Absolute Max

There is no simple rule. It will usually depend on the aging/failure mechinism.

With leds higher current means more heat and shorter life. For most applications leds are more than reliable enough so you can push near the limit.

On the other hand I like a nice margin on a fets safe operating area. Otherwise some unexpected glitch blows them up.

Standard electrolytics are normally specified with 2000 hours life at

TTL is specified at something like 5.5V max. It's designed to run at

Depends on the expected failure mechanism.

For an LED, do some basic math to estimate junction temperature at Imax. If it's reasonable, the concern would probably not be sudden failure, but light output degradation over time.

Tantalum caps, derating to 33 or even 25% of rated voltage is prudent. For film or ceramics, 100% is usually safe. I've tested 10 volt ceramics to 120 volts and ran out of power supply before they failed.

Interestingly, the standard 0402 through 1206 surface-mount resistors can all dissipate the same power (at a given hot-spot temperature) if you heatsink their end-caps to fat copper pours. So you can really push small resistors if you're careful. Conversely, poor cooling can kill a part even within its datasheet specs.

Sometimes we push parts, especially semiconductors, well past their abs max ratings... as much as 2:1 on occasion. It's a tradeoff between risk and performance.

Testing to destruction is a great way to get an idea of what your actual margins might be. It's fun, too. We routinely fry parts to see what will happen.

John

TTL would actually not fail until ~7V

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

That sounds right, long time since I've used TTL. But I wouldn't run it at 50% of 7V.

I wonder if anyone has written a book on such topics? Would be nice to have some reference.

It'll work there, but you won't get much of a VH ;-)

...Jim Thompson

Theory and experimental result of failure mechanisms are discussed in a book by HP, "Optoelectronics Fiber-Optics Applications Manual," (2nd ed.) These include flexing of the bond/wires, if memory serves. But there is an entire chapter on the subject there. It's a little old, but I suspect that although there may be some new additions to the list the ones listed there still remain in some fashion.

Jon

Thanks, I'll look that up.

Do you use Olive Oil for deep fry?

es

the

is

is

Generally the abs max compress from the process/reliability people, not design. In other words, you are given a blah blah blah voltage process in which to design. Now you could do something stupid and achieve less than absmax.

It's a little worse than I describe since the absmax isn't an over temperature spec. Worse yet, it varies from semi to semi. I've worked at companies where they would check functionality at abs max, which I always thought was dumb. That is, why stress the part. [The wafer itself had it's test pattern flogged to insure devices are working at absmax, no diodes are breaking down at lower than spec voltages, etc.) Some do a QA test at absmax, rather than stress every part. Some just do a lab test to look for obvious design error that make the part not function at absmax.

The book should be fairly easy to find. I think HP put out a bunch of them at one point (gave them away, even.) 1981, I think, is the pub date. I'd be willing to try and copy out parts of that section for you as a personal matter but I suspect you can get your own copy without much fuss or money. It will definitely give you some things to follow up on.

Jon

OK, with proper safety precautions, no doubt.

One criterion is how much the part costs or how much you can afford to spend.

Murphy says that if the part is really expensive, you will probably perform the "to destruction" experiment even if you didn't want to.

I've seen them online for 40$+ bucks but I thought I saw it used for 4$ somewhere ;/

Try:

I think I saw some cheap ones there. Conditions were "good." They also seem to list a couple of pristine copies, but _expensively_.

Their standard shipping rate is $4 (a penny less) for all sources of books so a $4 book will set you back a real $8.

Jon

-- An equal right to an opinion isn't a right to an equal opinion.