Ping: Vbe Thompson et al

Widlarizing ;-)

Tim

Follow up: Yep, they're still made, and decent. Those, the TL061/2/4, and TL081/2/4, same story.

Jellybean, ~20 cents @ 100. Datasheets advertise that common mode range includes V+, but guaranteed specs only go up to (V+ - 4V).

Cheers, James Arthur

Cheers, James Arthur

Production volume - and a small die - are presumably the dominant factors.

The LT1014 - which I've used - is on offer from Farnell for $A 6.i6 50+

The LM324 is $A 0.16 100+ - cheaper if you buy a full reel.

The dual LM358 is marginally cheaper at $A 0.154 50+

I put both parts into the Cambridge Instruments preferred op amp list - which covered some 150 op amps for a wide variety of jobs. The LM324 and LM358 were listed as the cheapest available op amps, preferred if you didn't need anything better.

They're about $0.01 USD in China (SOIC).

--sp

NJR makes a lot of crap opamps, cheap, too.

Used tons of them back many years ago- they were actually way better than the specs guaranteed.

--sp

Round here, the Chinese stores only seem to sell "bankrupt stock", which is what you can buy from receivers when somebody goes bankrupt. The receivers haven't got any easy way of finding out a sensible price, so they sell the stuff for what they can get from anybody willing to buy it.

If Element 14 will sell SOIC LM358's in smallish quantities for $A 0.154, i t's likely that other broad-line distributors will sell at similar prices.

OK, damn them with faint praise. ;-)

I design using specs, not hopes and wishes.

Wasn't 'hopes and wishes'- it was 100% testing in an automated mass testing oven. Nothing else would meet the cost/performance requirements set by the market participants.

These days no particular effort or triple-digit intelligence is required. Just read data sheets and pick parts. Commodity work for commodity rewards.

--sp

End of life? What reject rate. Testing isn't free, either.

?? You're talking about exceeding the datasheet.

I concur. A previous employ, JRC was a standard non-critical choice for all multi-channel applications. Used a whole mess of them in each product. The y would also design in 10-ohm carbon comp resistor "fuses" in series with e ach IC's supply pins, to limit the failure effects.

-RS

Yikes! That inspires confidence! I can't imagine telling people in a design review "the resistors in all of the ICs are to prevent fire". Yes, we do have fire mitigation in mind but not to protect infernal ICs.

In one-offs, it's done all the time. That's because few if any datasheets tell the whole story. Case in point: what leakage does one expect from a mylar or polypropylene or polystyrene capacitor? It'll be a lot lower than the datasheet guarantees.

It's useful to exceed 'the datasheet' on occasion. Many engineering designs come from test-to-failure experimentation.

Reject rate was around 20% for the critical tasks but many more were used for non-critical tasks, so never any waste at all. Like the aluminum material flow in the new Ford F-150 where the scrap is fed back into the production stream.

Of course the testing wasn't free (the test system itself was expensive too) but the best choice of 'guaranteed' part available at the time was too expensive for the market. Many competitors were doing much worse.

You'll find the same thing with tested parts (matched transistors) etc. used in precision instrumentation of the day. These days we'd probably just buy better parts and/or throw some software at it.

Sometimes the software just hides real deficiencies in the hardware.

--sp

For a one-off, no problem. For a million, specs mean something.

The difference is that Ford's scrap is being sent back to the smelter, not put directly back in the production stream. You're re-reeling parts? Ick!

Unfortunately, the reverse doesn't seem to work.

Sometimes some important thing isn't specified. And sometimes it's safe to use a part well beyond some datasheet spec.

Mosfet rise/fall time specs are sometimes pessimistic by 20:1. Surface mount resistors can dissipate many times their rated power if the PCB layout is right. Specified leakage currents are often 1000x (or maybe

This was a LONG time ago- through hole parts.

Watchdog timers are sometimes touted for that purpose, to my dismay.

--sp