Testing electronic components for manufacturers

[okjoe]

Anyone who could help me on this. Your input would be much appreciated. I am a distributor for electronic components. For some time I have outsourced my testing to another company and I thinking about bringing it in house. I am wondering if this is economically feasible and what kind of equipment it would take. I buy up to 100 different parts per month, sometimes quantities of 10,000 pcs. I need an automated way to do a functionality test. What equipment would I need? How do testing facilities do it? Any input would be great...

Thanks,

OK JOE

 

[itsmoked]

Oh my gosh.. You haven't said what components or to what level of testing. IC's? Use a Colt tester they cost 4 or 5 million dollars a piece. Resistors? A multimeter $20.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[okjoe]

Here is a list of what I am bringing in this month.

TI/BB INA118U

Siliconix DG411DY

IR IRFP460

NSC LM2577T-ADJ

NSC ADC0820BCWM

Altera EPF6016TC144-3

Tripath TA2020-020

Maxim MAX232ESE

Maxim Max710ese

Altera atmega128L-8AC

Motorola DSP56002FC40

 

[IRstuff]

The variety of parts and complexity probably puts the utility and economic payback of testing at close to zero.

> You need to have personality boards for each part -- these must be custom developments with part handler mechanical interface
> You need to have test vectors for each part -- the only parts that can be tested turn-key are smaller FPGAs, processors and simple logic devices on small, pre-configured testers.
> You need to have a part handler to test the parts efficiently

TTFN

 

[benta]

Why do you need to test the stuff at all? Is it used? Normally, semiconductor parts come fully tested from the manufacturer.

Benta.

 

[lnphoenix]

okjoe, I see your dilema. I work for a small test lab doing IC qualification & screening for space & defense industries, many of them obsoleted. I design test adapters and write test software for opamps, ADC, DAC, modulators, demodulators, RF & microwave mixers, amplifiers, VCO and many other linear & analog ICs, not to mention digitals & FPGAs. So, in my line of work, I see your lists are very typical. But I am an outlier, not the norm in this industry for the reason below.

As far as I know of, this is an industry is heading to extinction due to complexity and expertise spanned across many disciplines, meaning high overhead in acquiring expertise (many engineers) with low payback due to very low volume procurement. How many engineers a company has to hire to have enoug expertise in all fields (look at your ICs list) to do a small volume business? Many of them would prefer to work for a big company rather than to work for a small company. And even a big company couldn't afford to do such small business while incurring so much overheads, so they also have to look for outside test labs to do the jobs for them. Due to the vast knowledge required across many disciplines, it's very hard to do in house and majority of test labs couldn't even to test certain parts for you if they are honest in what they're doing.

BTW, equipments are not the real issue here because you can buy cheap useable & dated test equipments to do the job but knowledges and price for using the test equipments & testing the parts are a very BIG issue.

 

[fortified]

Beyond the discussed (and not discussed) test issues, when it comes to testing components in particular for the government, there is the data. As a retired test engineer for a major defense contractor, I can assure you it is more about the customer buying good data, as it is about oversight of the process.
Standards have to be met, and agreed to by all parties involved.
Most reliable suppliers (and most are reliable) would not consciously provide parts and /or data that was not accurate. Not to say it has not happened, but it is rare.
When it does happen, the events are recorded and remain on file on a master list. This list figures greatly in the supplier selection process as contracts become available.
Another part of the supplier selection process, and a less interesting one. When you agree to do testing, and provide government accepted data, you take on a formidable task.
Often, the documentation could easliy choke a horse.
You would have to have masochist tendencies to do business with the government and do less than what is expected of you. Many on-site, or visiting source inspectors have little technical knowledge (probably making them more dangerous is some ways) but the ability to shut down a line with a phone call.
On the bright side, in some 30 years I saw many successfully completed programs, the result of all parties involved understanding what was expected of them and not hiding problems in trash receptacles.

 

[lnphoenix]

It's true fortifiied, that in addition to the burdern of testing devices to milspec where many today ICs with finer geometry are not specified to operate under extreme conditions of -55C/+125C, we have to put in considerable amount of time to process data. This sub-contracting test industry also evolved to automate as much as they could and utilize as many "push buttons" workers with minimum knowledge to stream line their operations as efficient as possible. As I mentioned in my previous post, the cost of testing ICs to milspec is no longer economical when a IC socket is cost $100 for a finer pitch device and up to few thousand dollars for a GHz operating range RF device, while defense contractors are operating with budget constrain (with a their own waste & dead woods), this sub-contracting test industry is also adapt to remove as much as many complicated tests and to employ as "many people who do the job American doesn't want to do (with that price? you got to be out of your mind)" as possible. Of course, it also fits in the grand scheme of a bigger system to reduce cost as much as possible with the new strategy currently employed to emphasize to system quality probability strategy instead of component quality probability of the past. I have my own reservation about this scheme because it tends to obscure and masks the complex parts in the ocean of many simple parts' statistics.

PS: My conjecture is okjoe is more concern about whether he bought counterfeit parts than testing parts to milspec. But the differences are not that great so it would be a natural step for him to move up the chain.

 

[itsmoked]

Frankly I think we are wasting our time on okjoe. He's a one time poster and hasn't been back in almost a month.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com