> what frequency is that figured for?
Typical "470K" input might be:
0.1uFd coupling cap
470K resistor to ground
100pFd grid capacitance
~~200 Megs grid leakage
So:
Ignoring the caps appropriately, we have 470K||200Meg which is 470K for any reasonable purpose
At utter DC the 0.1ufd is "infinite" (assuming the coupling cap is perfect insulation)
At 17Hz the 0.1uFd is around 100K capacitive, in series with the 470K. Ignoring quadrature, it would be 570K, which is "not much different" from 470K. Someone with a better grasp on the Imaginary might say it sums to like 480K, pretty-nearer "470K".
At 17KHz the 100pFd is around 100K capacitive to ground, so the "470K" is violated (common)
At 1KHz it is more like 1.7K in the coupling cap, 470K resistor, and 1.7Meg grid capacitance, so rough it up as around 370K and falling.
Over the interesting part of the spectrum:
DC - infinite
17Hz - ~~480K
170K - 470K
1KHz - 370K
17KHz - 100K
This input should perhaps be specified as "470K||100pFd".
It isn't, because we'll usually have hundreds of pFd of capacitance in the input cables, so if 100pFd is a problem, it isn't entirely the input's problem.
Another cheat is to say "at 1KHz", and leave you wondering what happens elsewhere: flat? slope? roller-coaster?
In commercial products, you can pretty well bet there is a chunk of capacitance, hundreds of pFd, in the input, just so that the rare customer living under a radio transmitter is less likely be be a problem. Most sources can drive that, and many listeners will not notice a dB of slump at 20KHz, but any listener will notice the hockey-game bursting into his Mozart.
On another hand I have a report of a tranny amp with 470K DC, a coupling cap, 47K to ground, and some bases which are probably >1Meg + 50pFd total. The user mistakenly "measured 470K", at DC, not understanding that there was another 47K hidden behind a cap. The manual claims "47K", but clearly the real midband impedance is no higher than 470K||47K= 42.7K. The stray capacitance is like 200K at 20KHz, so the impedance drops below 40K somewhere above 100KHz. I forget what the coupling cap was, but this amp is probably 50K to 40K from 5Hz to 50KHz, and no unexpected kinks beyond this range.
I have another report of an amp with 9K input. It is probably cap-coupled, so probably headed up to infinity at DC, though the rise should not start until well below the audio band. It probably tends to zero at high frequency, but it would take a HEAP of stray capcitance to swamp 9K resistive anywhere near the audio band. So this one is a solid 9K suck at "any audio frequency". We could perhaps call it "a 9k resistor in the audio band" and drop the word "impedance". Except "impedance" sounds more impressive on the sales-sheet.
Here's another amp with a bootstrapped input. From 100Hz down below 10Hz it looks a lot like an L-C resonance. Indeed the input network is similar to the contraption used in inductorless graphic EQs.
> when we say the input impedance ...is 10kohm...
In modern custom, you drive it with a low-Z source which is entirely happy about loads somewhat below 10K to anything higher, plus whatever cable capacitance is involved. i.e., you set up a "don't much care" condition. A chip op-amp with 100 ohm series resistor will drive anything over 2K and anything less than several hundred pFd with ease and authority.
