LCD: Making consistant contrast over large voltage range

The following are two circuits I came up with. Currently, I'm using a standard HD44780 LCD (1604A). which uses 5V for power and it is connected to other circuitry. All my circuitry is fed through a voltage regulator in a TO-220 type package that delivers a steady 5VDC. I'm also using the circuit pictured on the right to setup the contrast (simply a pull-down resistor).

In the future I will power my circuit with batteries and there will be times when the battery is ready to die. With my current setup, the contrast fades too much when the voltage is below 5V.

The simplest, yet tedius solution that everyone likes is to replace the 2.2K on the right with a potentiometer but the problem with that is I have to change the knob each time the battery state changes which is a pain.

So I thought of using some sort of simplistic voltage regulator circuit that I can easily add to a PCB from parts commonly available.

The internet suggests a circuit setup on the left. I selected the resistor and diode values because I believe based on ohms law and diode voltage drops that I won't burn anything and that the current to the VEE pin is roughly the same.

Then again, maybe I need to include another resistor or something.

Basically I'm trying to make a fixed version of the circuit shown on the right so that the LCD contrast stays at the same level at a much wider range of voltages (at least down to 3V).

How do I modify the circuit on the left to achieve this?

and the diode is a zener.

Well, first, that's not really a "pull down resistor."
Second, you need to be able to vary the voltage to Vo (what you're calling VEE?). For that, you need an adjustable voltage divider, which is usually provided for by a pot between 5v and ground, with the wiper going to Vo. A 10k or 20k pot configured thusly usually does the trick! Your circuits are providing a constant voltage, so the contrast will also remain constant.

But it would be impossible for a standard regulator to provide its specified fixed voltage if the input voltage is too low (for example: a half-dead battery). So the potentiometer won't do well for me.

Use PWM to feed Vee (1K resistor and 10uF capacitor from PIC PWM pin), and monitor the battery voltage with your micro-controller - adjust the Vee voltage via PWM in response to the falling voltage.

Or, much easier, use a decent battery and better design so you can maintain a steady 5V.

I use a 5K potentiometer as per Nigel's tutorials and have never had to readjust one after it is set. The one I measured is at 1.1 volts.
I don't change batteries until the circuit quits.

I just want to let everyone know that I'm not going to use the potentiometer setup.

Nigel Goodwin said:

Use PWM to feed Vee (1K resistor and 10uF capacitor from PIC PWM pin), and monitor the battery voltage with your micro-controller - adjust the Vee voltage via PWM in response to the falling voltage.

Click to expand...

I'm using an AT89C2051 but I'd like to avoid wasting CPU cycles for trying to determine battery voltage. But then again, a 10uF connected between VEE and GND/VCC is an idea but it won't be an ultimate solution.

Or, much easier, use a decent battery and better design so you can maintain a steady 5V.

Click to expand...

I've already done this part. I'm going to in the end use 4000mah RC car batteries. Even though they last longer than standard store batteries, I still want to make a regulator so people can see a low battery symbol on the screen in normal lighting even if the LCD backlight is a bit dim.

mik3ca said:

I'm using an AT89C2051 but I'd like to avoid wasting CPU cycles for trying to determine battery voltage. But then again, a 10uF connected between VEE and GND/VCC is an idea but it won't be an ultimate solution.

Click to expand...

What I proposed would be an 'ultimate' solution, as it meets all your requirements, assuming your processor has hardware PWM?.

Why do you consider monitoring the battery voltage a 'waste of CPU cycles'?, it takes almost no cycles at all - is your processor so busy it can't afford to do that?.