Resistor Sizes Vary Greatly?

[schmitt trigger]

And I would venture to say that ZERO people on this forum are doing anything in volume!  Let's not kid ourselves and be real for a minute!

- Wil

The highest volume product I was involved with, that was utilized across the whole vehicle spectrum of one of the Big-3, sold 2 1/2 million units a year.
The board even had dual footprints for a relay, to leverage the cost between two different relay manufacturers.

[floobydust]

It's interesting to think that there are a few people who think world's largest resistor manufacturer publishes deliberately wrong data and scams their customers, while their actual customers, who buy these parts in billions, have nothing bad to say about the parts, nor any difficulty designing them in.

Think about it, and it's actually pretty close to flat earth tin foil stuff.

Everyone fails to document important details in resistor specs. A huge number of resistors are carrying a pulsed current these days. Just how big can that pulsed current be before you are obeying the overall, say, 1W rating, but causing such intense momentary heating that bad things will happen? I never see that properly defined.

Never? For example, that very resistor posted by OP, see datasheet pages 5-6: https://www.yageo.com/upload/media/product/productsearch/datasheet/lr/YAGEO%20FMP_datasheet_2023v4.pdf Or do you mean this is somehow improper, if yes, what's wrong with it?

I just looked through the data sheet. That one does seem to properly address the concerns of a designer of modern systems. Well done YAGEO.

Please provide any numbers for OP and the thread, compared to armchair opinion- What power would you use the FMP100 for? How hot? Have you heard of math?
Experts in the beginner's thread kindly give a usable power dissipation, temp rise even for operating the part at 1/2 or 1/3 of rated P₇₀ of 1W. 

OP's Yageo does a scam-eo by not providing heat-rise/thermal resistance data for their FMP family. Sorry Yageo fanboys "world's largest resistor manufacturer" means sweet FA. Is it incompetence or deception that the data is missing?
Ohmite does the same - no heat rise/thermal resistance for their parts.

Perhaps it needs tinfoil fashioned into a heatsink, unless you can provide any numbers. Something engineers like to work with.

[floobydust]

Vishay response about the resistor test jig used:
"The resistor is mounted on a PCB (see Fig) which is set up vertically, with the resistor horizontal. Using an infrared camera, a thermal image is made of the resistor, thus defining the hot-spot and solder-spot temperatures."
---------------------

To me it looks like max convection cooling and a lot of copper too. Not mentioned but the test board I believe is EN 140400 (for SMT is 2.3.3: FR4 100mm x 65mm x 1.4mm, 35μm Cu, pad 2.0mm width).
So it's a rosy scenario compared to a horizontal PCB in an enclosure.

[Siwastaja]

some of the products that I designed and produced are used even by multi billion dollar organizations AFAIK.

Same here; I can say I have designed specialty electronics for some of the largest corporations on the planet; and while it sounds fancy, it really isn't and it happens easily! Those multi-billion organizations tend to be quite paralyzed due to sheer size, and outsource a lot of experimental work, through weird and long chains, at end of which some random guy they have never heard about builds something random in a week and gets paid $5000 for the job, and is never heard again. Such projects end up in some drawer, but allows these large corporations to explore ideas which are impossible for them to explore due to their internal bureaucracy.

On the other hand, if you build say some interface box (let's say USB-CAN interface) or a programming jig or similar essential tool, chances are it will be sold in tens of thousands and even largest of corporations buy it, or at least their engineers buy them bypassing their purchase departments. Many hobbyists have succeeded in becoming "professional" large producers this way with relatively simple products.

[Siwastaja]

So it's a rosy scenario compared to a horizontal PCB in an enclosure.

There is nothing "rosy" in it, it's the standard way of rating all components. You as a designer are responsible for understanding what Ta means. It's not the temperature of your house, or temperature of Antarctica. It's the temperature inside your case, around the resistor (what else could it be?), and you are the only one who can calculate/simulate/test/measure the thermal performance of your case. It''s a ridiculous idea that a resistor manufacturer would start guessing how typical boxes work and somehow nudge that number into their ratings. Instead, a responsible manufacturer arranges the measurement so that such application-specific conditions are removed from the equation. Selling a part with pessimistic ratings because a test mimicked blocked airflow in a tight case serves no one's interest.

Besides, the amount of copper seems very fair, it is very easy to use more copper than that, for example there seems to be no contiguous ground plane at all, and a 4-layer board would perform significantly better. If they wanted to con you, they would do that.

re vertical, understanding convection is basics for a designer. If you opt for a fan-less design, then you need to be pretty damn sure to provide room for air flow with very little pressure drop; if the board is vertical, don't place anything sensitive above the hot part, and remember not to block the air flow from below either, and make sure the product is operated at correct orientation. If the board is horizontal, consider adding cutouts or cooling holes under the resistor; they are usually free-of-cost. All of this works exactly the same even if you use a larger resistor with smaller power density!

But I guess you are beyond salvage. You are blinded by the idea of a conspiracy and see clear evidence of there being none as there being one.

[shapirus]

It would be interesting to see what will happen to that small resistor when there is actually 1W of heat generated in it. Will it burn? Will it crack? Will it glow red and keep working? Will it survive, but change color? Will it retain its original resistance after cooling down?

Testing this, however, will require a power supply that can provide 68.5V (and then make it lower as the resistor heats up), which is on the less common side.

@Wil_Bloodworth do you have a lower value resistor of the same series? For example, a 150 Ohm resistor will require only 12V to produce 1W of power.

[wraper]

It would be interesting to see what will happen to that small resistor when there is actually 1W of heat generated in it. Will it burn? Will it crack? Will it glow red and keep working? Will it survive, but change color? Will it retain its original resistance after cooling down?

Testing this, however, will require a power supply that can provide 68.5V (and then make it lower as the resistor heats up), which is on the less common side.

@Wil_Bloodworth do you have a lower value resistor of the same series? For example, a 150 Ohm resistor will require only 12V to produce 1W of power.

Resistor marking may discolor and PCB pads may delaminate over time depending on their size, material, resistor mounting height, cooling. In any case, generally you don't want to run parts at 100% of their power rating, it reduces reliability significantly.

[shapirus]

Resistor marking may discolor and PCB pads may delaminate over time depending on their size, material, resistor mounting height, cooling. In any case, generally you don't want to run parts at 100% of their power rating, it reduces reliability significantly.

It's so small that running it at 500-700 mW would be just as interesting.

[Siwastaja]

By all means test it, but I have boring news to you: nothing special happens. Manufacturer tests them at rated power given the test procedure image shown. You can, of course, wrap some paper around this resistor and make the paper discolor or even ignite, or block the convection cooling or thermally insulate the resistor and make it fail, but

Discoloration of the resistor may or may not happen and does not indicate failure because the markings exist for part identification during assembly.

If it fails, it's big news. So far, people who talk about resistor power rating conspiracy have produced zero evidence, even when the test is really easy to arrange. Absolutely zero, not even circumstantial. If there is something to it, and anyone proves it, I'm happy to stand corrected and join your camp.

Of course, some derating is almost always a good idea. With components for professional designs, manufacturers do not much or any safety margins as that would only fudge the reality; it's a convention that this task is for the designer to do (it could be the other way, too, but this is what the convention is, and it only works when everyone follows the same convention).

If you want to criticize an almost con-like manufacturer convention, look at how tantalum capacitor voltage ratings work, and how manufacturers recommend 40% derating in separate documents because the parts are literally rated to some dozen hours of lifetime, and after that they burst in flames violently! Another commonly used front page specification which carries no practical purpose and exists only to mislead young players are MOSFET ratings at Tj = 25degC.

[floobydust]

It's so small that running it at 500-700 mW would be just as interesting.

Nobody here will stick their neck out, give a number or calculation of the usable power dissipation for our old friend the 1/4W body resistor.
I get a good laugh out of that, armchair experts galore that can't put out.

Soon OP's 1/4W sized part will be rated 2,3,4W... and it ends with something called A LIGHT BULB 

[wraper]

If you want to criticize an almost con-like manufacturer convention, look at how tantalum capacitor voltage ratings work, and how manufacturers recommend 40% derating in separate documents because the parts are literally rated to some dozen hours of lifetime, and after that they burst in flames violently! Another commonly used front page specification which carries no practical purpose and exists only to mislead young players are MOSFET ratings at Tj = 25degC.

It's note like that really and the problem is not the lifetime, it's about failure rates. Technically manufacturers don't lie about their voltage rating. If you spot weld or attach them with conductive glue, that voltage rating is fine. However they get damaged during reflow and if they are run at high voltage, instead of gradual self-healing, they can burst in flames. Majority of them will last for a long time but you certainly won't like single digit % failure rates over a short time period.

[floobydust]

I find component manufacturers give maximum ratings. Was always taught, even 40 years ago to derate parts for long life.
Voltage-derating tantalums, power-derating resistors.
But nowadays it's a bigger trap because noobs and beginners don't really know this and the numbers are further stretched as competition heats up (pun intended).

A tech specified a replacement snubber resistor, original out of stock and into production went one of these "miniature" resistors. I wondered why it was too small. Of course it ran super hot, even though lead-bending was up off the board.
I phoned the resistor manufacturer (not Vishay) and flamed them and the product manager told me "um, well these parts are intended for fan-cooled systems". You couldn't say that on the datasheet?! "No because it's evaluated to legacy standards (convection cooling)." 

[shapirus]

By all means test it, but I have boring news to you: nothing special happens.

This is a prediction that is likely, but not certain, to be proven correct.

Manufacturer tests them at rated power given the test procedure image shown.

That's a good point regarding the test procedure. Is it in the resistor's datasheet? Does the datasheet specify test conditions (in particular, forced vs natural air flow, as mentioned in one of the posts above)? The answer is "maybe": it depends. If someone is curious enough to find the datasheet and see what's inside, then the test can be reproduced to see if the resistor can handle the claimed wattage and not only what we subjectively expect a "1W" resistor to be able to handle.

If it fails, it's big news. So far, people who talk about resistor power rating conspiracy have produced zero evidence, even when the test is really easy to arrange.

This is precisely why a practical experiment sounds interesting to me, especially considering all the prior discussion. And especially because it's the beginners section of the forum.

I have some small (in terms of size) resistors with a suspiciously high claimed wattage in my parts box. Today it's too late, but tomorrow, if I don't forget, I will test some of them, if I find suitable values to use with my lab PSU, and see what happens.

[Siwastaja]

I find component manufacturers give maximum ratings. Was always taught, even 40 years ago to derate parts for long life.

Yeah. Remember though that derating is not a silver bullet. It's just the cherry on the top; important but only part of the deal. The foundation should be understanding and then calculating/simulating/testing all mechanisms behind those limitations. When you handle that all, you can go pretty close to limits. You add derating to cover for uncertainties; the more uncertainties you have, the more you need to apply derating, but also the less confident you are whether it's truly enough, so you cover by adding even more derating.

Voltage-derating tantalums

Yes. But rules of thumb can be tricky. Aluminum electrolytic caps are formed at something like 130-140% rated voltage and actually sustain some overvoltage pretty well, and it's rare to see failures because somebody ran an aluminum elcap exactly at rating (failures go up only when significantly exceeded). Some MLCCs easily withstand +500% overvoltage without any reduction in lifetime. And then there are tantalums, which, as wraper correctly points out, fail in high numbers if used at rated voltage, after the soldering process.

[wraper]

Some MLCCs easily withstand +500% overvoltage without any reduction in lifetime.

Not true, they withstand 500% but often for not that long. I've seen reports of this feature used by scummy laptop services in Russia. When doing some inexpensive work, like cleaning dust and replacing thermal paste,  they also solder multiple low voltage MLCC in parallel to charging connector and then wait for returning customer to claim expensive motherboard repair is needed.

[Siwastaja]

I said some, this means designs that are clearly close or identical to higher rated parts, but for some reason (e.g. supply chain related i.e. demand) rated for lower voltage. I.e., those with suspiciously low ratings for package size.

If the part is 25V 1uF 0603 to begin with then it likely cannot take 100V. On the other hand a 6.3V 0.1uF 1206 part could take 63V. My point is, MLCC voltage ratings are quite arbitrary and sometimes there are huge margins before failure.

Soldering a physically large MLCC with excess heat and maybe giving it a mechanical bump would be a great planned obsolescence trick, too. Make it crack, measure to verify it isn't an instant short, and some time later it becomes a short, possibly aided by humidity.

[shapirus]

I have some small (in terms of size) resistors with a suspiciously high claimed wattage in my parts box. Today it's too late, but tomorrow, if I don't forget, I will test some of them, if I find suitable values to use with my lab PSU, and see what happens.

So the closest I had (which I'd say extrapolates well to the one in question) is a Hitano MFR-SS ("super mini size") 2W metal film resistor. 560 Ohm in my case. 9.0 mm x 3.0 mm.

Here's the test setup.



I kept it running like this for ~15 min. There was that typical smell of hot resistors, but otherwise, just as predicted by @Siwastaja, nothing special happened. No burning, no discoloration, no resistance change after cooling down. Temperature rise above ambient was higher than specified in the datasheet, but that's most likely due to the fact that the datasheet assumes (without explicitly stating it) that the specified temperature rise is measured when the resistor is soldered in a PCB, which acts as a heat sink, and the resistor's leads are shorter, creating a shorter path for the heat to escape.

The 1W resistors of the same series are smaller, 6.0 mm x 2.3 mm, and will probably be comparable to the one that was the starting subject of this topic.

[Siwastaja]

That's as expected. Such small resistor probably is at thermal equilibrium after just 15 minutes, so any more test time would be to prove absence in resistance shift or slow deterioration. More interesting would be see how it performs after, say, 10000 hours at this power. Since I realize the obvious difficulty testing this, maybe try accelerated aging test at 3W. If it is able to run at +50% of rated power for a few hours (or even 15 minutes), without changing the value or failing, that would give further confidence on the rating.

[shapirus]

That's an interesting scenario. However, my PSU goes up to only 36 volts, so no luck there: with 560 Ohms it'll be only ~2.3W, which isn't enough, and I don't have lower values of these.
Theoretically, though, resistors of this type are made from nichrome film over a ceramic core (so the datasheet says), so I would venture to guess that anything below glowing red isn't going to damage or degrade them (at least in a short run).

[floobydust]

Checked and found the same beliefs I'm mentioning here in The Art of Electronics:

"A word of caution: It is our belief that one should not place complete reliance on the kind of curves provided by manufacturers" ...
"Generally, though, it’s better not to “twist the dragon’s tail”; our advice is to derate resistors by 50%."
Paul Horowitz, Physics and EE prof. Emeritus at Harvard.
Winfield Hill,  Director of the EE Lab at the Rowland Institute at Harvard University.

Definitely give them a ring regarding my conspiracy on not blindly trusting manufacturer's datasheets 

I think the "50%" resistor power-derate rule was OK many decades ago, it might even go back to the days of point-point wiring with carbon resistors.
But nowadays it's not enough, on a PCB and with high temp materials, practical power is much less. Let's say 1/3 of P₇₀ looks good to me. Unless fan cooled. IS safety standards max. is 2/3 of rated power to be considered infallible.

SMT resistors are also something to look at, best in another thread as they have "hidden EE common knowledge" as I call it traps as well regarding power dissipation.
I was just reading about the infinite heatsink test (two copper blocks) while measuring surface temperature and look forward to that spec on a resistor datasheet 
It's needed for science, just as power mosfets use the same spec - but EE's need to know practical, achievable values.

[tggzzz]

"A word of caution: It is our belief that one should not place complete reliance on the kind of curves provided by manufacturers" ...
"Generally, though, it’s better not to “twist the dragon’s tail”; our advice is to derate resistors by 50%."

Definitely give them a ring regarding my conspiracy on not blindly trusting manufacturer's datasheets 

I think the "50%" resistor power-derate rule was OK many decades ago, it might even go back to the days of point-point wiring with carbon resistors.
But nowadays it's not enough, on a PCB and with high temp materials, practical power is much less. Let's say 1/3 of P₇₀ looks good to me. Unless fan cooled. IS safety standards max. is 2/3 of rated power to be considered infallible.

SMT resistors are also something to look at, best in another thread as they have "hidden EE common knowledge" as I call it traps as well regarding power dissipation.
I was just reading about the infinite heatsink test (two copper blocks) while measuring surface temperature and look forward to that spec on a resistor datasheet 
It's needed for science, just as power mosfets use the same spec - but EE's need to know practical, achievable values.

It has always been the case that you need to read and understand what is not being guaranteed/required[1] in a data sheet. No change there.

EEs have always needed to understand the theory of reliability and practical failure mechanisms. No change there.

EEs have always needed to work out how practical components will behave in their circuits. No change there.

Data sheets have always stated specifications involving ideal conditions[2] that won't apply in a real circuit. No change there.

[1] The legal profession also has to do the same, e.g. to get their client acquitted of charges, or to invent tax avoidance schemes.

[2] The physics equivalent is the famous "assume a spherical cow".

[floobydust]

Withholding measurement data is the same as lying by omission.
Nobody here can ballpark OP's yageo scam-eo resistor temperature operating at 1/2 power or anything reasonable.
How is this the EE's fault?

[Siwastaja]

That text you quoted is decades old. You once again proved yourself wrong on your claims that new resistors are a conspiracy. How long are you going to keep making yourself a laughing stock? I mean, the credibility you have left is basically zero. Keep wrapping more tinfoil and keep providing zero evidence.

Once again, I have zero problems using moden high power density resistors significantly over 50% rated power. Other professionals agree. No proof of otherwise, not a single bad experience by a professional who actually designed this stuff in has been shown. The burden of proof is on who makes the claim, and extraordinary claims require extraordinary evidence.

50% derating suggestions hide some assumptions (e.g. PCB orientation, possible blockage of cooling, how close multiple heat-generating parts are close together, etc.). It's a rule to ease thermal design for those who don't have time or energy to think about it. There is no need to follow such practice if you understand thermal design and can simulate or practically measure the performance.

Exactly as with all other component types, manufacturer cannot predict how the customer uses the part and tests how it performs in isolation. Just like you can drive faster when there is no other traffic, and thus cars have higher top speeds rated compared to what you actually can drive at.

[tggzzz]

Withholding measurement data is the same as lying by omission.

Caveat emptor is a useful principle.

If a component doesn't have a specification for X, then don't use it in an application where X is important. Either use a different component where X is specified, or qualify/test the component.

Classic example: MLCC where the voltage decorating curve is not specified.

There is no substitute for understanding and for experience.

[Siwastaja]

If a component doesn't have a specification for X, then don't use it in an application where X is important. Either use a different component where X is specified, or qualify/test the component.

Or work with the manufacturer to get the specifications you need; they know how to qualify/test their parts. Yes, there is cost to this. Basic principle of any business is offering different level services to different customer tiers who pay differently. This applies to components as well. If you are a poor hobbyist or startup, then work through the night and test it yourself. If you are a large business, just the information alone that you are willing to buy in millions will get the manufacturer to offer data just for you.