What do Oil Caps do and how do they effect sound?

Although Sean's post raised my eyebrows too, I think it's pretty academic to focus on how a device might theoretically continue to operate to some degree when it's damaged or broken, when obviously if it doesn't sound right we're going to want to repair or replace it ASAP.

Though I'm certainly no engineer, I think many of us realize that yes, vacuum tubes can be more robust than transistors under certain unusual conditions having little-to-nothing to do with home audio (some kinds of radiation exposure, thermal stress, higher voltages), and yes, they can be made to last a pretty long time under relatively harsh conditions where transistors are at a disadvantage (in radio transmitters, microwave magnetrons).

But when it comes to home audio, I think all we tube lovers can admit that our vacuum devices will be much more likely to fail, and anyway need a hell of a lot sooner replacement, than silicon semiconductors. In fact, tubes will audibly degrade over their useful lifetime to a much greater degree than transistors. And the ways they usually fail prematurely, such as losing their vacuum, blowing like an incandescent light bulb, or getting intolerably noisy in one way or another, are going to cause us to replace them, not marvel at how they might still be able to function a little bit. (But I would think the same goes for a failing transistor - I've just never had one go bad yet). In addition, tubes are generally more sensitive to some stuff that does find its way into the audio environment, like vibration and accidental impact.

Anyway, I've never thought "graceful" when one of my tubes died, except to the extent that it didn't take out anything around it, knock on wood (and once it did). But regardless, this whole debate is irrelevent to the question asked, and I'm not quite sure why Sean brought it up in the first place, unless he meant to imply that amps using large numbers of output tubes to develop higher powers would entail discouragingly high retubing costs...

Bombaywalla: I have pulled apart dozens of high current RF transistors and found blown junctions in them. All of these were still working albeit at reduced output due to the loss of internal conductors.

Many transistors use several junctions in parallel to handle the current levels needed, so blowing a few simply reduces the gain and max current potential. In effect, they have compacted several transistors into one case and they are all working together to share the load. Blowing one of the internal junctions inside of a transistor is no different than running a dozen external transistors in parallel. One, two or three might fail, but the rest of the circuit will continue to work at reduced capacity. That is, so long as the transistors that failed blew open and not shorted. If the transistors are running in matched push-pull pairs, noticeable distortion may occur, but the circuit can be made to work again by simply deleting the the mate to the transistors that went south. Once again, it will not be as powerful, but it will work and should maintain a reasonable semblance of linearity.

As far as tubes go, they are in a continual state of decline once they are fired up. While most experience a very slow and gradual death, it is more likely for a tube to "pop" at random than for a transistor just to up and let loose. On top of this, it is not uncommon for a piece of equipment to experience major damage when a tube lets go. If you doubt this, try searching the archives here pertaining to tube failure and start reading about the fires that have resulted.

Other than that, i agree with everything that Albert had to say. High rail voltages are a must if you want realism. Then again, you have to have enough current to maintain that level of voltage when under load or the rails will sag. As such, unless you have a speaker that is of noticeably higher than average impedance across the band, at least a reasonable amount of current will be required from the output section in order to maintain linearity, dynamic headroom and a solid soundstage. As the voltage sags under load, so does the impact and spatial characteristics. Sean
>

Both solid state and tube amps produce a voltage. Current is what results when voltage is applied to a load, such as a loudspeaker. An amplifier might be capable of high voltage output under no load condition, but if it has little current delivery capability this voltage cannot be maintained into a load. Your preamp typically can output a signal of several volts, enough to play a speaker quite loudly, however if you hook it up to a speaker I doubt you would hear anything. The preamp can only deliver enough current to maintain the voltage into a load of several tens of thousands of ohms, not 4 or 8 ohms.

Tube amps have high "rail" voltage (plus and minus 400 volts or more) so that the output tubes can apply a high voltage to the primary winding of the output transformer. The output tubes cannot deliver a great deal of current. However, the output transformer is wound with a step-down turns ratio, so that the secondary output voltage is down to a handfull of volts, as appropriate for the speaker, but the necessary current capability is there. You might think of the output transformer as similar to the gearbox of a sports car, which permits a small engine reved up to 7000rpm, to accelerate the car from a standing start.

Correction in order here bomby:

"845 tube. This is a radio broadcasting type tube & the grid is biased at 1500V! "

No, sorry - not correct. max PLATE voltage on 845 is 1250v, and ASL doesnt' run them that hard. They would not bias the grid higher than the plate(!) and infact the grid is going to have about 70-80v of negative voltage applied, to run them in class A1.

And let's stop talking about 'rail' voltage w.r.t. tube amps - it's called B+. Rails apply to sand amps.

Eldartford makes a good point I was going to mention - B+ on tube amp is somewhat irrelavent in any case, since it's almost always going to go out through an OT, which has a massive step down ratio. (5000 : 8 or more)

The only situation which B+ would be used to drive speakers directly would be a high voltage transmitting tube output connected directly to the panel of an electrostatic loudspeaker. There were a few commercial attempts at this I think, none sold today that I know of. It's an idea which has a lot of merit but unfortunately is rather difficult to implement well, or at all. There are some more worthwhile and successful attempts being made by the DIY tube crowd, but that doesnt' apply to commercially available designs. Most of the time you would need a bigger tube that even 845 to do it as a SET amp (the only way it would be worth doing), meaning either a tube like a 212 or 849 or maybe 250/450TL, or else a more modern planar triode ala Eimac or Svet ceramic types possibly. (talking 1500-3500+v on the plate of those tubes)

-Ed

Thanks for all the wonderful response. It was a pleasure to read them all. This will definitely help me understand my new Cary tube preamp and also help me when deciding on a new amp and weather to go SS or Tube when the time comes.