The Truth Should Out Page 3
The ABX cannot, however, switch a device's inputs and outputs simultaneously. If you are comparing outputs from two devices, their inputs must remain in parallel across the source. This could be a liability under certain conditions, as when two power amplifiers present different load impedances to a preamplifier which does best with a fairly high load across its outputs. The amplifier which would normally sound best with that preamp will be handicapped by the reduction of its input impedance due to the paralleled impedance of the other amplifier, and neither amp may sound as good with that preamp as it could. This may or may not be significant with any pair of devices, but it is the kind of uncontrolled variable which tends to undermine the validity of any scientific experiment.
And speaking of power amplifiers, the ABX is not really designed for testing them all. While the manual—which itself is not as clear nor comprehensive as it could be --shows how the relay box can be connected so as to switch loudspeakers between two power amps, it cites the relays as being rated at 100 milliamps, which equates to a power-handling capacity of about 0.08 watts into an 8-ohm load. With most loudspeakers, this would elicit an output level of about 80dB at 1 meter, which is not quite adequate for serious analytical listening. (Most audiophile systems are played at 90 to 100dB on peaks.) Adding further to the difficulty of handling power amplifiers with the ABX is the fact that it has no level-set adjustments with which to match the gain of the amplifiers. Level matching is extremely important in A/B testing, so if the more-sensitive of the two power amplifiers does not have its own input level-set adjusts, these should be added externally. (They should be 100k ohm log potentiometers, 0.5W, located as close to the power amp as possible to minimize HF losses. These pots are available as "raw" parts from any electronics supply store, but they are not commercially available as preassembled devices from any source. I feel that ABX should at least make them available as an extra-cost option.)
Even with preamps (instead of power amps), there are a few problems which ABX should address themselves to. The lack of input switching means that one cannot do valid comparisons between preamps with a phono cartridge as the signal source. And since the biggest difference between the sounds of preamps is the way they handle impulse material from cartridges, it is essential for preamp evaluation that a phono source be used. With the preamp inputs paralleled, the cartridge will see half of its recommended load impedance, resulting (usually) in a rolled-off high end. Inserting a resistor in each channel to bring its load up to the requisite 47k ohms will reduce its output by 6dB, and cranking up the gain on both preamps to compensate may make hum and/or hiss audible. And if one preamp happens to be a little noisier than the other, any listener can take the increase or decrease of noise as an unfair clue as to which preamp is currently occupying the X slot. (The best approach to preamp testing would be to lift the phono-input load resistors from one preamp, allowing the other to provide the load for both.)
As it now stands, the ABX is most easily used for comparisons between tape recorders, signal processors, and the high-level sections of preamplifiers. But that is not the end of potential problems with this. Some devices, due to poor design, leaky capacitors or misadjustment, have small amounts of DC across their output or input connections. When one of these switched DC-offset connections is switched into the system, it will cause a click which, again, can be picked up by any perceptive listener as a clue to what component has just been switched in. High-value low-wattage resistors from the signal path to ground at all of the receptacles on the ABX will prevent this with most components, although not with power amplifiers. For valid comparisons of amps, any DC offset adjustments in them should first be trimmed to eliminate every trace of switching clicks.