In brief: LED output depends on the current that flows into them, regardless of the voltage drop that the LED incurs in. Device-to-device variability, and other factors, make the voltage drop different in different LEDs, which makes a constant-voltage power supply (the one currently used in the design) a bit less stable and a bit more subject to calibration and calibration drifts.
A simple off-the-shelf linear regulator such as the IHB120-0.2 (if one would like to power all the LEDs in a single strip) or lower voltage such as the HB48-0.5-AG (spitting the LEDs in multiple strips, probably better for maintenance) can be used to very stably power a voltage-to-current regulator such as the CL220N5-G which is inexpensive and can be used either for single strip or for multiple strips in parallel as described in its datasheet (Fig 3-5).
This setup would guarantee a more constant LED-to-LED brightness as well as a more stable in time brightness (since the residual ripple is extremely low). Moreover, there is no need for current-limiting resistors, and so if they are left out of the board no power dissipation on them (however the linear voltage supply will have an efficiency of just about 60%, but that is already similar to what is installed on the IPS board).
Of course all of this could be implemented on the IPS itself rather than as an external solution as described here
This is potentially related to #40 and #14 and definitely to https://github.com/gzweigle/DIY-Grand-Digital-Piano/tree/main/hardware/releases/aceA00 as well as https://github.com/davidedelvento/piano-conversion/tree/QRE1113GR-4x-four-holes/hardware/sensorboards/2040-centered-4x-QRE1113GR
In brief: LED output depends on the current that flows into them, regardless of the voltage drop that the LED incurs in. Device-to-device variability, and other factors, make the voltage drop different in different LEDs, which makes a constant-voltage power supply (the one currently used in the design) a bit less stable and a bit more subject to calibration and calibration drifts.
A simple off-the-shelf linear regulator such as the IHB120-0.2 (if one would like to power all the LEDs in a single strip) or lower voltage such as the HB48-0.5-AG (spitting the LEDs in multiple strips, probably better for maintenance) can be used to very stably power a voltage-to-current regulator such as the CL220N5-G which is inexpensive and can be used either for single strip or for multiple strips in parallel as described in its datasheet (Fig 3-5).
This setup would guarantee a more constant LED-to-LED brightness as well as a more stable in time brightness (since the residual ripple is extremely low). Moreover, there is no need for current-limiting resistors, and so if they are left out of the board no power dissipation on them (however the linear voltage supply will have an efficiency of just about 60%, but that is already similar to what is installed on the IPS board).
Of course all of this could be implemented on the IPS itself rather than as an external solution as described here