Hi I've built a pcb and housed it in a small metal enclosure. I've run the box for two hours and opened the enclosure to touch the components on the board. The hottest components the two DC-to-DC converters on there. They are warm to the touch, not hot. In normal operation, this box will be running without shutdown for indefinitely. How do I know if I need cooling? Thanks!!

 

You need cooling if the temperature exceeds the maximum operating temperature.
If after two hours you do not burn your fingers then you should be ok.

 

But be sure the components were working under their worst case operating condition for some time just prior to you checking the temp.

For something that is critical or for production, you could mount some thermometers on the key components and then datalog the temperatures as the thing is put through its paces. Then you use the thermal performance data for each of the components to determine if it is getting close to its limits.

 

I once worked with an engineer who calibrated his finger to well within 5°C so he could just touch a component and tell you it's temperature.

Two hours in still air with the cover on is essentially steady state. While I forget his exact method if you can leave your finger on it then you have no worries. I believe his scale maxed out at about 70°C (where you had to remove your finger least you burn it).

 

If you put a drop of water on the component surface, and it vaporizes instantly, then it is too hot!

If it can be touched permanently, it's OK.

If it is too hot to touch, then maybe a problem, cooling might be required.

Generally temperatures should not exceed 70C. And you need to take care for slow creep up.

But OP tested for two hours already.

A picture + information about the currents/power levels would be helpful.

 

You should test it at the highest ambient temperature the device will ever see.

 

You should test it at the highest ambient temperature the device will ever see.

Or just correct each reading for the higher ambient..
If all temp tests are done at 25 deg C and it could be in a 50 deg C ambient just add 25 deg C to all measurements.

 

Or just correct each reading for the higher ambient..
If all temp tests are done at 25 deg C and it could be in a 50 deg C ambient just add 25 deg C to all measurements.

That is not safe since if the ambient temp is higher, the ability to carry away heat might be lower.

25C ambient and you only may get 50C creep after many hours.

55C, and it might get higher than 80C.

 

I think this is a case of over-engineering.

After two hours the op touched the components and they were warm.

Case closed! (Done at post #2).

 

I think this is a case of over-engineering.

After two hours the op touched the components and they were warm.

Case closed! (Done at post #2).

So it doesn't matter if the box was running under idle conditions for the last hour? It doesn't matter if there was no load attached to it?

I still say it is important for him to run it under worst case operating conditions, as best he can. There's no indication that this was done.

 

That is not safe since if the ambient temp is higher, the ability to carry away heat might be lower.

25C ambient and you only may get 50C creep after many hours.

55C, and it might get higher than 80C.

Totally safe and accepted by ALL standards/approval agencies. It's done ALL the time in the professional world.

 

Totally safe and accepted by ALL standards/approval agencies. It's done ALL the time in the professional world.

OK good for these agencies. Some would perform a test at actual expected temp. And I have tested for temp. developement myself, so I can speak of temperature creep. Not just switch on, measure a few minutes. You need to test hours.

How will it be if the PCB has heated up, slowly, can it still carry away the same as cold PCB?

And things that are sold commercially do blow up, that is a fact.

How many appliances indeed will have a certification for 50C is another interesting question.
Interesting that hypothetical ALL agencies would add 25C, but indeed not many appliances would have a guarantee for 50C.

Maybe expensive gear in the professional world will have such a guarantee in written.

Professional here maybe means medical, Avionics, Nuclear and the like?

 

In case of doubt, use eletronic temp. monitoring, or forced cooling.

OP needs to measure and give numerical information about temp.

 

And I have tested for temp. developement myself, so I can speak of temperature creep. Not just switch on, measure a few minutes. You need to test hours.

How will it be if the PCB has heated up, slowly, can it still carry away the same as cold PCB?

The time it takes for a device to stabilize thermally and ambient correction are 2 totally different animals.

Once a device has stabilized, correcting for elevated ambient operation by simply adding the ambient delta to the thermal results is 100% acceptable and very reliable.

 

I think this is a case of over-engineering.

After two hours the op touched the components and they were warm.

Case closed! (Done at post #2).

Not really, electronic power devices can go into thermal runaway, where they are "warm" under ideal conditions but once they get too hot from ambient or load issues the resistance of power devices goes up and they get even hotter again, and you get that vicious circle happening.

When I was working in electronic repair it was common to use blankets to test or simulate thermal runaway faults.

Since the OP asked about a sealed container with DC converters within it that will be run continuously I think it's wise to test it really well, and if it's for a commercial product that's another whole thing again!

 

The time it takes for a device to stabilize thermally and ambient correction are 2 totally different animals.

Once a device has stabilized, correcting for elevated ambient operation by simply adding the ambient delta to the thermal results is 100% acceptable and very reliable.

Ambient correction is theory.

Thermal creep is real world. It can cause slow degradation + eventually even thermal runaway. You never know the critical trip point when it starts moving towards that. Until you run extensive tests + engineer it very well.

Line of last defense is often a one-blow thermal fuse.

If it is just 50C and the circuit is well understood then let say, forget about it...

 

Ambient correction is theory.

Thermal creep is real world. It can cause slow degradation + eventually even thermal runaway. You never know the critical trip point when it starts moving towards that. Until you run extensive tests + engineer it very well.

Line of last defense is often a one-blow thermal fuse.

If it is just 50C and the circuit is well understood then let say, forget about it...

Maybe its a language barrier thing but I'm not sure where you are going with this "thermal creep" thing. I thought you were using it to mean that it takes time for a product/components to reach "thermal stability" when turned on/operating at 100%. Which I fully accept/agree with. (typically 2-4 hours for an electronic device..I've sat through and performed multiple thermal testing sessions) Now though, I'm really not sure what you are talking about.

Are you familiar with the "bathtub" curve for electronic system failure?

Again..mathematically correcting for a higher ambient approval IS accepted/done by all "nationally recognized testing laboratories"/standards organizations.. including UL, Intertek/NTS, IEC,etc... Its not a "theory" its based on math/proven testing.

 

Maybe its a language barrier thing but I'm not sure where you are going with this "thermal creep" thing. I thought you were using it to mean that it takes time for a product/components to reach "thermal stability" when turned on/operating at 100%. Which I fully accept/agree with. (typically 2-4 hours for an electronic device..I've sat through and performed multiple thermal testing sessions) Now though, I'm really not sure what you are talking about.

Are you familiar with the "bathtub" curve for electronic system failure?

Again..mathematically correcting for a higher ambient approval IS accepted/done by all "nationally recognized testing laboratories"/standards organizations.. including UL, Intertek/NTS, IEC,etc... Its not a "theory" its based on math/proven testing.

Are you seriously saying I read "thermal stability", and can't understand it?

Interestingly you write"nationally recognized testing laboratories".

We don't have any US made domestic goods here in Europe.
Even if some domestic goods have UL logo.

Show me something written for domestic goods that certifies approval for elevated ambient temperature.

Anyway, I think we can end our discussion here. It's not covering a project we are participating in. We don't even know the test subject from the OP!

But it's interesting to know the subjects of mathematical temp. correction- isn't it rather professional equipment would use specially rated components?

Take for instance cars- they don't just add 25C and then it's approved. It is maybe one step among many others when doing tests.

 

18 posts later the op is sitting back and chuckling, see what I started

 

Are you seriously saying I read "thermal stability", and can't understand it?

Interestingly you write"nationally recognized testing laboratories".

We don't have any US made domestic goods here in Europe.
Even if some domestic goods have UL logo.

Show me something written for domestic goods that certifies approval for elevated ambient temperature.

Anyway, I think we can end our discussion here. It's not covering a project we are participating in. We don't even know the test subject from the OP!

But it's interesting to know the subjects of mathematical temp. correction- isn't it rather professional equipment would use specially rated components?

Take for instance cars- they don't just add 25C and then it's approved. It is maybe one step among many others when doing tests.

No..Im saying I read "thermal creep" and have no idea what you are talking about or using those words to reference about this post...

Not sure where the stipulation of "domestic goods" fits into this either.

As for "products" holding certifications for high ambients look at many motors, solar equipment, or any piece of equipment for a communication/telephone system (switches/routers/dslams,etc..) Often the equipment is stuck into little more than a box outside baking in the sun on top of a hill. Its been certified for the higher ambient that environment will see due to solar loading,etc... And the way that approval is granted is by running thermal tests and taking the final numbers and adding the delta between the test environment ambient and the expected real world ambient to each of those thermocouple points. If those final numbers don't exceed the rating of each of the components then..poof..certification is granted for higher ambient operation.

Heck I've even been part of some joke of an international test (IEC/EN/CE) where a thermal chamber was used.. BUT all the do is stick your product in there and simply turn it on so the power LED comes on or whatever (not force it to 100%, 50% or even 10% fully loaded). Then after x amount of hours at maximum "rated" temperature they open the doors and if the unit is still on its considered passed/approved.. That's just crap and proves virtually nothing there. Big difference in something just being on and something running at its maximum. They also do the cold side too which was neat..They can actually make it snow inside those thermal chambers.

I'm done.. I do this stuff all day long and don't need to argue with you about how it works..