Hi Tim,
The rule of thumb guideline is commonly stated that way, but that is a misleading oversimplification IMO. In particular, it does not take into account that designs which use a coupling capacitor at their output, such as the majority of tube-based preamps and source components, will have output impedances that rise significantly at deep bass frequencies from their specified nominal values (which usually are based on a mid-range frequency such as 1 kHz). An example, which is by no means as extreme as many others I have seen, would be the measurements I linked to above for the CJ preamp you were considering.
I would state the rule of thumb guideline as follows (I’m quoting this from a post I made in another recent thread:
The input impedance of the amp (or other component that is receiving a line-level input signal) should be at least 10 times the output impedance of the preamp or line-level source component that is driving it, at the frequency within the audible range for which that output impedance is highest. Which in the case of preamps or source components having capacitively coupled outputs (such as the majority of tube preamps) will usually be at 20 Hz. And the output impedance at that frequency will often be far higher than the specified output impedance (which is usually based on a mid-range frequency such as 1 kHz), because the impedance of a capacitor rises as frequency decreases.
That doesn’t mean that there will necessarily be a problem if the guideline is not met. It depends on how the output impedance **varies** as a function of frequency. What it means is that there **won’t** be an impedance compatibility problem if the guideline **is** met.
If Stereophile has reviewed the preamp or source component, the measurements section of the review will usually indicate the output impedance at 20 Hz as well as at other frequencies. But if only a nominal impedance can be determined, such as a manufacturer’s specification that is presumably at a mid-range frequency, to be safe I would suggest a ratio of 50x or preferably even 75x.
Regarding the ARC LS-12, though, I would consider all of this to be superseded by ARC’s recommendation of a 10K minimum load, as indicated at the arcdb.ws link I provided.
As you are probably aware, the load impedance seen by the preamp would be the product (multiplication) of the input impedances of the sub and the power amp divided by the sum of those two numbers. The result will be less than either of the two numbers.
Also, if you were to connect the balanced output of the preamp to one of those components and the unbalanced output to the other, that result would most likely apply to the signal on XLR pin 2 (as well as to the signal on the RCA connector), but not to the signal on XLR pin 3, since on schematics I have seen for various other ARC preamp designs XLR pin 2 and the RCA connector are wired directly together. (And my understanding is that it is not uncommon for that to be done in many designs from other manufacturers). Which is probably just as undesirable as applying the excessive load to both of the signals in the balanced signal pair, and perhaps even more undesirable since it would unbalance the impedances and the amplitudes of the two balanced signal lines, at least slightly.
Best regards,
-- Al