Re: Operating temperature derating

I don't think that there are any standards that manufacturers use to make datasheet abs max specs.

I like to test parts to destruction and estimate what they can actually do. For example, max voltage ratings can sometimes be safely exceeded, or a surface-mount resistor can dump a lot of power if it's well heat sunk with copper pours.

Thermal imaging can quantify hot-spot temperatures.

Pushing parts past specs is only a good idea if there is a serious performance upside.

RF parts are especially likely to be under-rated. Like a 2 volt rated schottly detector diode that starts to leak at 7. Or RF amp parts that actually run at 2x rated max voltage in normal use.

GaN fets are interesting. As are mosfets that avalanche at some unspecified voltage.

A bit of air flow can really change things. I assume that most specs are based on still air.

I've seen a few parts that were unreliable within their abs max specs and needed to be derated from there.

Blow up some parts. It's fun.

Some fellow on Youtube a while back managed to get a kilowatt out of a TO-220 packaged device rated at 100W by blasting it with plumber's Arctic Spray.

Probably because it's common in RF to end up with the part being subjected to voltages double its normal supply rail figure.

Best fun you can have with your clothes on as they say.

When I use RF parts in time domain/pulse applications, I have to make up my own rules.

It's rare to get even basic DC curves on RF parts. The data sheets say, literally, adjust the bias until it works. Or AC couple in and out and don't ask questions about "volts". Which opens up the possibility of novel biasing schemes.

Clothes on and safety goggles.

Yeah, RF biasing - well, everything to do with RF in fact is a blast, isn't it? No wonder the guys at the cutting edge of it have always coined the BIG BUCKS.

Indeed. It's amazing how many decibels you can get out of even a 10uF electrolytic.

No wonder modern electronics - consumer electronics I mean - doesn't last five minutes. :(

There are two basic limits to operating environment temperatures.

Junction temperature limits are usually proscribed below book Tjmax for an agreed mtbf. Sometimes the environmental limits can be extended under reduced performance guidelines, without exceeding agreed Tj limits.

For a whole product, there's usually other component limiting factors that occur first - electrolytic and film capacitors, insulation system ratings and surface touch restrictions.

Performance is measured with these specified limits in mind.

or not.

RL

The guidelines are derived from empirical data and accelerating factors are determined. The arrhenius equation comes to mind.

Predictors determine the design guidelines, but, yes, varying degrees of physical testing is a standard element in design and product verification.

Commercial considerations of free enterprise have always pushed human behaviour towards banditry, so what can I tell you?

RL

What exactly IS your concern?

RL

That's entirely wrong. Whether you make spears or helicopters, you need repeat business.

Communist thugs are the genocidal bandits. They have a captive market and no competition.

We’ve had chipmaking folks here (including Miso, iirc) say that the abs max ratings are primarily marketing driven, like most of the rest of the datasheet.

You find out what your major customers are going to do with the part, and derate it to match. That way you sell parts and still get to have a quiet life.

Of course there are things like delamination, overvoltage and electromigration that cause lifetime issues rather than prompt failures.

Cheers

Phil Hobbs

Given that you expect to get what's promised when you pay for it, outside of banditry, I only expect performance and a reasonable service life.

I often kick mysellf in the head when I realize that shortcomings in product performance were actually predicted in the written specification - that the performance that I was expecting was not only outside published spec, but might not be physically possible, using current materials and techniques.

RL

Tell it to the bandits. There are plenty of suckers out there and paper/internet bumph easily generated to suit them.

Compared to the large-scale kleptocracies that replaced them? . . .

RL

You can buy Sony or cheap no-name Chinese junk. Your choice. Would you prefer to not have a choice?

The US and Europe have antitrust laws that pretty well ensure multiple suppliers and competition. The real kleptocracy, in the USA or in Russia or in China, is the government.

If you don 't want to buy a car from Honda, GM, Ford, Toyota, Kia, Tesla, VW, or BMW, walk in the rain and snow like most of your ancestors did. Few of our ancestors could afford a horse.

I used to be in the iPhone design team. At the time we used a 40 deg C as the maximum ambient temperature.

As part of the development we would run a "thermal virus" software to cause the CPU to dissipate an approximation for the maximum possible.

Under those conditions the internal temperature could get to the 70 deg C region. If excessive temperatures were reached the CPU would be throttled to avoid damage.

I was working on the display/touch hardware; LCD displays stop working at about 75 deg C (they just turn black). Since the display was within a couple mm of the CPU there was not margin.

Under less stressed conditions the internal temperature was much lower.

I suspect that the battery is probably the most sensitive item for storage temperature while not operating, especially if fully charged.

kw

In the US, the govt ensures that wall street is not inconvenienced. Assets are stripped to pay dividends, labor juggled for a similar benefit. Industry in north america is just not delivering.

You're forgetting BYD, Brilliance, Chery, Changfeng, Geely, Great Wall or any of the european-name-brand subsidiaries in India.

I've only ever paid for one car - had to pay for the telephone pole it was totalled against, as well (it was that long ago).

Anyways - a bit of a non-sequitur , neh?

RL

Or maybe you are the United Fruit Company.

Not the only ones.

You'd probably find that surface touch temperature limits are exceeded before that - screens having direct access to the external environment.

LCD screens also become unreadable at ~ -20C. Again - don't confuse it with the Fahrenheit scale.

He's talking about aging, loss of capacity and charging voltage tolernce restrictions. Lithium battery 'cook-off' temperatures are much higher.

Industrial grade components and equipment are designed for higher operating ambients - but equipment designed for personal (hand-held or pocket) use assume the limits for human physical comfort and safety.

Vehicular operating environmental limits differ from consumer equipment standards.

What's your actual problem?

RL

It has something to do with the human body at reat, in the shade, failing to be cooled by sweat in moving air at a certain temperature. Cooling has then to be obtained by artificial means. It probably IS a 'nice round number', as some are affected earlier than others. A body will already be avoiding direct sun exposure with hats / umbrellas, shade trees etc.

Sorry to hear about your buddy. Senior citizens or health compromised people have to take special care in the heat.

How do you know that was the problem with the phone? What were the phone's symptoms?

Already said - automotive component temperature limits are set by other standards. That being said, it's a test of your vehicles integrity to do hard work at elevated temperatures. For the benefit of passengers, some attempt is made to keep the cabin below 40C.

A company will have stated limits for a project in the earliest design stages. They will use limits expected by their industry, unless there are reasons to do otherwise. Stupidity would be a pretty bad reason.

Are you serious? Touch temperature is the surface temperature in contact with operator, usually the hand. If it's not meant to be held, you can get away with visible surface temperature warnings. Anything made to be touched doesn't get an exemption.

Safety standards are pretty explicit w/r to operator environment.

Humans do not, in fact, operate well over any environmental extreme. That's why we wear clothes; live in temperature-suitable dwellings, don space suits etc.

Commodities respond to market requirements.

I think, at this stage, you're just being fractious.

I'm out of here.

RL

On 6/12/24 12:26 PM, Don Y wrote: <...>

The phones may not stop working but may not meet performance specifications. For example applications may not work as fast as at lower temperatures.

I've never heard of lower level software being compromised at high temperatures so basic phone calls would be ok but action video games would run slower.

The ASICs I worked on had a -25C to +75C operating range. When the CPU was idle the phone would be too hot to hold if the internals were at 75C.

Automotive devices that expect to be in the engine compartment are designed for -40C to +125C or +150C.

kw <...>