Tube Amp for Martin Logan Speakers

Bifwynne, if I may, not trying to split any hairs, but the term SOA has a very specific meaning for ss devices. The Safe Operating Area defines the combinations of current and voltage across the device, that it can withstand for a certain time before being destroyed. For instance, a power transistor might be rated for 100W dissipation, but that's not the full story. 10 amps at 10V, being 100W, would be OK indefinitely. BUT, 1 amp at 100V would NOT be OK indefinitely, although it is also 100W. That 1A/100V combination can only be withstood for, say, 100 milliseconds, and then it will let out all the magic smoke*.
Now over to amplifiers: they also have an area where they will operate as designed. As you noted, drive the level too high, they will clip. Drive too much level in too low a load and they may eventually overheat and break. One common accident on the test bench is full power at 20kHz - not many amps can hold out on that, and of course during music reproduction, they don't have to. But all this is not referred to as 'the SOA' of the amp.

* The reason is that at higher voltages, the current tends to hog a small part of the transistor die, so that small spot will heat up much more that the rest of the die. That small spot cannot handle the 100W the whole transistor could handle. It blows.

I would like to thank Jan Didden for laying this furfie
"NF feedback DOES NOT effect Z=out or damping factor" out in the open, in his usual calm and palpable way.
His is a bottomless pit of correct information over at the much more techincal forums, along with the usual others like Nelson Pass, John Curl, and many others, there should be more of them to police bad info on forums.

http://www.wordsense.eu/furfie/

Cheers George

George don't make me blush :-).
Anyway. Atmaspere's statement that the INTERNAL Zout doesn't change has another side to it. It suggests that you can open up an amp and point to a component and say see, there's the output impedance. This is not the case. Apart from a few minor things like the resistance of the cabling from PC board to speaker connector, the Zout is sort of virtual. One cause for instance is the fact that the gain of the output transistors drops when you request more output current, so the output voltage drops a bit and that is seen as 'Zout'. As Bitwynne noted, what nfb does is counteract the drop in output voltage so it appears that Zout is now less. But the 'internal Zout' and the Zout with nfb are equally virtual.

Jandidden, keep in mind this thread is not about transistor amps. But more importantly you have to be careful here not not violate something called Kirchoff's Law.

Actually you will find that such is impossible as it is a law of physics, and not part of legal code :)

This law is known also as the law of energy conservation and simply states that the amount of energy in an electrical network is equal to the amount of energy going into that network.

What feedback does not affect output impedance:

Now if we take two output circuits, one with a high impedance, for example the single ended output of a tube preamp, and that of a transistor power amplifier we will see that due to the lower output impedance of the amplifier that it will drive a lower impedance.

Now if what you and George say is true, that adding negative feedback lowers output impedance, it then follows that if we add feedback to the preamp circuit its output impedance will become so low that at some point we can drive a 4 ohm load effortlessly.

But what we find is that is not the case. Now you may argue that the preamp output is not relevant, so let's take the case of an SET with a 10 ohm output impedance on the 8 ohm tap. If what you say is correct, its output power will increase if feedback is added when asked to drive a 4 ohm load off of that tap. But again, what we find is that the 4 ohm output power is unchanged.

The reason for this is if negative feedback really did decrease the output impedance, the resulting circuit would have the increased current to drive a lower impedance. That of course would violate Kirchoff's Law.

Of course, the real way to provide for greater current ability is more output devices, larger heatsinks, output transformers, power transformers and the like.

IOW, what is happening is that the term 'output impedance' as used with the Voltage Paradigm is a charged term that actually refers to servo gain in the output circuit and not the actual impedance of the devices involved (all types of which have an impedance greater than zero ohms).

One might state that the issue here is semantic- I point it out simply because its use in the context of teh Voltage Paradigm leads to a lot of confusion- but that is how the audio industry is set up.

If you are having trouble following this, it is because you are operating within the Voltage Paradigm. The word 'paradigm' has to do with a platform of thought; quite often viewpoints outside of that platform are hard to think about or might be considered blasphemous.

A further note- people have accused me of making up the two Paradigms (voltage and power) that I mention in the article at this link:

http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php

In case I run into that I refer them to this google search on the Fisher A-80 amplifier

https://www.google.com/search?client=ubuntu&channel=fs&q=fisher+a-80+amplifier&ie=utf-8&oe=utf-8

The very first hit you get at this link is a YouTube image of the damping control of the Fisher amplifier. It is marked 'Constant Voltage' at fully counter clockwise, 'Constant Power' at noon and 'Constant Current' at fully clockwise.

This is because there was a time when not all loudspeakers conformed to the voltage model, despite George's and Bruce Rosenblit's remonstrations. Such speakers are still made today, and in increasing numbers. Any speaker than can be driven successfully by an SET will be an example.

ESLs are also an example as it will be found that their impedance curve is not based on the resonant impedance of a driver in a box. If you could show that the impedance curve of the speaker (which in most ESLs varies by about 10:1) is also an efficiency vs frequency curve then you would have an argument that the speaker is a Voltage Paradigm device. IOW the ML ESLs are a low impedance Power Paradigm loudspeaker.

Atmasphere, your conclusion that someone (certainly not me!) would think if you lower the Zout to zero in a preamp, it can effortlessly drive 4 ohms, is rather far-fetched. As I'm sure you know, Zout has NOTHING to do with what you can or cannot drive, and I would expect that most people here would be well aware of that. A D-type battery cell has a Zout of less than an ohm but nobody would expect it to be able to start their car. :-) . I really have no idea what gave you that particular idea.

BTW I gave the example of ss amps' Zout being caused by for instance the transistor gain droop with increased load current. In a tube amp, there are causes for 'Zout' like the internal output tube(s) resistance and transformer losses that cause a drop in output voltage if you increase load current (for instance by connecting a lower impedance speaker). My point was that this Zout in the amp is largely 'virtual' in the sense that you cannot point to a resistor in the amp that would be the Zout. Now what nfb does is making the drop in output voltage with more load less so, thus causing the Zout to appear lower.

Actually it is quite easy to measure Zout on your amp. Just connect an 8 ohms resistor as load and set Vout to say 8 volts. The Iload is therefore 0.5 amps. Now connect a 4 ohms resistor and let's say the Vout drops to 7 volts. The Iload is now of course 7 V/4 ohms = 1.75 amps. So we now know that Vout drops 1 V if the current increases 0.75 amps. Therefore Zout is 1 V / 0.75 amps = 1.3334 ohms. Easy, no?