder defects to less than 50ppm. ICT can help catch the remainder of component defects, such as defective or out-of tolerance components. Many EMS Customers are starting to drive more of the testing back to the EMS Supplier. They have more confidence and reduce their in-house costs when they received testing electronic assemblies.

Hope this helps. Feel free to e-mail me if you wan t more information.

Brian W.

reply »

Scott Cook


Re: un-tested assembled circuit reject rates | 8 December, 1999

Tim, What Brian writes is all true. However, let em put it into a sales / marketing perspective.....

The REAL issue in your situation is: Who is responsible for the failures which WILL result in shipping untested product?

Typically, your customer ASSUMES that a visual inspection is acceptable. They ASSUME that any product which fails in a unit will be fixed BY YOUR PRINCIPAL, under warranty. This is how you wrap dollar bills around every board you ship. Your confidence level on shipped product should be low. There WILL be failures in the unit, and it's not really your fault. Yes, you can tighten your process windows. But until you develop a test strategy which both the customer and your principal can live with, you are shipping defects. Even then, you will get escapes, as Brian says. It's just that test can make the escapes and resulting rework more manageable.

IMHO, this strategy of shipping untested product is an equation ripe with failure. And it includes unfair expectation levels on the part of the customer. Your principal can't win!

I think some major education of the customer is in order. If they "can't afford" to test, then that translates to "it's easier to put the onus of rework costs bsck onto you". It's unspoken that they are passing the true cost of either test or rework back onto your principal. And you must ask yourself, "Is this a customer I want to continue to do business with?" I think not. A true partnership starts up front, with communication.


Scott Cook

reply »


Re: un-tested assembled circuit reject rates | 8 December, 1999

Well said Sotty!!!

reply »

best reflow oven

Fluid Dispensing, Staking, TIM, Solder Paste