Ive been measuring the temperaturs of the heatsinks on the amplifiers ive neen working on over the last few months, and they are all very different, but the one im working on at the moment has a very high temperature (53c)
most i have worked on get up to at most the mid 40's
can this affect transistor performance and eventual quality of output?
so a bit like a PC processor, the hotter it gets the less efficient, without the heatsink and fan it would soon fail, as would the transistors
with the processor it slows as it gets hotter, or too hot and i wondered if the same could happen to BJT'S?
most i have worked on get up to at most the mid 40's
can this affect transistor performance and eventual quality of output?
so a bit like a PC processor, the hotter it gets the less efficient, without the heatsink and fan it would soon fail, as would the transistors
with the processor it slows as it gets hotter, or too hot and i wondered if the same could happen to BJT'S?
A silicon transistor will work up to 150 degree's C.
However not very well at that temperature.
I wouldnt call 53 degrees high if the amplifier is at full power.
However not very well at that temperature.
I wouldnt call 53 degrees high if the amplifier is at full power.
ok understodd, but would they still work more efficiently if you kept the temp low, say 10c regardless of output
The heatisnk temp. tells you little about the actual junction temperature as the thermal resistances between heatsink and junction are unknown for the moment.
A 53°C heatisnk is pretty hot IMHO but as always it depends.
Is this a class-A or AB-amp? What´s the power dissipation at 53°C?
If you run an amplifier at elevated temperatures you have to watch out for thermal runaway and if they are still operating within their safe operating area.
A 53°C heatisnk is pretty hot IMHO but as always it depends.
Is this a class-A or AB-amp? What´s the power dissipation at 53°C?
If you run an amplifier at elevated temperatures you have to watch out for thermal runaway and if they are still operating within their safe operating area.
I think 53 is fine. Most devices are rated to 70c, the standard commercial temperature limit. I can't find a reference anymore, but I seem to recall the rule of thumb was "For every 10c increase in operating temperature, lifetime is cut in half". That is how accelerated testing is done. So it is more about reliability I think than performance.
You clearly can and should not advise something like that when you don´t know the application.
What do you mean by that? A case temperature of 70°C? Max. junction temperatures are more like 125-175°C.
This is only true within certain temperature boundaries.
Accelerated tests or "high temperature operating life"- or htol-tests are executed at different temperatures.
With help of the Arrhenius equation you can then calculate the lifetime of a semiconductor at a given temperature.
Yes, sort of but the performance could be very short though ;-)
The safe operating area of a semiconductor has to be derated linearly dependent on temperature.
If you elevate the temp. enough you could easily be driving it outside of its SOA.
53°C might be fine but it really depends.
Class A amplifiers run at high temperature and dissipation all the time and they work fine. If you dont burn yourself on them they're OK 🙂
Could it be you are mixing rules for electrolytic capacitors here? Transistors don't really have a stated lifetime.
You really need to work back from heat sink temp to case temp to die temp.
Values will be in transistor data sheet.
Values will be in transistor data sheet.
We routinely have 45 degrees ambient here, so getting 53 on a heat sink is low...
Also, where on the heat sink was this measured?
How, in the sense what kind of sensor? Within its sweet spot of range? Best to measure the device, or as close as you can get on the other side of the device on the back of the heat sink, that means heat sink mounting position, from where the heat will be transferred to the rest of the heat sink.
Basically, is the temperature stable or increasing?
If the heat input from the device is equal to the heat sink dissipation, it is okay.
The device must be properly mounted so that no hot spots or air gaps exist, with proper fitting practices.
You can check up on the safe storage and operating temperatures in the manufacturer's data sheet, and try to stay lower than that.
As a thumb rule, stay within 75-80% of the max. continuous ratings for power, voltage and temperature to get the best long lived performance.
Also, where on the heat sink was this measured?
How, in the sense what kind of sensor? Within its sweet spot of range? Best to measure the device, or as close as you can get on the other side of the device on the back of the heat sink, that means heat sink mounting position, from where the heat will be transferred to the rest of the heat sink.
Basically, is the temperature stable or increasing?
If the heat input from the device is equal to the heat sink dissipation, it is okay.
The device must be properly mounted so that no hot spots or air gaps exist, with proper fitting practices.
You can check up on the safe storage and operating temperatures in the manufacturer's data sheet, and try to stay lower than that.
As a thumb rule, stay within 75-80% of the max. continuous ratings for power, voltage and temperature to get the best long lived performance.
Last edited:
Does a 10degC Increase in Temperature Really Reduce the Life of Electronics by Half? | Electronics Cooling
Their lifetime isn´t stated most probably due to cost reasons.
Also the structures in a typical TO-247/TO-220 transistor are very big and don´t vary so much. Manufacturers should have a pretty big ability to foresight/estimate their lifetime by now.
You can find some reliability data though (look for values MTBF/MTTF or similar):
https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwjNmIf-ycvxAhXI_qQKHepGDr8QFjAAegQIBRAD&url=http%3A%2F%2Fwww.hp.woodshot.com%2Fhprfhelp%2F4_downld%2Fproducts%2Fxrs%2Fat41511_r.pdf&usg=AOvVaw0SMH7UvSG4XMHgDQedJDOq
All thermally activated failure mechanisms cause a halving of the MTTF with a 10 degree C increase in temperature. 10 is usually assumed for simplicity, even though it may bit be *exactly* 10. It is always close.
Transistors do have lifetimes - but you have to look up manufacturer and PROCESS specific curves for MTTF. Some processes will give lifetimes of as much as a million hours at 150 or 200C. But don’t expect that from all silicon processes. And ON Semi epi-base will be different from Toshiba triple diffused, and different still from Sanken LAPT. In addition, rated maximum often is not dictated by transistor die failure or degradation - but by other things in the assembly process. Die attach methods are often the limit - that’s what limits the TO264 and its brethren to 150C even though the same die will go higher in a TO-3. The cheap molding compounds used in commercial IC packages limit it to a miserable 70C (where the more expensive military version can go to 125). Again, these other failure mechanisms still induce failures that follow the same rate curve - a doubling for every 10C. It is just the nature of heat.
Transistors do have lifetimes - but you have to look up manufacturer and PROCESS specific curves for MTTF. Some processes will give lifetimes of as much as a million hours at 150 or 200C. But don’t expect that from all silicon processes. And ON Semi epi-base will be different from Toshiba triple diffused, and different still from Sanken LAPT. In addition, rated maximum often is not dictated by transistor die failure or degradation - but by other things in the assembly process. Die attach methods are often the limit - that’s what limits the TO264 and its brethren to 150C even though the same die will go higher in a TO-3. The cheap molding compounds used in commercial IC packages limit it to a miserable 70C (where the more expensive military version can go to 125). Again, these other failure mechanisms still induce failures that follow the same rate curve - a doubling for every 10C. It is just the nature of heat.
Maybe not so needless to add:
If using transistors out of their SOA these reliablity data do not apply.
Failures can happen in an instant or at least severely accelerated.
If using transistors out of their SOA these reliablity data do not apply.
Failures can happen in an instant or at least severely accelerated.
Secondary breakdown is primarily voltage-activated. It’s dependence on voltage is much greater than it’s dependence on temperature. But that particular failure mechanism also has a similar temperature dependence - doubling every 10C. You see the curves push out a bit for pulsed operation, because of the decrease in temperature. But you also see the voltage breakpoint move because it’s following a different curve. This curve depends on everything - process, die layout, size - anything that can cause temperature non-uniformity - you name it. It is specific to a transistor *type*, and somewhat dependent on who made it. No two lines even following the same recipe give exactly the same result.