I can only assume you are talking about the LED's you are driving. That's not what I was asking.
Was interested in how you circumvent the LDR properties which are not only light sensitive but also voltage dependent.
I assume the audio input is attenuated by the LDR's which are both light and voltage dependent.
One series and one shunt as most of these attenuators work.
When there is an audio voltage across the LDR harmonics are added because of non-linearities.
Did you solve this by 'modulating' the LED current with the audio input voltage to compensate for the non linearities ?
You are I think confusing open faced LDR types with encapsulated audio purpose LDR's which are not light sensitive.
The attached datasheet shows the NSL32SR3 and its specifications.
http://lunainc.com/wp-content/uploads/2016/06/NSL-32SR3.pdf
I hope to turn around your understanding of audio purpose LDR's. Audio purpose encapsulated LDR's are actually quite linear in terms of reproducing music, however they require quite a leap of faith away from conventional thinking to start showing you their actual capability, namely that the anode and cathode of the led are properly driven and not just assumed as " anything will do to drive after all its just an led ", somewhat lazy modicum of thinking. Have I got your attention ?, or are we lost already in the anything will do and it can't make any difference room. If so let me open the door for you.
A lazy LDR circuit would see a 7805 with parallel resistance and parallel capacitance - and internal resistance of 4.3k between output and ground
in parallel with the anode and cathode. We begin by removing any such parallels.
All of the needs of a NSL32SR3 in terms of being an attenuator are met by having just 2ma or less of current available,we begin by creating a high impedance away from the general supply, we then make the series anodes separated from shunt anodes - however some minor conversation still between the two is beneficial. The signal side is a simple L pad there is no wired modulation or cross connection from signal side to anode and cathode side. Whatever we do on the anode and cathode ( you will have to accept my experience of 11 years ) either improves or does not improve the signal side ability to be a average to good vs exceptionally good variable resistor.
Where the circuitry attends to all of the anode and cathode needs ( and they are believe me - beyond vast ) the LDR' signal side if you are good or beginning to be good at what they really need - responds by widening the audio sound stage to accommodate, and in the process I perceive ( awaiting measured confirmation ) becoming more linear.
I know you dislike subjective ( which I am doing my best to change with having my product measured ) but here is some anecdotal independent comment. Look away if you really have to.
https://www.stereo.net.au/forums/to...peed-diy-ldr-pre-vs-stereo-coffee-diy/?page=2
An approach I experimented with many years ago was to use 4069 inverters to see how far one could invert and reinvert each inverter
with DC supply the anodes and see if it changed the resultant audio, I stopped at 18 inversions realising the importance of creating a very high impedance. My efforts were then toward having the volume potentiometer in the cathode half and sensing and re using the available current.
I no longer use 4069's , but instead a combination of p channel mosfets , n channel mosfets, thyristors, voltage reference ic, a dual opamp , TL431's LM336's ( the 336 is a great device ) LM317's , 2N5088's etc.
I hope that begins to show there is far more than you first think or have read elsewhere to audio purpose LDR's which are and will always be
a great alternative method of attenuation.