sampleaccurate wrote:
Your right, I copied it wrong, the article says 40 watts into 16 ohms, not 4 ohms. Now I just need to wrap my solid state brain around the counter-intuitive idea of higher power into higher impedance loads instead of the other way around. It makes sense to me only for an amp with a very high output impedance, but this one is supposed to be very low as you mention.
How did you measure the output impedance?
The reason is that current is the limiting factor. These tubes can be pushed hard; with the anode current able to peak at something of order 1.75A maybe (way above the 0.6A continuous average indicated in the data sheet). But, that is about the limit. So, with the peak current constrained like that then power, being the product of resistance times current-squared, gets bigger roughly in proportion to the speaker impedance. (At least over the range of impedances around 8 ohms or so.)
The impressively low output impedance of the OTL is a consequence of the negative feedback. I think I saw an article saying it would be about 11 ohms for the "modified Futterman" type circuit using 6C33Cs, if there was no feedback. With feedback, this is reduced to about 0.3 ohms or so.
The output impedance can be defined by assuming the output of the amplifier can be modelled as a perfect voltage source (zero impedance) with a series resistor Rint which represents the output impedance of the amplifier. To measure Rint, we first set the amplifier to drive a resistive load (say 8 ohms), and measure the voltage across it for a fixed input signal. Now, keeping the input fixed, add an additional resistor in parallel with the 8 ohm resistor, and measure the output voltage again. It will now have dropped, of course. A simple calculation from these measurements allows us to determine Rint.
I did my measurements by first using an 8 ohm load, and then paralleling another 8 ohms across that for the second measurement. So if we say that the perfect voltage source is producing V0 volts, and that we measure V1 across the 8 ohm load, and V2 across the 4 ohm load of the two paralleled 8 ohm resistors, then
V1 = 8*V0/(8 + Rint) and V2 = 4*V0/(4 + Rint),
so having determined the ratio V2/V1, we can solve for Rint.
Of course, the idealisation described above is only valid as long as the current is never required to exceed the maximum the tubes can pass. So the impedance is low as long as the amplifier is still within its current capabilities. I made my impedance measurements with the input signal low enough that the output currents were way below the limits.
But if, for example, I had made the first measurement using a level that would imply 25W power dissipation into 8 ohms, then the output voltage would have collapsed down dramatically upon paralleling an additional 8 ohms across the load, and the resulting measured "output impedance" would have been way higher.
I hope this helps!
Chris