How hot is too hot for a heat sink ? Wet you finger and hear it sizzle?

[ez24]

"Wet you finger and hear it sizzle" is from Art of Electronics 2nd ed.  I would rather use a IR meter that I got from Harbor Freight.  So as not to get burned, has anyone compared their finger sizzling to a temp gauge?

I am testing little buck dc to dc modules with tiny heat sinks.  The first test was at 3.3 v 2 amps and I could hold my finger on it for 3 seconds (145 F).  The second test at 12v 2 amp, I did not have to remove my finger (much cooler 125F ).  But I want to see if I can burn out the modules so I would like to run them to "sizzle".

But again I think I would rather use my IR than take a chance at burning my finger 

thanks

[c4757p]

The wet-finger test is very common. Perhaps not in electronics, I suppose? My mother always used it to check if the iron was hot, I remember. As long as you're smart enough to know what "wet" means, you won't burn yourself.

[Asmyldof]

Too hot for the heatsink?
Say about 15 degrees below melting should be your peak.

Too hot for whatever is attached to the heatsink? Ah, more complicated.
Your alternate question: wet-finger-sizzle should start at around 80 deg C, work well at about 95, and start to be painful anyway at around 130deg C, I'd estimate.

Back to the too hot question:
That's a matter of the flow of heat. Your converter transistor or chip connected to it creates heat, this flows away into the heatsink through a certain path. These chips are usually advised to stay under 130deg C. But, they will be significantly hotter than the surface of the heatsink. This is because the materials between the chip itself and the heatsink are hindering the flow of heat, or insulating it thermally.

That's why we want high quality paste on CPUs, if the paste is better, more energy can pass through it per unit of time, limiting the increase of the chip's temperature.

Then there's a small graduation along the heatsink, the tips will be slightly cooler than the point it touches the chip.

All in all, I'd say, for a small heastsink and a normal type DC/DC chip, that 70 to 80 degrees Celcius on the heatsink's surface will be the approximate temperature where the chip itself will start self-destructing by crossing the upper limit and snowballing.

[suicidaleggroll]

Checking for sizzling water isn't going to accomplish much unless you're nearing 100 C, which might be well above the chip's operating temperature once you take into account the thermal resistance of junction to surface and surface to heatsink.

I typically grab the heatsink with my bare skin after it's reached steady state.  If I can hold it for 10+ seconds without pain, it's fine.  If not, it's too hot and needs to be modified.

[Howardlong]

Checking for sizzling water isn't going to accomplish much unless you're nearing 100 C, which might be well above the chip's operating temperature once you take into account the thermal resistance of junction to surface and surface to heatsink.

I typically grab the heatsink with my bare skin after it's reached steady state.  If I can hold it for 10+ seconds without pain, it's fine.  If not, it's too hot and needs to be modified.

+1.

[ConKbot]

Dont rely on an infrared thermometer too much unless you can adjust the emissvity setting on it.  Even then it wont be that reliable for low emissivity items.  We had a Fluke IR imager at work, and I was trying to measure the temp of a heated mandrel made from aluminum. The machined surfaces were nice and shiny, and had a very low emissivity, the saw-cut end had a higher emissivity due to not being as shiny, there were reflections off of it despite it being 350F, etc... The only way I could get it to take a even remotely accurate spot reading was to put a thermocouple on the item, and tweak the emissivity setting on the imager until the temp for that particular surface matched the thermocouple.  Otherwise the best I could do would be to estimate "somewhere between 120 and 500"

TLDR: IR based thermometers and imagers are sketchy at best on aluminum items, as the emissivity changes wildly depending on the surface finish.

[Ice-Tea]

Dont rely on an infrared thermometer too much unless you can adjust the emissvity setting on it.  Even then it wont be that reliable for low emissivity items.  We had a Fluke IR imager at work, and I was trying to measure the temp of a heated mandrel made from aluminum. The machined surfaces were nice and shiny, and had a very low emissivity, the saw-cut end had a higher emissivity due to not being as shiny, there were reflections off of it despite it being 350F, etc... The only way I could get it to take a even remotely accurate spot reading was to put a thermocouple on the item, and tweak the emissivity setting on the imager until the temp for that particular surface matched the thermocouple.  Otherwise the best I could do would be to estimate "somewhere between 120 and 500"

TLDR: IR based thermometers and imagers are sketchy at best on aluminum items, as the emissivity changes wildly depending on the surface finish.

A thin piece of tape or a blob of paint will solve this (tipex works as well, btw )

[TSL]

Or Matt black high temp paint that car guys use to paint their exhaust systems is good for evening out emissivity.

 Adheres to nearly any surface and most of them are good to 600degC. Get a can of VEO if you need 800degC.

TT

[T3sl4co1l]

By the way, when you're snooping around with a thermal cam or IR point-and-click thingymabob, look for the highest temperature you can find.  Shiny is the reason: a shiny surface, with low emissivity, largely looks like its surroundings.  A mirror.  But even a mirror can give up its secrets, under conditions.  What you're looking for are acute inside corners: even if made of mirrors, such shapes reflect light down to the root, where it's eventually absorbed.  Thus making a pretty good black body.

Example: you might have a silvery aluminum heatsink with shiny chromed screws in it.  Most of the heatsink is useless because it's shiny and metallic.  But you can pick up a glint of strong emission from the cross on the head of the screw, and in the gaps around it, where it touches the heatsink.  Aha, interior reflection, black body emission!

The temperature may still be hotter than this reading yet, but it's at least a good start.

Lesson #1: a black (i.e., not glowing) object might not look hot, but you still don't want to go and grab it.  (Ouch!)

Lesson #2: a black (i.e., not emissive) object might not look hot, but if you can see some spots on, or connected to it, that look hot, you can at least guess what the rest of it is doing.  Maybe.

As for semiconductors, what they're rated for depends, but it's quite common to run heatsinks over 100C.  Semiconductors usually die above 150 to 200C, depending on packaging and design. The internal junction temperature is necessarily hotter than the case, which is, in turn, hotter than the heatsink.  Calculating how much of each requires knowing only the power dissipation and thermal resistance between each pairing.

So do be careful -- 150C isn't that much hotter than 100C, and transistors may be near or into the failure range under that condition.  But the other point is, if it's a responsible design, it is absolutely possible to operate like that, without it being dangerous.

Tim

[CatalinaWOW]

While many silicon devices can operate at die temperatures of 125C or more, reliability suffers as temperature goes up.  If you are doing something where a wet finger test is an adequate measure then this probably isn't an issue (unless you are building a supply for your dream lab that you expect to use for the next 25 years).  If on the other hand you are designing a railroad speed controller you will probably want to pay a lot of attention to reliability and may want to change your heat sink design to keep temperatures lower.

I actually would not operate hobby equipment that hot, simply because over time I would forgetfully set a hand or arm or elbow on the device and burn the crap out of myself.  But you may be more mindful of such problems.

[ez24]

Your alternate question: wet-finger-sizzle should start at around 80 deg C, work well at about 95,

These chips are usually advised to stay under 130deg C

From the datasheet http://www.xlsemi.com/datasheet/XL4015%20datasheet.pdf
The max junction temp is 125C  and I assume it is up to the user to figure out how not to exceed this because 125C at the junction will be hotter than on the heat sink.

This is what I am testing -->  http://www.aliexpress.com/item/Wholesale-1pcs-DC-Step-Down-Converter-DC-4-0-38V-to-1-25V-36V-5A-75W/32285636716.html
For many days I have run it at 2 amps at 3.3, 5, and 12 volts.  It puts out the most heat at 3.3 v at 2 amps

I typically grab the heat sink with my bare skin after it's reached steady state.  If I can hold it for 10+ seconds without pain, it's fine.  If not, it's too hot and needs to be modified.

I moved the IR meter all around until I found the max temp which was 56C  (at 12v it was 45C).  See picture.  I was only able to hold my finger on the fins for 3 secs.  So this failed the 10 sec rule. 

This module has been doing very well and actually puts out better ripple than the input (subject of another post soon).

I have used a thermistor to double check the IR and they agree if I am careful.  I have to move the IR around a lot to find the max temp.   Somewhere I have some high temp black paint and if I can find it, I will experiment with it.   Do you think black paint will help the heat dissipate off of aluminum?

I actually would not operate hobby equipment that hot, simply because over time I would forgetfully set a hand or arm or elbow on the device and burn the crap out of myself.  But you may be more mindful of such problems.

In my case this is good advice.  I do not think there will be cases that I will be running 2 amps at 3.3 volts (max temp) but I will accept 56C even though it is hot.  It takes a little effort to get to the heat sink because of the other components.

wet-finger-sizzle should start at around 80 deg C,

I will use 80C as my max temp when I test the module at higher temps as an experiment when I can find my paint.  (it was $25 a pint).  Then I will test the wet sizzle method.



thanks everyone

[Asmyldof]

80deg C isn't really high temperature paint domain quite yet, though close.
80deg C is very much compatible with hobby store acrylics and the like. And still also the suggested correction fluids.

In fact for below 80deg temperatures you might look for thin-coat or single-coat paints instead (which then again rules out hobby paint and correction fluid), as 0.1mm of black polymer based paint will impede the energy dissipation more than 0.01mm of black polymer based paint.

To be honest, while I have 800deg C paint, I have never really looked at the composition, so it might be thin film or some reasonable heat-transfer thing.
I have also never tried intentionally heating it beyond 200, I really should.

(and I don't mean the can, though it might be fun as well)

[saturation]

Heat receptors in your skin kick in at about 40-50^oC, in which case you'd describe it as hot.  As temp rises above 50^oC, the degree of discomfort depends on you.  Damage occurs at >80^oC and up, and you'll start to feel heat related pain >65^oC.

http://users.rcn.com/jkimball.ma.ultranet/BiologyPages/P/Pain.html

This relationship is helpful because many max operating temps, be it passives or actives, as a rule of thumb are operationally optimal ~100-150^oC, and keeping the operating temperature about half is just at the painful touch component.

Thus, as a rule of thumb, if your heatsink or casing feels hot but not enough to let go, there is a very good chance you have enough heat sinking.  Once it becomes too hot to touch, you have to check your thermal calculations carefully.  If it boils water, its >=100^oC and the junction will be far hotter too.  Note this is all tied to room temp, aka ambient, too, in order to dissipate the heat if its just passively radiated.

A caveat, if you keep your items cool, regardless of its maximum operating temp in the datasheet, which means good enough to touch, it will not suffer thermal related aging stress.  Even if you device can operate well above 'touch', it may not live long in terms of operational hours. 

Many products designed for longevity will often have thermal management were sinked items are touchable at quiescence and room temperature, e.g. HDTVs power drivers or music power amps.  So you can use just your fingers to troubleshoot thermal related issues or diagnose possible failures.