[AutumnCrown]

I have an HD800 S. For argument's sake, let's imagine that I want to drive that thing at ridiculously loud volumes. Say, maximum of high power mode of an IDSD micro BL:

https://ifi-audio.com/products/micro-idsd-bl/

That would mean that it can give off a continuous power output of

about 332 mw at 300 ohms, right?

And because the maximum continuous handling capability of the HD800 S is 500 mw, it is literally impossible to damage the HD800 S with this amp, right?

What if you put all of those 332 mw in a sine wave at 20 hz? Does the "power handling capability" consider the possible damage from over-excursion of the diaphragm?

 

[castleofargh]

Long-term input powermax. 500 mW as per EN 60-268-7

knowing what that measurement method really is would help. I couldn't find that paper or even the specific part about power measurement( all the google results I tried were under paywalls). when nothing is said I expect that the max power handling means that it survived a few seconds at that level. but here the "long term" information throws me off. is it as opposed to instantaneous max power, and defines like 2 seconds at 1kHz? or does it really mean long term as in minutes or hours under test? IDK.
I also don't know if potential damage would come from the physical force applied on the diaphragm, or if it's a matter of the coil heating up too much(which is not unlikely).

 

[AutumnCrown]

knowing what that measurement method really is would help. I couldn't find that paper or even the specific part about power measurement( all the google results I tried were under paywalls). when nothing is said I expect that the max power handling means that it survived a few seconds at that level. but here the "long term" information throws me off. is it as opposed to instantaneous max power, and defines like 2 seconds at 1kHz? or does it really mean long term as in minutes or hours under test? IDK.
I also don't know if potential damage would come from the physical force applied on the diaphragm, or if it's a matter of the coil heating up too much(which is not unlikely).

So from what I have read, it seems speakers usually blow out due to excess heat on the voice coil. Here is a very helpful analysis of this: http://sound.whsites.net/articles/speaker-failure.html

As you note, the power handling is an unspecified amount of time. The article talks about what is a typical test, as noted below. One thing I do not understand from this article, is that it implies that running a narrower band with the same power results in more heat:

"If such power tests were ever repeated with the drive signal from the amplifier being a pure sine wave at 250 Hz (or octave band, pink noise centred on 250 Hz ) [instead of the usual 50-500 hz pink noise band] the same speaker would be quickly be destroyed by the extra heat dissipation. It is very easy to measure these parameters, but the results are so unflattering almost no maker does so or publishes the results today."

Am I misinterpreting this, or do different frequencies, with the same power, result in different levels of heat? Is it because the speaker is much more efficient in wider bands? I thought that headphones were so inefficient that there was little more to lose in terms of extra heat from extra inefficiency.

 

[castleofargh]

I'm actually not sure. I know that I should avoid a bunch of things when doing routine measurements because the papers I learned from mentioned to be careful about them, but I don't necessarily know why.
a wild guess here would be that if the test signal has higher freq content, then it's not maintained too long at a given output like with a low freq, but also I'm guessing the driver is dissipating heat better thanks to the extra movements? IDK.
maybe @jagwap or other guys in the business can tell you what really goes on when they come by.

 

[jagwap]

As stated the test conditions are not clear. 500mW pink noise with a 6dB crest factor is likely to be less distructive than a sine wave at a low resonant frequency of a mechanical part of the driver. However the HD800 impedance plot is very flat for a dynamic driver, sugesting it does not have many extreme resonant issues low down.

I am more familiar with loudspeaker drivers. There the long term rms power is generally the 100 hour capability. The burst rms is usually double that roughly, and the peak power quadruple the long term power. But again this is usually 6dB CF pink noise. A sine wave at a low minimum impedance will do a lot more heating, and at resonance a lot more excursion.

Then there is the question of all of this being unclipped power. If you put unclipped 6dB CF pink noise out of your ifi, the rms power will be half that of an unclipped sine wave. That is because a sinewave has a crest factor if 3dB compared to the pink noise.

Now put heavily clipped signal of any type through the amp, and all bets are off. The rms power could be up to double.

I remember a talented colleague doing heat tests on a guitar amp, using pink noise. Then he stuck AC/DC through it at the same apparent level. It got hotter. His conclusion, with a huge grin: AC/DC are louder than physics... Sadly the real reason was the limiters used caused more disipation.

In answer: don't do it.

If you have and you broke your HD800s, talk nicely to their service department. I've had three drivers on my HD800 replaced free of charge by them and I have no idea what killed them (I know not clipping). They are wonderful in the Hong Kong service department, if you can find them open.

 

[AutumnCrown]

Now put heavily clipped signal of any type through the amp, and all bets are off. The rms power could be up to double.

If you have and you broke your HD800s, talk nicely to their service department. I've had three drivers on my HD800 replaced free of charge by them and I have no idea what killed them (I know not clipping). They are wonderful in the Hong Kong service department, if you can find them open.

I did put a heavily clipped signal into the headphones for about 1 second, but it was at least an order of magnitude lower than the maximum continuous input. It was a heavily clipped music signal with pot at 1 O'clock on an amp setting where max power at 3:30 is only 30 mw into 300 ohm. So I figure it was putting out less than 30 mw, even with the clipping. I think, though correct me if I am wrong, this gives me a comfortable cushion, so not surprisingly, the headphones are fine. Just got me curious about this stuff. I'm not going to do it again anyway.

I have to admit, I am a bit shocked that you have had to replace HD800 drivers 3 times, and I will admit that I have had to do the same once, which is surprising because they have a reputation for reliability. Are you by any chance using a certain amp which begins with the letter L, and has a reputation for killing drivers?