Why I built an LED luminosity tuner

5/31/26   -  Project completed

The device is testing very well, accuracy is right on, it's temperature and displayed readings are stable.  The LEDs are within the tolerance of my light meter when measured on the eeZ LED and powered by lab supply with the resistor values shown. 

It looks like I started this over 6 years ago! Glad I finally found some time to finish, the folks who have one love them.

5/16/26
Worst case scenario: A 48" round trip for 28ga wires to check SMD LED in position:

28 ga wire is 64.9Ω/1000 ft @ ambient, which comes to 64.9mΩ per ft.  Total resistance for 48" is 259.6 mΩ

At 20mA, we are at 5.19mV of voltage drop across the wire, this amounts to a  173uA error at max current (this is 0.8% error)

At 1 mA, the drop is 260uV which is 8.6uA 

At this time, the error introduced by 4 ft of 28ga wire (worst case) is within the margin of error and not significant to justify kelvin connection. 

5/13/26 
1. Experimenting with log amplifier so that the lower current gets more throw since the human eyes see logarithmically, this is challenging as adding a diode to the feedback limits the dimming range, I will have to possibly drive a BJT and add a subcircuit that provides delta R to the feedback.  This will take some thought and experimentation. 

2. Some of the feedback from a design review on LinkedIn has brought up the topic of adding easy SMD support.  There are several features I would like to implement when this is done

  a. kelvin connection

  b. A method of providing hands free clips so that 0603 SMD LEDs can be measured in place for LEDs under a control panel and or diffusers. 

NOTE:  this can be achieved by soldering wires to the LED and attaching to the eeZ LED as it is currently designed.  However without kelvin connectors, the measurements are slightly off due to the voltage drop on the wire leads.   Analysis and assessment needed to understand the effect on variance and significance.

UPDATES:

5/3/26  -     Software update: 

Show resistance as 1% E96 values available on the market.  

Implemented 96 value array was saved to flash (free up RAM)

const int E96_TABLE[] PROGMEM = { 100, 102, 105, 107, 110, 113, 115, 118, 121, 124, 127, 130, 133, 137, 140, 143, 147, 150, 154, 158, 162, 165, 169, 174, 178, 182, 187, 191, 196, 200, 205, 210, 215, 221, 226, 232, 237, 243, 249, 255, 261, 267, 274, 280, 287, 294, 301, 309, 316, 324, 332, 340, 348, 357, 365, 374, 383, 392, 402, 412, 422, 432, 442, 453, 464, 475, 487, 499, 511, 523, 536, 549, 562, 576, 590, 604, 619, 634, 649, 665, 681, 698, 715, 732, 750, 768, 787, 806, 825, 845, 866, 887, 909, 931, 953, 976 };

The measured resistance is converted to 100's value and is compared with a switch case to next highest value.  Then the value is changed back to show the correct decade. 

5/5/26 -   Implemented 14 bit oversampling of the 10 bit ADC.  It brings the current resolution from 35uA to 2.2uA. I had a lot of math overflow issues to contend with, have had to use 32bit UL casting to ensure no overflow.