RGB 7-segment display

So I bought a spool (5 m) of 12 V RGB LED strip for testing. I didn't actually have a use for it planned, but it crossed my mind that I could make a 7-segment type numeric display out of it. I'm thinking 2.5 digits as 7 + 7 + 2 = 16 segments.

The idea

As a first target, it could display an ADC value and then perhaps a temperature. If I can get a temp sensor inside a sauna, it could neatly show the progress from ambient to 60 - 100 °C. Being an RGB strip, I could make it change colour based on the reading as well. Target temps would give yellow to red numbers.

Wiring the RGBV lines will take some work. I'll need some substrate to tape the string on and it might be neatest to drill through it and do the wiring on the backside. Finally, I could add a protective sheet of something on top and possibly another board to protect the back.

The claimed 72 W would mean 0.72 W and 0.72 / 12 V = 60 mA per 5 cm. Having a battery option would be nice, but 12 V and 32 * 0.06 A = 1.92 A at full blast doesn't make that an obviously easy/cheap option. Logic will also need a regulator or a second supply voltage.

The strip is divided into 5 cm pieces, each containing three series connected 5050 RGB LEDs. Anodes (+) are common and they have resistors supposedly matching 12 V nominal voltage. For some reading distance, two lengths per segment seems nice. That will make the digits maybe 23 cm high and 12 wide (plus blank margins). Expecting 120 mA per segment fully lit. Total strip used will be 16 * 10 cm = 160 cm out of a spool of 500 cm. I will cut off some convenient length from one end that has a connector in order to save that, since it won't be useful here, but might be handy for something else.

Wiring scheme

It's 16 segments and three LED chains (R,G,B), so 48 switchable elements. That's obviously quite a few to switch separately. Six 8-line driver chips or seven 7-line ones, for example. 48 discrete transistors is a few as well. 8 * 120 mA = 960 mA per chip, too.

Multiplexing could be done by dividing the display into two areas and pulling two sets of RGB lines down simultaneously. One segment from both areas would get positive voltage at one time. That would mean six low sides (r1 g1 b1 r2 g2 b2) and eight high sides (two segments each).

The initial idea would be an ULN2003 and some PNP darlingtons I have around. I'll need to see if that's feasible. The high sides may need another chip or other scheme to get the base voltages right. Eight lines is a bit over the capacity of the most common chips like hex inverters and the ULN2003 . That's also 8 + 6 = 14 lines to drive from a controller, which sounds easy but does limit controller choice or further I/O.

Multiplexing will also lower the final light output by the effective duty cycle (1/6 or 1/8).

Measurements

12 pieces, 60 cm, 36 LEDs, all lit at 12 V drew 570 mA or close to 7 W. From there, 2 pieces should be under 100 mA. Prediction was 120 mA.

So, even an S9015 might survive :)