Fender Trans-Impedance Power Attenuator
With the Summer 2006 announcement of the Princeton Recording amp, Fender has announced its intention to ship a tube amplifier with an interesting new bit of technology: The Trans-Impedance Power Attenuator (U.S. Patent #6,816,009). In this article I'll discuss, in a mostly non-technical manner, what makes Fender's attenuator so special. If you want to delve into the technical details for yourself, consult the patent.
A "normal" attenuator (Power Brake, Hot Plate, etc.) puts a power resistor across the amp output to make sure that it works into a suitable impedance, and another resistor in series with the speaker to reduce the power reaching the speaker. The problem with that approach is that a speaker presents a "complex" (i.e. not purely resistive) impedance to the amplifier, but the load on the amp becomes more purely resistive as you increase the attenuation. This loss of direct coupling from the output transformer to the speaker alters the behavior of the amplifier.
The Weber Mass is one attempt to circumvent the problem of decoupling the complex speaker load from the amplifier. The Mass is a coneless speaker that's used instead of a load resistor across the amp's output. The coneless speaker presents a complex impedance that's very similar to a normal speaker. The rest of the Mass is just like any other attenuator: a resistor to reduce the amount of power reaching the speaker. The Mass has a couple problems: the minimum attenuation is 3 dB (the coneless speaker and the normal speaker each get half of the amp's power at minimum attenuation), and the impedance reflected back to the amp still varies depending upon the amount of attenuation.
London Power's Power Scaling reduces the power delivered to the speaker by reducing operating voltages within the tube power amplifier. This is effective over a wide range of power outputs: delivered power is roughly proportional to the square of the operating voltage. Also, since the output transformer remains connected directly to the speaker, the complex impedance of the speaker is always reflected back to the output tubes. However, the magnetic flux in the output transformer decreases along with the power output, and some of the "character" of the output transformer is lost at low power. The power supply also sees less demand for current at low power outputs, which also changes the behavior of the amp.
Maven Peal has its own technology that appears to be similar to power scaling, with the addition of a way to simulate a heavily-loaded power supply at lower power outputs.
Fender has, without going into the technical details1, come up with a way to reduce the power to the speaker while both reflecting the complex load of the speaker as-if it was still directly connected to the amp and also continuing to exercise the power supply, output tubes and output transformer at unreduced power levels. I would expect this approach to preserve most of the the relevant behaviors of a cranked tube amplifier while allowing the speaker to be run at any desired volume.
This analysis is all based upon my own reading of Fender's patent. I haven't actually seen, much less heard, a Princeton Recording amp. On paper, though, Fender's technology looks as if it would address most of the shortcomings (save the decrease in speaker distortion at lower power levels) found in other approaches to power attenuation. I'm looking forward to finding out whether their first implementation of this new technology delivers the goods.
Note 1: Briefly, Fender's attenuator is a variable-gain SS power amp driving the speaker, plus a current-feeback loop. The speaker current is sensed and fed back to the output of the tube amp. The current feedback is scaled in such a way that the tube amp appears to be pushing the complex impedance of the speaker no matter how much you attenuate the actual signal pushing the speaker. Undoubtedly there are some second-order effects lost due to increased speaker linearity at lower power input... Even power-scaling hasn't solved that problem. OTOH, any passive attenuator attenuates the back-EMF caused by the complex impedance, so the Fender approach is a huge step in the right direction, IMO.