I'd also be tempted to see if it genuinely needs 63v, or if the designer specced that because they were cheap/no more expensive at the time. Could even be that that was a part they already had on their system. I've done that in the past.
If it's over-rated, then 50v caps are available that otherwise meet the spec. But you'll need to look at the circuit carefully to be sure.
Oh, it does.
The power supply on this device has a slightly unusual NAD feature, called PowerDrive.
The marketing blurb for this is total tosh ( it's hi-fi, after all ), but what it amounts to is this:
The main transformer has two sets of tappings: 'regular', and 'high voltage' (my terminology)
Each is connected to it's own bridge rectifier.
Under normal load, the 'regular' tappings feed the regular rectifier, and charge the reservoir caps to around +/- 50v.
If the load exceeds a threshold, then a pair of SCRs fire, and the higher-voltage DC from the 'high voltage' tappings and rectifier are switched on to the reservoir caps, pushing them up to slightly over 55v.
Basically a crappy solution to a poorly regulated under-spec PSU.
Make the base PSU able to hold up under load!
But there's not a marketing name for that...
To give them due credit, it does make the system somewhat more efficient. However, charging GBFO capacitors can't be done quickly, and it's not a big change from 50 to 55 V.
I've repaired power amplifiers with other odd power supply arrangements that improved the efficiency.
One had power rails at ±42V, ±84V and ±126 V (and about £200 worth of capacitors) and it would run power the main transistors from whichever rail would be enough. It would change very quickly, so running at 1 kHz and full power, each output cycle would be created using 8 changes of power rail voltage.
Another had two separate power supplies, which were in parallel but were switched to series for large voltage peaks.
Those used unregulated power supplies with really heavy transformers. Another arrangement that I have seen described is to have a switch-mode power supply, where the output voltage changes in response to what is being amplified, and stays a few volts above the output voltage required, following the waveform. The output transistors then remove the switching noise, but they never have a large voltage across them, so they don't get too hot.
±55 V will give around 1500 W into 4 Ohms if run as push-pull, so there is quite a lot of power to be handled. I've seen quite a few amplifiers where the power transistors have exploded, and anything that can keep the heating down is a help.