is it dangerous to run a tube power amp unloaded?

[ccmccull]

I saw this article today and was struck by the gray box titled "Speaker Loading & Valve Amplifiers"

https://www.soundonsound.com/techniques/understanding-impedance

"Whereas most modern solid-state amplifiers are virtually bombproof in terms of whether their outputs see proper loudspeakers (of any nominal impedance) or a short or open circuit, most valve amplifiers are far less tolerant. In fact, the majority of vintage valve amps will self-destruct if driven without the correct speaker load attached! The reasons are complex and depend to some extent on the design of the output circuit, but can be boiled down to what are called 'reflected' impedances..."

I've never seen this kind of warning in the bottlehead manuals, so I assume our output circuits, e.g. in the S.E.X. I recently built, are designed to mitigate this problem. I'm curious how one designs such an output circuit.

The reason for my curiosity is that I was building a reproduction Fender 5E3 based guitar amp

https://www.mojotone.com/site/TT_SB2_Files/img/home/Tweed%20Deluxe%20Manual%20.pdf

Schematic here: https://www.mojotone.com/Amp_Kits/Tweed/Deluxe_5E3_SCH.pdf

and the build manual does warn about running it without a speaker plugged in. So I was wondering what the design differences between output stages are that would exacerbate or mitigate this unloaded operation issue.

Thanks for any insight or pointers!

[Doc B.]

That's often true of typical old school tube amps with lots of global negative feedback. But our amps with little or no global negative feedback don't go into oscillation if left unloaded. It's not a problem to run them without speakers, except for the fact that you won't hear much music.

[Paul Birkeland]

For a triode output stage, the maximum amplification factor in the output stage has an upper bound that is just the mu of the triode.  This also holds true for triode strapped pentodes.  The amplification factor of a pentode varies greatly depending on the load impedance, so it can get a bit out of control with an unloaded output transformer.  This is further exacerbated by circuits with little to no global feedback. 

[ccmccull]

Thanks Paul and Doc! You'll be happy to know that I didn't blow the 5E3, even though I did mistakenly run it unloaded for a few seconds...