Usage of 74HC595 for LED cubes: isn't it bad?

Other Hardware General Electronics
1

I've stumbled upon many projects, questions and instructables about using the widely available 74HC595 shift register to build LED cubes.

Just before wiring my own (it's a 5^3) I had a look at the datasheet and I've been surprised by reading the maximum allowed current through the IC is 70mA (absolute maximum): I do understand it's not common to light up all the LEDs on a plane, still, when such event occurs each IC will be asked to sink or source (depending on the cube configuration) 160mA, more than double it's absolute maximum!

My question is: am I wrongly reading the datasheet, missing something else or should all those LED cubes out there keep burning shift registers quite often? Even taking in consideration that kind of current is flowing for a fraction of a second, it's so much above the spec those registers should not survive for long.

Thanks to any good shepard willing to shine some light here...

2

Its a logic chip, not an LED driver. Just set the current lower (high brightness LEDs
only need a few mA to be plenty bright).

3

You are right, but many examples use either 68 Ohm or 220 Ohm current limiting resistors: at 5V that's way above the 8mA we should be constrained to in order to stay within specs.

Now, my question isn't "how can I use 74HC595 to drive my LED cube?" but "shouldn't we consider all those cubes faulty and bad design?" or, if you prefer, "how come those cubes seem so popular and don't burn out the shift registers?"

4

Use cd74ac164 to source current to individual columns, and N-channel MOSFETs to sink current from common layer;
Or, use '595 to drive P-channel MOSFETs to source current to common layer and TPIC6B595 to sink current from individual columns. TPIC6B595 can only pull low, so can be used to pull gate of P-channel MOSFET low to turn them on.

5

rlogiacco:
My question is: am I wrongly reading the datasheet, missing something else or should all those LED cubes out there keep burning shift registers quite often? Even taking in consideration that kind of current is flowing for a fraction of a second, it's so much above the spec those registers should not survive for long.

Thanks to any good shepard willing to shine some light here...

Short version: Nobody ever reads datasheets.

rlogiacco:
"shouldn't we consider all those cubes faulty and bad design?"

Yes. A lot of their builders don't even use resistors.

rlogiacco:
"how come those cubes seem so popular and don't burn out the shift registers?"

Dumb luck.

6

That's an answer 8)

For sake of measuring the shift register life expectancy in such configuration I'll run a test: it should be funny to blow one of them up ]:smiley:

7

Won't be that dramatic. Things will just stop working. You may notice some parts are overly warm.

Short version: Nobody ever reads datasheets.

Just the engineers here.