thermal life test

Thread Starter

jismagic

Joined Sep 7, 2013
77
I have been asked to do life test of electronic product for a period of 1 month.
The thermal profile is defined and can be programmed in thermal chamber.

I will have to monitor the product every 1 hour for its functionality.
i cannot sit and watch , so i decided to automate product monitor.

One question i have is - some of the output can load high current. do i need to load them 100% every 1 hour while i do life testing? can i test with no load?
No load makes my life easy

give your thoughts

thanks
jis
 

crutschow

Joined Mar 14, 2008
34,282
So is the device is continuously powered when going through the thermal cycles? I would think you would want to test with the rated load on the circuit. Is that a problem?
 

Thread Starter

jismagic

Joined Sep 7, 2013
77
Yes, the device is powered ON all the time. there are 9 products that go thru the life testing. there are 4 ouputs for each product , each one around 5A rated.
that means i have to switch ON and OFF 4*9=36 outputs . it is a kind of big mess for me. the power ratings of my design go high.

if i can do with no load, that makes my life much easy.
What goes wrong if you do life test with no load.?

thanks
jis
 

Brevor

Joined Apr 9, 2011
297
that means i have to switch ON and OFF 4*9=36 outputs . it is a kind of big mess for me. the power ratings of my design go high.
Been there/done that... My testing required me to test with static load and also dynamic load, at high, low and normal input voltages, at both -55 degrees and also at + 120 degrees. Testing with load is the only realistic way to test them.
 

ErnieM

Joined Apr 24, 2011
8,377
If you test with no load you are not testing the condiotio it sees in the real world.

Why turn the loads on and off? I've set up temp chambers for environmental stress testing using a single 5V switched output to control a relay. The relay sits in various power switches so I can cycle the input power of multiple devices. All loads are always connected.

Put a small fan on the load resistors if you are worried.
 

mcgyvr

Joined Oct 15, 2009
5,394
A whole month long test?

Some things to google for you
HALT/HASS testing
MTBF (bathtub curve)
burn in test

Most thermal tests for certifications are only required to be maintained until temperatures have stabilized while the device is 100% fully loaded. This can typically take from 1 hour to 3 hours for most any electronic device. Then you simply ensure that the temps of all components do not exceed their ratings like the UL RTI value for plastics, stated max IC temps, touch temperatures, etc...

And if you aren't loading it 100% or simulating whats possible in the real world then whats the point?

I do thermal testing with devices that can be pulling up to 2000 Amps..
You should see our "load banks" and power supplies.
My test lab gets to 100+ deg F simply from the heat being dissipated in the load banks. Its quite toasty :).. But we just measure the elevated ambient and correct as needed.
Our products get rated for up to 70 deg C ambients so we just correct all the measured values to account for that.
 

AnalogKid

Joined Aug 1, 2013
10,986
I think the longest test our stuff has gone through was 72 hours. Ramp from -40 to +85 in 30 minutes, soak for a fe hours, ramp down, soak, repeat. Tough test.

I agree with Ernie, why turn the loads off?

ak
 

ErnieM

Joined Apr 24, 2011
8,377
We have a relay endurance test that's been running continuously all year so far. Have 9 8 foot baseboard heaters in a storage unit just outside our building to provide a 9000 watt dummy load.
 

Thread Starter

jismagic

Joined Sep 7, 2013
77
thanks for all replies.

why turn the loads off?

if i have no load, i can avoid high current- so no heat :) , high wattage power supply, it will be easy to monitor if output really turns on.

I agree its no real life condition. but 3 months conitnuosly loads ON with 15-20A * 9 products is very scary.
i will be more worried about my test set up than my product if i do this :)

thanks again.

jis
 

mcgyvr

Joined Oct 15, 2009
5,394
now its 3 months?
How can you have a "thermal life test" if you product is just turned on not doing anything?
Whats the point?

Its "very scary" you can't understand that.
 

Thread Starter

jismagic

Joined Sep 7, 2013
77
if a output can load a max of 5A, if i just do the life test with a load of , say 0.5 A , does it affect the test result?

if a output can load a max of 5A and the product switch on/off the load based on some xyz conditions , does it still make sense if i verify the product still switch ON and able to turn OFF than verify turn ON and OFF with full load?

3 months is a typo - back to 1 month :) or many hours...

thanks
jis
 

ronv

Joined Nov 12, 2008
3,770
It might help to know what you are testing and the specs. What temperature are you planning on. It could be that the temperature you will test at will offer plenty of stress to your components with a light or nominal load. This sounds more like a life test as opposed to a destructive test.
 

AnalogKid

Joined Aug 1, 2013
10,986
if a output can load a max of 5A, if i just do the life test with a load of , say 0.5 A , does it affect the test result?
YES. Whether it is a switch, a power source, an amplifier - whatever. Testing something at only 10% of its rated capacity is not a "life test" of any kind.

ak
 

ronv

Joined Nov 12, 2008
3,770
We really can't answer your question without knowing more about your circuit and your test. Lets say for example you have a bunch of logic and then some power electronics that run 5 amps. If your design is speced for 40C operating temperature and you run it at a temperature of 75C certainly the logic will be stressed. If your 5 amp part only runs at 50C with 40C ambient it will also be stressed at 75C and .5 amps. The reverse is also true if the test takes the parts well outside their maximum temperature.
 

mcgyvr

Joined Oct 15, 2009
5,394
Its really all about "heat dissipation".. I highly doubt something only running at 10% is going to dissipate the same amount of heat as if it was running 100%.
But as ron has stated we really don't know what your device does.. Sounds like you don't though either.. I assume this is a "professional" product and as such there is probably an "Engineer" you should ask at your company to find out how it should be tested.
 

ErnieHorning

Joined Apr 17, 2014
65
You should at least be testing each output at the maximum current and operating temperature that you’re advertising that your product will work at. That doesn’t mean that every output is on at the same time unless you say that it can be. There are a lot of products that state what the total current can be (i.e. 5 amps per output, 15 amps total). We’ve got a few baseboard heaters too. They’re the cheapest 1500 watt resistors @ 120 volts you can buy.

Typically my products have a microcontroller on them and we load special code that cycles one or more outputs at a time. We just connect a counter to each output and make sure they all have the same counts at the end of the test.

If there is no micro, I manipulate the product with external hardware (i.e. PC, embedded controller, etc.).

Typically we're cycling between -30 & 140°F for a week. We also do life at room temperature too but then we're cycling the outputs 500 thousand to a million times.
 

mcgyvr

Joined Oct 15, 2009
5,394
As I stated before your test needs to simulate what a real world "fully loaded" usage will be or the test is just pointless.
If the end user could have all 4 outputs running at 5A constantly (and your instructions/specs state thats acceptable) then you need to test to that.
As 1 output will/could contribute heating to the others.

You don't need to match ambients as you can mathmatically correct the results..
ie.. if max stated ambient is 50deg C and test environment is 25 deg C you simply add 25 deg C to each test points measurement.
 

Thread Starter

jismagic

Joined Sep 7, 2013
77
okay!

the product is a automotive electronic device.
Btw what is a baseboard heater? ("They’re the cheapest 1500 watt resistors @ 120 volts you can buy.")
 
Top