Attenuator puzzles

In Stereophile, November, 2000, page 105, reviewing the Hovland HP-100 preamplifier, Michael Fremer says: "The 31-position (2db steps) custom-made, low-mass, coin-silver-contact stereo volume control is configured as a series attenuator instead of the more conventional shunt configuration: the source sees a constant impedance and the listener is ‘...not forced to listen through high-value resistors' at certain levels." (My guess: the inner quote is from the manual.) This passage puzzles me in several ways. First on a fact: are shunt attenuators really "the more conventional" choice in active preamps? Can anyone name an active preamp that has one? Preliminary to further puzzles, some attenuator basics. For each channel, a shunt type has a fixed resistor in the signal path and multiple resistors selectable by rotary switch in the return path. (A return path potentiometer is another option.) A series type has a chain of resistors, arranged around a rotary switch, going to ground at one end, with input coming into the other end and output leaving from a switch-selectable point in the chain. In a series type using a K-position rotary switch, there are K-1 resistors in the signal and return paths combined, at any volume setting, with less in the signal path at higher settings. A ladder type uses two switch decks to switch-select a pair of resistors, one each for signal and return paths. So there are only two resistors in signal path and return path combined, for any volume setting. A shunt type is like a ladder in this regard. Back to the Stereophile passage. On the inner quote, why are high-value resistors inherently bad? And why are they put in the plural, since a shunt has only one each for signal and return paths? And whatever the badness is, how does a series type avoid it with its joined multiple resistors that add up to a "high-value resistor", in signal or return paths, "at certain levels", namely, high levels for the return path and low levels for the signal path? It is arguably bad for the signal to go through multiple solder joints, as it does at low levels in a series attenuator. This is the only badnessI can find regarding "certain levels". But despite the apparent intended thrust of the inner quote, this hardly shows the superiority of series over shunt! Another question: why didn't Hovland use a stepped attenuator? Why put serious money into a top quality switch and then use it suboptimally? The CAT preamps use a series attenuator, my old ARC SP-15 did, older Sonic Frontiers top preamps did too. Series is surely "the more conventional" choice. It takes only about half as many resistors as ladder, and half the switch decks, but their costs needn't be prohibitive. Why pass up a chance to make a product better, in one major respect, and steal a march on some very prestigious competitors? But a final question: with all the incoherence I've cited, and my view of shunt as an implausible rival to other types in active preamp design, I'm no longer sure what attenuator type the Hovland HP-100 in fact uses. Maybe the inner quote was a poorly stated argument for ladder over series. As Sherlock Holmes might say, when an impossible reading has been eliminated, an improbable one may be all that remains. Their switch itself sounds like the Shallco used in the Blue Circle Galatea's ladder, available from The Parts Connection to passive preamp enthusiasts like me.

Showing 1 response by carl_eber

Tom, this may sound like a stupid question, but when you use the phrase "return path", are you referring to resistors that are connected in parallel, or are you referring to resistors that are in series with the negative pole's path after they've passed the attenuator (or perhaps there is no difference)? I thought a "shunt" used both parallel and series resistors, an "L pad" if you will. It's the series resistor that determines how quiet the signal gets, and the parallel one is there to keep the impedance proportional to what it was before any "padding down" has taken place (i.e., the higher the series resistance, the lower the parallel resistor value will be)....Or so I gather.