Regarding BAT's statement as quoted in Dave's post:
It is also important to understand the role of the output coupling capacitor, present at the output of BAT preamplifiers. It is the size of this capacitor that will dictate the minimum power amplifier input impedance, not the output resistance of the preamp.
That is absolutely true. It should be noted, though, that he refers to output resistance, not output impedance. The effects of the output coupling capacitor will be reflected in output impedance, which is what John Atkinson measures.
"Is the requirement for minimum input resistance dictated by the output resistance of your preamps?"
No. However, this is a popular misconception. It is common to apply some old rules-of -thumb to things like preamplifier to power amplifier interfaces and state that there should be some magic ratio between the output resistance of the preamp and the input resistance of its load. People commonly mention numbers in the 10:1 or 20:1 area, some as high as 100:1. Unfortunately, there is no truth to such claims. As many of us know, one can perfectly transmit a signal in a system where the load impedance is equal or even lower than the source impedance (witness any cable TV system).
The key factor is how much the output impedance VARIES as a function of frequency. A purely resistive output impedance will not vary at all. An output coupling capacitor will cause the output impedance to rise at low frequencies, to a degree that may or may not be significant in relation to the load impedance.
The 10:1 rule of thumb guideline is, IMO, properly stated as follows: The input impedance of the amp should IDEALLY be at least ten times greater than the output impedance of the preamp, at the frequency for which the output impedance of the preamp is highest.
The frequency for which the output impedance of the preamp is highest will usually be 20 Hz in the case of a preamp having an output coupling capacitor, such as most tube preamps.
That goal will assure that there is not an impedance compatibility issue. If that goal is not met, the results may or may not be ok, depending on the specific numbers that are involved, and especially on how much the preamp output impedance VARIES as a function of frequency.
IMO, based on JA's measurements the 10K minimum load recommendation for this preamp is marginal at best, and I would not assemble a system that way. Dave, I believe that the 10K/leg input impedance of the Bryston amp you referred to would be 20K as JA and others generally define it (the sum of the two legs), and even that resulted in a few db of rolloff at 20 Hz, as you indicated.
Regards,
-- Al