Temperature to Light Intensity with voltage linear

[PhillDubya]

I am currently helping design a circuit that would alter voltage linearly to changes in temperature.

Temp goes up, and red light gets brighter due to linear increase in voltage; while blue light intensity goes down, due to a inversely linear decrease in voltage with temperature.

The opposite would be true when temps get cooler. I.E., the blue light gets brighter as the temps go down, and the red light gets dimmer.

What I am considering is using PWM from a 555, and sending that output to an array of 3904npn transistors which will input to a ULN2803 Darlington array to drive the led's.

One 555, plus one 3904 array, plus one ULN2803 per LED set. This would be set -up for increase in temps from a diode.

Another one just the same, except set up for a decrease in temps.

I know a uprocessor using an ADC + PWM clock might be a solution, however, that is off the table due to the complexity, as the person I am helping has no experinece with it what so ever.

Any ideas for improvement, especially for an easier, if possible, solution.


Your inputs are appreciated.

Thanks

Edit: to be clear, there will be a lot of LED's thus the reason for a driver, and why I am not running them directly from the 555.

Probably 20 LED's per color, so roughly worse case, 400mA consumption.

 

[crutschow]

You could just invert the output from your PWM to drive the other set of diodes. That way when the duty-cycle increases to brighten one set of LEDs the duty-cycle will be decreasing to dim the other set of LEDs.

 

[ccurtis]

I don't understand the need for the 3904s. With a 5 volt input, each darlington in a ULN2803 consumes 1.5 mA input current. One 555 can drive 200 mA. That's 133 LEDs (200 ÷ 1.5), at the very minimum. More, if you drive more than one LED with each darlington output.