torturing caps

without

Depends a lot on the diode. Maybe 10 uA for a 1A, 400 V silicon diode. Maybe 100 uA for a 50 A, 20 V Schottky diode. Maybe 10 nA for a fast small signal or microwave detector diode. Maybe Maybe Maybe. And maybe any of these will ruin the diode.

In the end it is how much risk (including liability) do you want to take?

=20

So is your claim that Fairchild is outright lying when they state "100% final test"?

Best regards, Spehro Pefhany

By that logic you can only attach leads to 4% of the diodes.

Here's another overview, specifically for diodes that shows that 100% are electrically tested after packaging:-

Has anyone here other than myself actually been in a diode production facility?

Best regards, Spehro Pefhany

Nothing? Come on. Stop pulling our collective leg.

Assault and battery, of course. ;-)

Cheers! Rich

The Crazy Old Hen was, I think, proposing I investigate the physics of the current decline. I was asking him for suggestions how.

I have enough gear to measure the electrical behavior accurately, but that wouldn't satisfy a COH. I don't think anything would satisfy a COH.

John

So buy 25 test machines.

If SPC tells you that the failure rate is, say, 0.1%, you'd have to trash every production batch. I really don't think anybody can manufacture diodes, or any semiconductors, that have sellable failure rates straight out of production, untested.

John

So is your claim that Fairchild is outright lying when they state "100% final test"?

Best regards, Spehro Pefhany

-------------------------

LOL,

"During limited production, where components must meet defined reliability goals, samples from a minimum of three lots are taken for extensive testing. These samples must meet or exceed defined goals in order for the product to be considered qualified and transferred to the reliability monitoring program."

Again, what the hell is sampling/lots for if you test every one 100%?

Sorry my expectations were so high. I actually thought people in this forum had at least a kindergarten education.

You outta learn to actually read instead of just finding key words that you think support your claim. You surely must be a liberal. If you think you aren't you might as well join as you would fit right in.

Clumsy attempt at misdirection. I said 100% of the parts are tested. Do you understand the difference?

LOL.

Ad hominem..

US political ad hominem.. even more lame.

Final score:

F- in debating skills, 0/10 in relevant knowledge.

Best regards, Spehro Pefhany

So buy 25 test machines.

If SPC tells you that the failure rate is, say, 0.1%, you'd have to trash every production batch. I really don't think anybody can manufacture diodes, or any semiconductors, that have sellable failure rates straight out of production, untested.

-----

That?s 25 per production run. What happens when you have 10 production lines? What about 100? After all you generally are making multiple components.

What you fail to get is that if your materials and equipment are perfect your components will be perfect and no testing will be needed. There will be no variation in the components and all will work exactly the same under the exact same conditions. What makes components bad are imperfect materials and equipment. It's much better to put in the time to make these closer to ideal then producing a bunch of s*it components that are going to fail all the tests you want to do.

Your real problem is that you are "thinking" and know vary little about quality control, statistics, and economics.

A real engineer wouldn't waste time performing tests that are useless...

I wonder why "untested" JAN-TX transistors cost about 30x what commercial parts cost.

John

whatever.

Wafer yield on some ICs is under 25%. Do they ship 75% bad parts?

I doubt that even simple parts like diodes get below maybe 1% bad parts, out of assembly. Maybe Spehro knows about diodes.

John

Back in the early '70's I was selling temperature-compensated zeners to the military, in lots of 1000, 100% burned-in and 100% tested, priced by the lot.

Their incoming acceptance inspection allowed for 3 bad units per lot, so I always made sure to ship 3 bad units ;-) ...Jim Thompson

I'm sure you've heard the (presumably) apocryphal story about the Japanese semiconductor supplier that provided the specified 3 bad units per 10,000 in a separate box..

Best regards, Spehro Pefhany

Wouldn't surprise me if it were true. Some of those MIL acceptance requirements were hilarious.

But I did have a bad situation, reminded by another thread that talked of voltage compliance testing... Hughes croaked a complete lot of these zeners by inserting them backwards in the tester... they didn't have the sense to stop the testing when 100% failure was observed :-( ...Jim Thompson

Don't know the exact yield, but you're probably close. They told me the chips from the periphery of the wafer were a lot more dodgy.

Best regards, Spehro Pefhany

Clumsy attempt at misdirection. I said 100% of the parts are tested. Do you understand the difference?

LOL.

Ad hominem..

US political ad hominem.. even more lame.

Final score:

F- in debating skills, 0/10 in relevant knowledge.

On a sunny day (Sat, 18 Jun 2011 12:40:10 -0500) it happened "Stretto" wrote in :

I think you make a lot of contradictory staments. Also you cannot generalise one manufacturers for all. Some may test all, some may not. Clearly some do not test all, as I had defective resitors AND transistors. I asked somebody at that time and IIRC I was told not every thing is tested. Also it is possible the tests go wrong, and give a right if something is wrong. So if everything is tested, but the test was wrong, then sampling again later could miss the defective parts. There is no need for personal attacks, there is plenty of possibilities. Again, I think only milspec and space parts are more extensively tested. J.L states that .4 percent undetected failure rate woul kill a manufacturers businees. But could not the failure rate for modern machines be much lower, maybe 1 in a million or less? Because how ever you slice it, you need to explain my defective trannsistro (empty can), and resistors with a much lower value than the color bands indicate in a tape strip. Also explain all my defective (10) MAX232CPT (could be rebranded, but still). Anybody from MAXIM here who wants those for verification?

As you scale up production volume, you scale up all the manufacturing equipment. Why not?

A machine that tests 20-40K diodes per hour might cost $30K. Run that

Semiconductor test is a big business.

Basic silicon wafers have defects. Then you diffuse it to make the junctions, and hope it's perfect and uniform across the whole wafer. Dicing creates more defects. Then you have to solder the diced diode down to nailhead leads on both sides, or solder one side to a lead frame and wirebond the other. Then mold and cure the epoxy case around it.

Do you think that can be done to PPM quality levels, without testing?

I don't.

I don't think any semiconductor, from diode to FPGA, can be sold without testing every one. I don't know the test yield of, say, power diodes, but it sure isn't low PPM. Things like processors and FPGAs are sold with test yields below 25%. Some chips, like flash and drams, are built with redundant stuff, tested, and bad blocks swapped out for good ones; otherwise yield might be 1%.

There will be

What sort of electronics are you designing lately?

John