1
\$\begingroup\$

I’m using a 74HC595 chip (shift register) NPX semiconductors, connected to and arduino board. Basically the shift register switches on and off 8 leds in sequence.

Everything seem to be working, code and pin wiring to the chip are fine.

The issue is that sometimes the circuit seems to have some problems and the leds are all frozen. I've noticed that when I move the breadboard in some specific locations or simply pass the hand over some wires all the circuit starts working fine. Clearly I’ve checked all the wiring connections and from that point there is no problem. What I’m suspecting is that there is some kind of EMF noise disturbing the setup.

From what I know only high frequency application can be affected by this problem, but I’m working with just 5V and nothing is going beyond 1MHz.

What can be the possible issues? What are the topics I need to study to prevent this problem? How to prevent it?

The following are links to: My video showing the possible noise problem The schematics I got from fritzing.org

enter image description here

enter image description here

I've added a 100uF capacitor, but nothing solved enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ Sounds like insufficient decoupling. \$\endgroup\$ – Ignacio Vazquez-Abrams Nov 22 '14 at 23:18
  • \$\begingroup\$ Looks like it too, now that I see the diagram. \$\endgroup\$ – Ignacio Vazquez-Abrams Nov 22 '14 at 23:21
  • \$\begingroup\$ @ Ignacio Vazquez-Abrams, So I need some capacitors? \$\endgroup\$ – Luther Nov 22 '14 at 23:23
  • \$\begingroup\$ Put short wires off both Vcc and GND and put a 100nF capacitor at the end of them. \$\endgroup\$ – Ignacio Vazquez-Abrams Nov 22 '14 at 23:33
  • 2
    \$\begingroup\$ Decoupling Capacitors should be as close as possible to the part that needs decoupling and you may need 1uF as well. The resistance at 100uF may to large to help... smaller value resisters tend to have lower ESR(en.wikipedia.org/wiki/Equivalent_series_resistance). Also remember it's not the clock speed that is the problem... it's how fast the edges are... Square waves can emit very high frequency signals at every transition so a 1hz signal could in theory emit 100Mhz pulses at every transition. \$\endgroup\$ – Spoon Nov 23 '14 at 0:04
4
\$\begingroup\$

This is either a decoupling problem, or an unconnected wire. As the commenters have said, you'll need to add some capacitors. Fit them as physically close as possible to the 74HC595's power pins (your diagram shows it too far away). Try at least a 0.1uF ceramic capacitor as well as the 100uF electrolytic. Try as much as a 1uF ceramic.

The other possibility is an unconnected pin. CMOS parts can be affected by electric fields when there are unconnected input pins. Have you connected all the unused input pins to either Vcc or Ground? Are you sure that the breadboard wires really are making contact? Solderless breadboards are notorious for unreliable contacts.

\$\endgroup\$
  • 1
    \$\begingroup\$ I agree that it's probably a bad breadboard connection, causing poor grounding and this glitch. As you and the others say, a few 100nF and 1->10uF caps placed appropriately (close to the IC) should fix it all up \$\endgroup\$ – KyranF Nov 23 '14 at 1:30
  • 1
    \$\begingroup\$ I wonder if it could be the notorious "breadboard power rails have gaps in them" problem? Some breadboards need short wire links in the centre of the power rails to connect the two sections. But looking at the still above, it doesn't seem likely. \$\endgroup\$ – John Honniball Nov 23 '14 at 10:52
  • \$\begingroup\$ @John Honniball placing an inductor or a capacitor solved the problem, the breadboard connections were ok. \$\endgroup\$ – Luther Dec 10 '14 at 0:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.