Why does 74hc595 work "better" without GRND (pin 8) grounded?

i followed shiftout tutorial in learning area figure out how serially control led matrix. ended in small pitfall game when attached joystick. fun i've had in while.

but when had played device while, noticed hade made omission should mean nothing lit @ all.



i had forgotten attach second ground row ground! still worked quite expected. decided bad idea keep ics ungrounded, dutily connected second ground row connected one.

what happened matrix flickering point of "nothing working". debugged , added delays , whatnot, , seemed work, not when things got fast. re-read article , noticed capacitor on clock pin , added wee bit big 1 (10uf) circuit. still behaving weird. dug out small 1 (22pf) , things started good.



so what's happening here? why apparently act more stable when ics aren't grounded, , why work @ all? dying in there when not grounded?

i'm quite curious why thought idea juble pins on led matrix instead of going 0-7 vcc , 8-15 v0.

here's video of final product if you're interested.
https://twitter.com/bleedo/status/931929808002535424

quote

i re-read article , noticed capacitor on clock pin

that's error in tutorial. shows 1uf cap on latch pin (not clock pin said). it's infamous error, nobody seems have ability correct it. there should not cap on latch pin. damage chip. there should, however, bypass cap each chip. these should 0.1uf, connected close vcc , ground pins of each chip.

quote

i had forgotten attach second ground row ground!

it took me while understand meant that. have written more carefully. meant ground rail on right side of breadboard. should not have worked without ground connection. dumb luck.

where led series resistors? not shown in diagram. value using? values need chosen carefully, avoid overloading output pins , overall maximum current 595 chip can handle.

Arduino Forum > Using Arduino > LEDs and Multiplexing > Why does 74hc595 work "better" without GRND (pin 8) grounded?

arduino