I just cleaned up and reserviced an old hitachi amp and while I was at it I put a 90mm muffin fan atop the heatsinks to boost the updraft. 12vdc fan running on 4.5vdc. Just enough cfm to work and really negligible noise. These are all tradeoffs, and for the purposes of the super high fidelity that most home equipment aims for, even the barest whisper of added noise is anathema to many. So this question is not for the purists.
With this little mod, the output transistors on the sink are much much much cooler to the touch (an unscientific but significant improvement).
My question is... to what extent is heat in an amp analogous to heat in an internal combustion engine? The engine is designed to perform optimally in a pretty specific range as determined in its design. Too hot is not good. Nor, less obviosly, is too cool. Is this at all true for electronics?
I would expect far better lifespan out of any amplifier given a little fan assisance regardless of how well it was ventilated or how large the heatsink to begin with. But do they actually perform better when 'warmed up"?
With this little mod, the output transistors on the sink are much much much cooler to the touch (an unscientific but significant improvement).
My question is... to what extent is heat in an amp analogous to heat in an internal combustion engine? The engine is designed to perform optimally in a pretty specific range as determined in its design. Too hot is not good. Nor, less obviosly, is too cool. Is this at all true for electronics?
I would expect far better lifespan out of any amplifier given a little fan assisance regardless of how well it was ventilated or how large the heatsink to begin with. But do they actually perform better when 'warmed up"?
It really depends on the bias circuit, and older often implies simpler and not so accurate over a wider temp range, though not necessarily. A cooler output stage needs more bias drive for the same bias setpoint. Of course forcing air through the entire amplifier might mean that every transistor is running cooler, so you can shift the operating point through every stage of the amp. Over the years I've stuffed fans around almost every amp I've had, depending on how hard I was abusing it at the time. When you're really running an amp hard pretty often you're more worried about it blowing up than becoming underbiased, and the heat generated is still enough to keep things in the normal range even with forced air. When you're barely running the amp over idle you may as well leave the fans off.
You can't compare an engine to a power amplifier 
There is no such thing as "too cool" with a power amp. The cooler it runs, the longer it'll last. The big problem is with the output transistors that produce the most heat. What you have is a silicon die welded to a copper substate & these don't happen to expand at the same rate when things get hot & sticky or contract at the same rate when switched off.
Given so many cycles of being turned on & off & used, ran at high volume etc if it's a class B amp, the silicon die eventually starts breaking away from the copper base & this leads to the eventual demise of the transistor. If you can keep it cool then it'll last longer.
In a class A amp it'll last longer if the thing is left switched on permanently as there will be no thermal cycling, eventually though all transistors will die
There is no such thing as "too cool" with a power amp. The cooler it runs, the longer it'll last. The big problem is with the output transistors that produce the most heat. What you have is a silicon die welded to a copper substate & these don't happen to expand at the same rate when things get hot & sticky or contract at the same rate when switched off.
Given so many cycles of being turned on & off & used, ran at high volume etc if it's a class B amp, the silicon die eventually starts breaking away from the copper base & this leads to the eventual demise of the transistor. If you can keep it cool then it'll last longer.
In a class A amp it'll last longer if the thing is left switched on permanently as there will be no thermal cycling, eventually though all transistors will die
Yes, thermal cycling is important to consider. This is why if you can rig it a thermistor on the sink to control fan speed is really boss. For lower cost class AB amps often the sink is too small to prevent big case temperature swings during a single song, no matter what the fan does. Also, in older amps, the relative dryness of the thermal grease is a big issue. But by far the biggest problem is the slope of max power dissipation versus die temperature. If you want maximum dissipation capacity you have to keep the temperature down.
In my high power amp in a pc case I use 3 fans.
One sucks air in and another blows it out the opposite corner.
I also have fan blowing air through the internal heatsink.
I found using a slow fan was useless, it just didnt pass enough air through the amplifier. So I use 3 large pc fans running on 12 volts dc.
One sucks air in and another blows it out the opposite corner.
I also have fan blowing air through the internal heatsink.
I found using a slow fan was useless, it just didnt pass enough air through the amplifier. So I use 3 large pc fans running on 12 volts dc.
Given the input so far, I would say.... EXACTLY!
There is no such thing as too cool, especially in Austin, Texas, except that most Texas Republicans, and that's most Texans outside of Austin, would say the Austenites are too-cool for the common school! But that's another matter...
So if cool is all good in electronics, unlike our internal combustion engine, and if my fan is significantly reducing the temperature of all elements, especially the output transistors (i wish i had some real temps to provide - maybe i'll measure them) and if (all these ifs) the fan noise is actually (and this is usually the case) below the ambient threshold produced by the road traffic, neighborhood kids, air traffic overhead, furnace and airconditioning, mice in wall joists, etc... then what's not to love about an aux fan in your integrated amp?
Thermistor control would be nice, and electronically basic enough to be within the means of most of us at diyaudio to employ, but not really necessary unless you were listening to very dynamic music, at very low volumes, at 3am deep in the countryside (or in your very own anechoic test chamber).
There is no such thing as too cool, especially in Austin, Texas, except that most Texas Republicans, and that's most Texans outside of Austin, would say the Austenites are too-cool for the common school! But that's another matter...
So if cool is all good in electronics, unlike our internal combustion engine, and if my fan is significantly reducing the temperature of all elements, especially the output transistors (i wish i had some real temps to provide - maybe i'll measure them) and if (all these ifs) the fan noise is actually (and this is usually the case) below the ambient threshold produced by the road traffic, neighborhood kids, air traffic overhead, furnace and airconditioning, mice in wall joists, etc... then what's not to love about an aux fan in your integrated amp?
Thermistor control would be nice, and electronically basic enough to be within the means of most of us at diyaudio to employ, but not really necessary unless you were listening to very dynamic music, at very low volumes, at 3am deep in the countryside (or in your very own anechoic test chamber).
Last edited:
The Arrhennius equation applies to MTBF for output devices. I understand that MTBF approximately halves for each 11C rise in temperature, due to doping migration effects which 'age' and eventually destroy the transistor.
Thereforce, you'd expect the output stage to last longer at lower temperature, and if the bias temperature management system is well designed there will be little difference in bias current regardless of temperature, within certain broad limits.
I found that moderate air circulation yielded around three times the still air cooling capacity in a small, multifinned heatsink. I could easily push 175W into a 180 x 180 x 50 heatsink with just 28C rise above ambient. Without forced circulation, dissipation dropped three times for the same rise. An amazing difference. Try idling an auto in heavy traffic with a defective visco coupler on the fan; they overheat very quickly.
Thereforce, you'd expect the output stage to last longer at lower temperature, and if the bias temperature management system is well designed there will be little difference in bias current regardless of temperature, within certain broad limits.
I found that moderate air circulation yielded around three times the still air cooling capacity in a small, multifinned heatsink. I could easily push 175W into a 180 x 180 x 50 heatsink with just 28C rise above ambient. Without forced circulation, dissipation dropped three times for the same rise. An amazing difference. Try idling an auto in heavy traffic with a defective visco coupler on the fan; they overheat very quickly.
Last edited:
Member
Joined 2009
Paid Member
All the above gets my vote too - it's also my understanding that temperature is the big thing, the lower the better is almost always a good thing to lengthen the average lifetime of components. But they will all die eventually - there are so many ways in which they fail.
In DIY a failed amp is a great opportunity to build a new one
In DIY a failed amp is a great opportunity to build a new one
In DIY a failed amp is a great opportunity to build a new one
For me I love the challenge of fixing an amp.
I have seen some horrible faults in my time.
One was after replacing blown output MOSFETs the amp still misbehaved.
In the end I took out all the semiconductors and tested every component.
In the end a MPSA42 had lost all its gain and had an HFE of 1 !
11 degrees may be newer as die quality has improved. We used to use 10. Note this is die temperature, not case temperature. As the case is much larger, it may not reflect any where near the same rise as the actual die. If the case stays only warm to the touch, it should last a good long time.
Also consider that transistors are not linear with tempature. Bias and operating range varies with tempature, so keeping a die in the 40 to 90 degree C range means you will probably stay reasonably in the design range. Over 130, you really start to see degradation.
Also consider that transistors are not linear with tempature. Bias and operating range varies with tempature, so keeping a die in the 40 to 90 degree C range means you will probably stay reasonably in the design range. Over 130, you really start to see degradation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.