I think I have finally found the answer, and the reason I probably don't have to worry too much about the Naim amp's high input sensitivity. The Naim manual states that there is: "Overload margin, on all inputs at all audio frequencies 40db", which is a multiple of x100 in voltage terms. With an input sensitivity of 75mV that implies a maximum 7.5V before the input overloads. I have read elsewhere that Naim pre-amps can cope with 7V at the input without a problem, and I presume this is where that figure comes from. Of course, that is around x10 the output voltage, so you would barely be able to move the volume control off the zero mark at 7.5V input. However, it also means it could cope quite happily with the Qutest at 3V as well, but again that would give you very little range on the volume control.
I misread the tape output voltage, which is 75mV. Presumably there is attenuation in the circuit somewhere to drop it to that level. The input sensitivity of the Headline is actually 820mV, so no chance of it overloading.
I put the questions to Kit Ryan, who very kindly responded with the figures from his modified version of the Naim line out boards that produce the signal going to the power amp. At default settings the figures are very similar to the stock Naim setup.
The maximum output from the pre-amp towards the power-amp is 5V before the pre-amp will clip. The input sensitivity of my power amp is 1V, so the power-amp will clip well before the pre-amp does. Most power-amp sensitivities are in the 1-2V range so the same will apply to any make of power-amp.
The inputs can accept any voltage (up to around 7.5V), but the stock pre-amp circuit will apply a gain of x13.3 when the volume control is turned right round to maximum. Where the "input sensitivity" figure comes from is the output voltage of 1V (to match the power-amp sensitivity) divided by the gain multiplier of x13.3, which equals 75mV (0.075V). Because of the high gain multiplier, if you put a higher voltage than this at the input, and then turn the volume up full, you could drive the
power-amp into clipping. The pre-amp will be fine.
Very usefully, Kit Ryan's replacement boards have three jumper positions to modify the gain multiplier - either x13.3 (stock), x4 and x1.3. So the input sensitivities at each jumper position are as follows:
Max = 1V/13.3 = 0.075V
Mid = 1V/4 = 0.25V
Min = 1V/1.3 = 0.77V
At the moment I have the gain jumper at x4, which means input sensitivity is 0.250V. I also have the Qutest set at 2V, so I cannot use more than 12.5% (250mV/2V) of the volume control range before there is a danger of the power-amp clipping. So I probably need to reduce the gain multiplier, or Qutest voltage, or both to give me more useable range from the volume control. My options are:
x1.3 at 2V = 0.77V / 2V = 38.5%
x4 at 1V = 0.25V / 1V = 25%
x1.3 at 1V = 0.77V / 1V = 77%
I need to give that some further thought. On the assumption that the M Scaler reduces the 1V output from the Qutest to 0.73V, at the x1.3 jumper setting I could use almost 100% of the volume control rotation before the power-amp is in any danger of clipping.
EDIT: Re the tape out, it is buffered, but the gain is effectively x1 the input voltage. So if the input from the Qutest is 2V then that is what goes through to the Headline. The Headline's maximum output voltage to the headphone socket is 3.7V, and the gain factor is x4.73 (13.5dB), implying an input sensitivity at maximum volume of 0.63V. That means I could use 86% of the Headline's volume control range without it clipping if I use the Qutest at 1V, being an actual 0.73V as modified by the M Scaler.