surface-mount resistor test socket

Buy another reel and ship what you have back, for credit.

What a PITA. We probably miss-placed 1000x more parts than were packaged wrong. That, of course, doesn't count the parts slightly out-of-spec, though who knows, some of our precision resistors (e.g. shunts) could have been in error, too.

I worked in a group that did incoming inspection. It's expensive and they gave it up once Intel got their act together (~1990).

We just did, bought a new reel when we found the bad parts on the old, year-old reel. The new one is worse, if anything, but same sort of bimodal distribution.

This is a precision thermocouple simulator. We use a 0.05% resistor in a ratiometric RTD measurement circuit. Since we calibrate that subsystem, the fact that the resistors can be as much as 1% off slipped through our process. Actually, as long as they are stable (which they seem to be) it doesn't really matter. But we use a second resistor on the board as a self-test on the RTD circuit, and those are occasionally failing the self-check.

There are places where we really want 0.1 or 0.05% resistors, where we don't calibrate around things.

John

Use two sets of the tweezers. Make a set of common handles for them.

Another disty/vendor is in order. Similar date codes?

Put bad ones there, too. ;-)

Makes one nervous when you've discovered the problem a year, or more, after the fact.

Well, sure.

John

I'll take 10% of the net profit you make from using this idea, if you decide to market it. Otherwise, feel free to use it for your business or personally at no charge.

Make a rectangular pipe with a vee shaped entry point. Pushing the chip through the vee to the rectangular part orients the chip. At the far end of the pipe are 2 contacts (half kelvin). Make a pusher that has the other two contacts to complete the kelvin connection. When you push the chip to the end take the measurement. The length of the pusher, when it "bottoms out" sets the correct pressure for the connection, and makes it identical each time. A trap door arrangement at the end of the pipe allows the measured chip to drop out easily.

Ed

So fill in complete garbage and sometimes sites will be like this one and give you the information regardless

(1 off) SMD-A $425.00 SMD-B $500.00 SMD-F $500.00 SMD-G $498.00

You can buy 'smart tweezers' that are both DVMs and rudimentary LCRs, as well as just two wire 'tweezer' terminated test leads.

I'd leave the parts in their tape, and pierce with sharp probes. Fiddling with smaller smd parts manually is a waste of time.

It might be advisable also to do a test coupon post-process, populated with a pretested series of parts, if this is a vendor part validation.

RL

It would be great if we could probe resistors still on the reel. We'd need four needle probes to do a proper 4-wire resistance measurement, and some mechanical stuff to align the probes with the parts and control the contact pressures. It's hard to get low contact resistance with probes punching through the tape

All that sounds tough for 0603s.

John

Don't you do a lot of manual assembly?

We stuff roughly half the boards in our products. If we have a big batch, 50 or 100, we send out kits to a contract assembler. But we do small batches and first articles in house. We have two Essemtec semi-auto p+p machines. A few common parts, like bypass caps, are fed from reels, but most parts are loose, in a carousel. So I don't really mind de-reeling the thinfilm resistors to test them.

The resistor rep came by today and gave us two full reels, to replace the ones we had the problems with. He mumbled something about getting

John

That's sorta what I thought.

Aren't they printed on the reels? Codes on the parts themselves?

They screwed up!

No. This company, weirdly enough, makes super-precision resistors in E24-series values, 270 ohms in this case. The marking is "271".

John

Oh, I didn't think anyone could be that incompetent. When you said they mixed up the lots, I thought it was just a packaging problem. That *is* bad. I wouldn't be using them anymore.

But they make phenomenally good, and affordable, super-precise surface-mount thinfilm resistors. The TCs are a few PPM.

Everybody in this industry screws up once in a while. TI does; Analog Devices does; Intel does. That's almost inevitable in a complex business. The good suppliers are honest, and they fix things as soon as they can; these people did all that. Hand delivered two full replacement reels today.

John

(Well, I don't know of a case where LTC messed up.)

Except they can't. They *will* make that mistake again and receiving inspection can't catch it.

I was involved in some downright crooked stuff with Intel and even worse, from our perspective, from TI. I'd have nixed them, too, if it were possible. In any case, one's misdeeds don't excuse another's. There are other resistor vendors.

I doubt we'll see that mistake again. But why can't inspection catch it?

John

ends

mixed

What assurance?

Because all resistors of that value, regardless of other parameters, are marked the same. Even in your sample(s) a good share of the resistors even passed a resistance test.

The LT1028 datasheet, where they cut the noise plot off at 100 kHz, conveniently omitting the big 300 kHz noise peak.

Cheers

Phil Hobbs