For a properly designed amp (say something you'd buy from Pass Labs today) used correctly (provide the ventilation specified by the vendor) - would you expect a Class A amp to have a shorter life than a more conventional amp that runs cool at idle?
The key assumption here is 'convectional' amps are on and at idle, or close to it for the majority of their (powered on) life - where a Class A amp is running hot by design (and better enjoyed hot vs. flipping on and listening immediately, so an incentive to turn it on, leave it on, let it get hot).
Even if you didn't burn out the 'important' parts (the actual power transistors making the heat), would you expect to see things like the caps and other parts burn out sooner?
Yes, very general question, just curious what your thoughts and experience are.
The key assumption here is 'convectional' amps are on and at idle, or close to it for the majority of their (powered on) life - where a Class A amp is running hot by design (and better enjoyed hot vs. flipping on and listening immediately, so an incentive to turn it on, leave it on, let it get hot).
Even if you didn't burn out the 'important' parts (the actual power transistors making the heat), would you expect to see things like the caps and other parts burn out sooner?
Yes, very general question, just curious what your thoughts and experience are.
Electrolytic capacitors would be the least reliable part in a high temperature environment.
Semis can be reliable if their ratings are observed.
Semis can be reliable if their ratings are observed.
Last edited:
Ran my 2nd hand Bedini 24/7 for at least ten years. The electrolytics in the feedback loop were replaced twice in those ten years, but then the Bedini ran bloody hot - the sinks are without a doubt too small. The case ran at a good 60 C. The PSU reservoir caps were replaced quite early on, but these were already quite old. All the caps went soft. Everything else was fine, though the PCB under the current drive stage got burnt and after 25 years the tracks started lifting in this area too. The heat also caused the insulation on all the transformer leads to start cracking.
I'm still busy rebuilding the amp, but with 'sinks that are 80% bigger, and new PCBs made with 2oz. tracks. Should help a lot.
I'm still busy rebuilding the amp, but with 'sinks that are 80% bigger, and new PCBs made with 2oz. tracks. Should help a lot.
Last edited:
Yikes! Do you think the amp should have been designed better from a thermal point of view (even if the sounded great)...like you would likely not see that happen on a something like a modern Pass Labs amp?
Exactly. Some designers clearly spend more time thinking about heat distribution than others. The quality of capacitors also makes a big difference.
Anecdotal evidence: two twenty year old amps, one from Krell, running class A and a second from Theta in AB. Not a single cap in the Krell exhibited any meaningful change in value or ESR. About 2/3 of the small caps in the Theta were dead, partly due to heatsink proximity and partly because of being really cheap.
It really depends on whether the designer actually understood thermal design. I've seen it go both ways. Being class A doesn't really mean much in terms of lifespan, since there's plenty of class AB or even IC-based amplifiers that run ridiculously hot.
As an example, Neve V and VR series consoles require almost a full-time tech to keep them running, mostly replacing caps and switches, both of which die from the heat. Some studios have been known to completely recap VR72s ever five or ten years, and they wouldn't do it if it wasn't necessary- recapping a 72 channel console is a massive job. The problem? There are an insane number of 5532s without a good way to get rid of all the heat, so everything just sits and cooks.
On the other hand, I've got plenty of amplifiers from the mid-90s which kick out a lot of heat, but are still running original caps, even after several hundred hours of running.
As an example, Neve V and VR series consoles require almost a full-time tech to keep them running, mostly replacing caps and switches, both of which die from the heat. Some studios have been known to completely recap VR72s ever five or ten years, and they wouldn't do it if it wasn't necessary- recapping a 72 channel console is a massive job. The problem? There are an insane number of 5532s without a good way to get rid of all the heat, so everything just sits and cooks.
On the other hand, I've got plenty of amplifiers from the mid-90s which kick out a lot of heat, but are still running original caps, even after several hundred hours of running.
The lifespan of any semiconductor circuitry tends to drop exponentially with operating temperature so yes class A are particularly at risk - however the cycling of temperature is often the most important factor - so a class B left permanently on at 100C could be less vulnerable to semiconductor failure than a class A running at 90C intermittently.
Capacitor dehydration however will be mainly about the total exposure time to high temps.
Temperature cycling effects can lead to mechanical stress issues too which over time can cause failures.
Capacitor dehydration however will be mainly about the total exposure time to high temps.
Temperature cycling effects can lead to mechanical stress issues too which over time can cause failures.
I have a Musical Fidelity A220. It's essentially an egg cooker - class A then A/B and it's normally on the too hot to put my hand on.
Constant heat/cool cycles aren't good for PCB solder joints/tracks, I suspect it's also not particularly good for the semiconductor devices either but above all the caps are essentially in an oven so even if they're 85degC or 105degC parts it does reduce their lifetimes and that depends on the parts used.
So basically what Mark said 🙂
Constant heat/cool cycles aren't good for PCB solder joints/tracks, I suspect it's also not particularly good for the semiconductor devices either but above all the caps are essentially in an oven so even if they're 85degC or 105degC parts it does reduce their lifetimes and that depends on the parts used.
So basically what Mark said 🙂