# Improving load regulation on voltage regulator

Basically, I'm building a device that powers a number of LEDs; each LED needs to have as consistent a power as possible (each eats 25mA at 3.3V) across the number of LEDs being lit up.

I'm using a L78L33C 3.3V regulator drawing from a 5V wall wart to supply power to the LEDs but I notice that each L78L33C has a load regulation of 60mV from 1mA to 100mA.

Since I need each LED to have extremely consistent brightness (+-20mV), I'm worried about the 60mV dropout.

Is there any way to ensure the power output for each LED stays exactly the same regardless of the number of LEDs I'm powering? Will putting multiple L78L33C in parallel with the LEDs help?

L78L33C datasheet: Here

• What exactly do you mean by "extremely consistent brightness." Unless you are buying carefully binned LEDs, just providing the same current won't cut it.
– jonk
Apr 26, 2019 at 7:50
• @jonk I could quantify our required brightness but I think it would just add unnecessary complexity to my question. For simplicity, the supply voltage should only fluctuate by +-20mV between different readings. We do QC our LEDs in controlled environments to ensure they output the same brightness with the same current. Apr 26, 2019 at 8:12
• So it sounds like you are binning the LEDs or selecting them based on light output. Is the environement temperature controlled too? What is the $V_f$ of the LEDs and how are you controlling the current to the LEDs or are you supplying 3.3 V directly across the LEDs? Apr 26, 2019 at 8:26
• @Transistor Yes, we do maintain a reasonably consistent temperature when testing the LEDs. Vf is 3.2V with a maximum of 4V. We supply the voltage directly to the LEDs, no current regulation. Apr 26, 2019 at 11:17
• I think your readers are wondering why you are running your LEDs on constant voltage and expecting consistent results. LEDs should be run on controlled current. Can you elaborate? Reading your question and comments it seems that you are missing some basic understanding of correct LED operation and use. Apr 26, 2019 at 12:14

1. Dropout voltage on the 78L33 is 1.7V. You are right at the limit. If the 5V drops just a tiny bit, that will be reflected in your output voltage.

2. The load regulation of the 78L33 is 60mV. If that is too much for your application, you need a better regulator.

3. LEDs are current driven. You ought to be regulating the current to each LED. You need a constant current source for your LEDs to ensure each gets the same current.

4. Even perfectly matched LED currents won't guarantee perfectly matched LED brightness. You'll either have to select LEDs that deliver the same brightness (test and measure LEDs for brightness) or you need to regulate the current with feedback from a light detector (photo diode or photo transistor or other sensor.)

If you just want a uniform brightness for lighting (eyeball uniform) then controlled current with all LEDs from the same model ought to do.

If you need brightness matched for some kind of precision measurements, you'll need to go with current controlled with feedback.

• As commented my question, yes we do QC our LEDs to meet our tolerance, even if our manufacturer can't. For 2), can I have a dedicated L78L33C for each LED to mitigate the effects of the load regulation? For 3), can I not assume that for the same voltage, each LED will take the same current? Or does it vary due to manufacturing tolerances? Does this mean we will have to QC for V/I across our LEDs as well? Apr 26, 2019 at 11:09
• I don't see any mention of how you are controlling the current. If you are just connecting them straight to 3.3V, then you will have problems. The brightness of an LED is a function of current. The voltage required for a particular current varies sharply over a short range, and varies by temperature as well.
– JRE
Apr 26, 2019 at 11:42
• A few millivolts of difference can make a large difference in current (and brightness.) This is why you need to control current rather than voltage.
– JRE
Apr 26, 2019 at 11:44
• Thanks for your answer. Apr 29, 2019 at 0:46