LCD Backlight control brightness adjustment & cathode pin purpose

I'm working on an STM32-based project to which I'm planning to connect an LCD (in my case, it should be 800x480, specifics irrelevant for the questions). I do have disco development boards with LCDs (F746 and F469), but I have never integrated custom LCD with custom MCU outside the dev boards. Did play with the dev boards for a while to get acquainted with them.

I've googled some questions of mine, of course, but I wanto to crystalize my knowledge before it solidifies in KiCad and will have a chance to explode in my face.

If you want some specific example, here's the LCD I'm planning to work with, so if you want to refer to some specifics, you can do it with this: LCD Datasheet PDF English

As for questions:

1. LCDs in general have separate backlight cathode (-) pin. From what I understand, it can be connected to ground, but doesn't have to be. It could also be connected to some inverting charge pump so that the anode voltage doesn't have to be that high. But ground also works. What matters is the voltage difference between anode and cathode, since it's just a circuit with a few LEDs in there (I remember about current limiting resistor). The only thing to watch out for, as I see it, is the ground current path in case of PWM-control (make sure it doesn't induce noise into other stuff). Please, confirm or debunk these statements of mine (and why if they're wrong).
2. Backlight can be controlled by PWM or by constant current source. First, how does it affect LED lifetime? (for example, the LCD I provided has 50,000h LED life ~ almost 6 years constant ON). I do realize it's very unlikely that a device with very specific purpose like mine will ever exhaust that time. But since I'm asking all the stuff about LCDs here, why not throw in all questions that interest me. So, LED-lifetime: PWM vs Constant Current? PWM is ON for less time, but there is turn on and off cycles and stuff. So if anyone knows, I would be greatful.
3. Other reasons for constant current or PWM? I mean, nothing stops me from setting up a Timer to MCU pin to PWM backlight chip's enable, but I guess I could also control a constant current source, although it seems to me it became de-facto standard to use PWM (I fix laptops' logic boards and also occasionally see tablets and phones, even had one TV to fix, so I meet real circuits made by proper engineers unlike myself, and I use schematics a lot). Also, constant current circuit may seem more difficult and expensive to control for my inexperienced eyes comparing to a single boost converter with enable pin straight from the MCU. Same story here as in question 1 here, please, confirm or debunk my view of the things.

Before I got any replies, I think I'm considering using PWM, and since my LCD wants ~30V, I will need a boost converter for it. Then there are questions about PWM frequency and stuff, but I guess this is where I'll have to take into account all info from LCD datasheet (no mention of backlight control other than diode parallel-sequencial structure) and specific backlight IC

To your Question 1: You are correct on why LCDs have a separate cathode. In fact, a lot of LCDs I have seen also have a separate anode, that way you can implement a completely separate circuit for the backlight and avoid coupling to your main circuit. PWM noise is naturally an issue but getting the higher voltage often requires boosters which add even more frequencies and noise to that circuit.

To your Question 2: The biggest influence on LED life is heat/power dissipation. Since most LEDs have a somewhat steady voltage, heat is directly proportional to current. It doesn't matter as much if you control current via a constant current source or PWM, what matters is the total charge/time ratio. The LCD spec sheet clarifies that the 50,000 LED life is defined by allowing the brightness to decrease to 50% of the original. This may or may not be acceptable to you so keep that in mind. Also, it specifies that this is at 25 degrees C and 20mA going through each LED string. There is no information on how this "life" is shortened if for example, you send through 25 mA, or how this life is extended if you send through 15 mA. Lastly, notice the 25-degree qualification, if the LCD is in a place where heat accumulates, the LCD will experience temperatures much higher than 25 degrees C, which will also wear the LEDs down much faster than the 50,000 hours. So in summary, it all comes down to heat; the hotter an LED gets the smallest the useful life.

To your Question 3: It all depends on the required efficiency and the specific parts you are using. If you use a constant current source that maintains the desired current by "absorbing" extra voltage and no PWM, you are wasting that energy much like a linear voltage regulator wastes the extra energy (i.e., converts it to heat) while maintaining a fixed output voltage. PWM on the other hand is more efficient as power is cut off. That alone is a big plus for using PWM versus constant current, especially when you want to be able to control the actual brightness. There is nothing wrong with using a constant current source with PWM to its enable line, especially if the current source was designed for PWM control but you can also use PWM on a low-side mosfet and accomplish the same thing. Keep in mind that a constant current source as implemented by most ICs requires some headroom above the maximum voltage creating the need for an even higher voltage to start with.

In your case, using a booster putting out 30V will allow you to power the LCD at full brightness. So you can either put an LED constant current chip that allows PWM control at the bottom but have to increase the 30V with the headroom required by the chip or just add a logic-level MOSFET and apply the same PWM signal to the gate.

• Thanks for your detailed answer. I have overlooked many details about lifetime, it's good you brought it up in detail. Seems like I was thinking right about everything else, for the most part at least. Turned out, it's not easy at all to find some 3.3V->30V boost IC (and 5$for a chip is a no-no). Looks like LM2733Y (2$ apiece) will do the job, even if efficiency is going to be 70-75% (with such a voltage jump, there is little chance for more efficiency, I guess, but PWM should save me some battery). Do you mind sharing your opinion on it?
– Ilya
Aug 23, 2021 at 14:07
• LM2733 Datasheet: ti.com/lit/ds/symlink/… It doesn't specify if I can PWM the enable pin or with what frequency, so I'm hesitant
– Ilya
Aug 23, 2021 at 14:07
• Glad the answer was useful (mark it as 'answered' if appropriate). Getting 30V from 3.3V is going to cost you in efficiency. Looking at the spec sheet for the LM2733, I don't even see charts for 30V output and 3.3V input. On page 8, the highest Vo for a 3.3V is 20V and the efficiency is under 70%. You will have to test this to find out if it will give you the 30V you'll need. Regarding PWM to the enable pin, it is unclear if it will work, and if so, you will have to carefully experiment to find the lowest frequency PWM that allows the booster to work while being invisible on the LCD. Aug 23, 2021 at 15:51
• The "Y" version has a chart for 3.3V->30V and it's around 70-75% at the currents that interest me; Still don't like it. Maybe I should just find an LCD that doesn't require going that high in voltage (finding good LCD with proper documentation and reliable supplier on the internet turned out to be MUCH harder than PCB design itself, and much more boring). As for accepting answer, I was hoping someone else would add something to it, but it's too late for that, I guess
– Ilya
Aug 24, 2021 at 7:59