Older versions of the disklavier could not detect accurate key/hammer position during playback as they only had a maximum of 2 beam, 4 position sensing to detect the velocity of the keys and hammers. This meant that playback accuracy depended on the data gathered during calibration of the piano and recorded in a 'piano table'. On very early ones, even this wasn't possible. Playback accuracy depended pretty much entirely on the skill of the engineer physically adjusting the sensor assemblies. To some extent, the last point also applies to newer playback only disklaviers as they don't have any sensors to allow calibration to the extent you find on full function models.
On the Mk IV, you have grey scale sensors which allow continuous sensing of the key and hammer positions.(Hammer sensors only appear on models from the DC3 upwards) This allows the Mk IV to not only sense velocities when a key is played, but also key release information. Using continuous feedback when playing back a piece allows it to do tricks like the ones you see in the "Ultimate Control" demos in the PianoSoft library included on the Mk IV. On the New Pro, you have an additional electro-magnetic sensor on each key solenoid. This additional feedback is used to give the 1024 steps for velocity, rather than the MIDI standard of 128(inc. 0).
Please note that I am using the term "accuracy" in the sense of being able to faithfully reproduce a performance, rather than ultimate resolution (except for the bit about the New Pro)
athomik
On May 20 2008, Kevin Goroway wrote:
>The Mark IV Disklaviers, do, in fact, "self regulate". Pro or non-pro.
>
>Although I think it was earlier described as something that is always ongoing, I'm under the impression that this only happens when the piano is specifically told to run its diagnostics.
>
>-Kevin
>
>