Things are a little more complicated than what might appear. There are several other factors to consider, this is not exhaustive. For general consumption.
and yes, impedance matching between pre and amp is important, as i mentioned.
Not so much as one normally thinks. RCA Radiotron Designers Handbook suggests a 5:1 ratio, amp input Z to preamplifier output Z. I suggest, as you have heard, 10:1 to be safe. But that is normally calculated at midband. What is really important is across the entire audio band.
My Preamplifiers are approximately 1.9k output Z, and works fine to 20k amp input Z. So a buffer stage is not necessary. Why not? What is the scoop.
1) Let's examine high frequency first. Example: we have an IC capacitance of 100pf and amplifier input capacitance of 50pf, so 150pf in parallel with the 100k input resistor (100k input Z at 1khz).
The reactance of 150pf at 20khz is ~53k ohms. So at 20khz, we actually have 100k ohm in paralleled with 53k ohms of capacitive reactance (called "I").
However, the preamplifier output Z is in parallel with "I" and 100K resistor, so not that bad. With 1.9k output Z, the high frequency response exceeds half a mhz (550khz) before -3db.
Believe it or not, the major limit on high frequency response is the volume control to tube grid input capacitance (including Miller effect). Not only that, the minimum frequency response (FR) occurs at mid resistance setting of the volume control. Using a 100k volume pot limits the HF more than a 50k or 25k pot.
2) Let's examine the low frequency response. Pretty easy as a coupling capacitor and following grid resistor. Here is some data.
Frequency 10hz. Grid resistor 100k ohms.
Coupling Capacitor -db
1uf .114
.82 .162
.68 .235
.47 .47
.33 .883
.22 1.76
.1 5.5 db
With 50k grid resistor.
1uf .38db
.1 10.
That is just one stage coupling. With multiple stages, dbs add. By the way, the 1uf coupling capacitor, - 0.116db at 10hz may seem small enough to not be perceived, but when additional stages and/or the OPT are considered, don't be so sure. By the way, bloated bass in tube amps is not so much damping factor, but more likely a power supply design problem.
my first foray into a tube preamp was the arc sp-9mkll. in a word - it totally sucked. it was worse than the adcom gfp-1a i was using at the time. reading this thread, and seeing that tmazz uses an sp-9 in his system made me go and read a lot of articles on it, and what folks thought, good and/or bad. one thing i gleaned was that the sp9 wants to see an amp w/minimum 60k ohm input impedance. i knew nothing about that back then. the amp i was using at the time has an input impedance of 22k ohms. maybe that was why everything sounded like it had molasses poured over it; the sound was so thick, heavy and lifeless.
With a smaller 22k ohms, the sound should have been thinner than using 60k ohms minimum, not heavier. This leaves several possibilities.
One, the room is quite bass resonant.
Two, the speakers are not correctly matched for the venue,
Three, the speakers are bass heavy to begin with, due to either their venue or electronic components they used
Four, the other electronic components in your system are not designed properly, thus bass heavy
The proper way I have found is to design the electronics first, since the preamplifier and amplifier can be listening tested for accuracy. After that, then I worked teeter/totter with speaker/venue.
even so, i wonder, as the sp-9 mkll's output impedance is 250 ohms, and the standard rule of thumb is the amp's input impedance should be at least 10 times greater. and tho arc recommends 60k ohm amp input, it also says "min 20k ohms". but, in any event, preamp-amp impedance matching is definitely something to consider.
Although the rule of thumb 10:1 is good, as shown above, the real culprit is the output capacitor size and quality, and also the power supply, which I did not address.
There are some weird "rules of thumb", such as concerning the -3db point, way out of line.
Cheers
Steve