Low MOSFET IDSS current

This was before your time. Things were a bit looser then.

I just used it as an example of how mistakes can happen. Certainly things have improved since then.

I hope so. But I'd like proof of what is going in the product.

I am planning on fixing it before there is a problem.

How? You are telling me they take each part out of the reel and test it? To what tolerance? And then they put each part back? And put the top tape back on?

Maybe the manufacturers do that before they put the part in the reel. But they don't give you any information on the distribution. They don't tell you what their test limits are.

You are a very small company and have got away with things like this since the beginning. Very few large companies could afford to operate this way.

I have grown companies from nothing to multimillion dollar business twice. I am planning on doing it again.

But setting up a business takes a lot of planning. I'm hoping to prevent a lot of mistakes I made before. Incoming inspection was a big one. I don't think it would take as much as you seem to believe to do a proper job. And that's only one of dozens of details I need to take care of.

I've been in the biz longer than HCMOS has been around. Thanks for playing though.

Well, real examples are much more valuable than fabricated ones. Just sayin g.

Cheers

Phil Hobbs

Good for you--I hope to do the same. Which ones were yours?

Cheers

Phil Hobbs

Yes, but that was at IBM. I also started long before HCMOS. But I was probably involved in incoming inspection much more than you. Every part in my designs had to have a written test specification. Often, I had to modify the spec written by someone else, then get it approved. Any new parts were a real pain to deal with.

Production testing for hard disk drives.

The new one will still be in instrumentation, but a different area.

Nope, long before that, and the whole issue is irrelevant to your *ahem* mi sremembering the 74HC7474 thing and then doubling down when challenged.

Steve, if you have anything specific to back up your contentions about the alleged misbehaviour of top-shelf distributors, I'd genuinely be glad to he ar about it, because I depend on them. If not, you really should just fold your tent on this one.

Cheers

Phil Hobbs

Hmm.

comes up with "couldn't resolve host name". Did your company have a different name?

Cheers

Phil Hobbs

ou

Not strictly true - even proof of principle prototypes can fail if a single part is out of spec.

Back in 1977 our prototype ultra-sound scanner stopped working about half a n hour after we closed the door of the rack, which we traced down to a Texa s Instruments TTL shift register which stopped working when the ambient got over 40 Celcius. It took us an hour or so to find the duff part, at the te st site, so we got home late that evening.

n

it

One of the harder to find faults on a Cambridge Instruments electron beam t ester back in 1985 turned out to be a resistor lead that had be soldered in to a via on a signal line rather than to Vcc. The signal line was mostly at Vcc, so the fault didn't happen often. I was the guy that actually found i t, but the narrowing down process had taken a couple of people a couple of days.

ust

w

You ship in volume, but fairly small volumes. Professional merely means tha t you get paid for what you ship. Steve Wilson isn't as enthusiastically in to self-publicity as you are, so he may have rather more standing than you think.

--
Bill Sloman, Sydney

Don't be silly. They make the parts, trim/bin/test them, and then they put them on the reel.

The guaranteed limits are on the data sheet. Sometimes they include histograms.

We sell a million dollars in a good month. We sell to the biggest aerospace and semiconductor companies on the planet. Our stuff is way better than the things they make themselves.

Burning holes in tape packaging to test resistors and caps is not an efficient way to do that.

Parts are amazingly reliable these days, orders of magnitude better than the nasty old days of early 74xx plastic chips. If you test millions of resistors and caps you'll probably never find one out of tolerance, much less DOA. Really. [1]

Skip that one; it's a waste of time.

[1] well, excepting NTC capacitors. Tempcos are all over the place, but at least the same in a reel.

PCBs are probably the worst things that we buy.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

As I said, it was a long time ago. Back in the 80's. There was no ebay. The only surplus vendor was Holtek. They had some interesting stuff, but we didn't buy much from them.

The only other place to buy semiconductors and components was national distributors. They screwed us royally sending us junk.

Things were very much different then. Pretty much absolute chaos as far as parts quality goes. We spent a lot of time trying to recover from bad shipments.

Nowadays, you probably don't have to worry about getting too many bad parts from modern vendors. So stop worrying.

The more I think of it, a better place to do the test is at the pick and place, just before the component is placed on the pcb. Then I will get a record of the value of each component and the location. It would take only a bit of software to convert the location to a reference designation.

An earlier poster mentioned another machine that does this, but he didn't know if they actually measured the value.

Make that Halted

Halted was a fun place to visit. Junk piled everywhere on the floor. Then they moved into fancy digs with everything organized on tables so it wasn't as interesting. So I stopped going.

Also, I got the part wrong. It was a plain 7474. We didn't use 74HC anywhere in the company at the time.

Anyone who would even think of buying production components from a junk dealer deserves what they get. They sell floor sweepings, and anything else that looks like a part. Did you buy production hardware from PolyPacks and RadioShack, too?

So it was marked "747474"? This is getting ridiculous.

And a machine that can test every component. ...and the money to pay for such a machine and time it adds to production.

And if the reel is bad, you shut down production until you can get more parts? Sure sounds like a great idea!

Our Universal machine does optical inspection, not electrical, on every part.

The more real problem isn't DOAs, it's parts damaged in process, or by electrical abuse. Lead-free temperature profiles are bad, as is water washing (for some parts) or solvent wash (for others).

We've recently had problems with presumably sealed relays. The conclusion: never get them near water.

We ran a couple of these boards

(and then threw them away)

through various soldering and cleaning cycles. We have setups that can measure contact resistance and leakage on almost every contact on the board.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

There was Halted and Haltek. That always confused me. There was even an electronic surplus place on Market Street in San Francisco.

The real estate boom killed most of them.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

The surplus houses were good for test equipment, cables, hardware, weird stuff for engineering and prototypes, not parts for production. I got started in picosecond electronics with flea market and junk store equipment and parts, when I was living on unemployment.

I also bought broken microwave and fast-pulse gear to take apart and study.

All this was under $200:

The Foothill Flea Market was great. Kids would empty their dad's garage into a pickup and haul it there to get rid of. Lots of old HP stuff.

There was a set of girlie-nudie HP postcards that the sales people gave away; I should have bought them.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

This is very alarming.

I'm thinking of ordinary components - resistors, capacitors, diodes, maybe inductors. Do any of these show up bad in your testing?

If so, how do you know where the parts were damaged? What are the symptoms - out of tolerance, open, shorted, etc?

How many other bad parts are on the board that didn't get detected?

How can pick and place damage a part?