0
\$\begingroup\$

I have built a device that consists of 10 PCBs connected by a backplane PCB. Each PCB contains various logic gates and a varying number of status LEDs for debugging purposes.

Never before in my life have I constructed any circuits even remotely this complex. However, by connecting some switches to the device inputs and some LEDs (and drivers) to the device outputs, today I have been able to confirm that, after weeks of designing and building, the device appears to be doing exactly what I designed it to do. Yay! :-D

However, something weird is happening: If I make more LEDs turn on, they all get darker! 8-O

It's as if the device is somehow running out of power. But I can't figure out how that can be... Here's what I do know so far:

  • The power supply is definitely still in constant voltage mode. It sees the current increase, and continues to supply the same voltage.
  • If I turn on more LEDs on a particular PCB, all the PCBs get darker, not just the affected one.
  • Various LEDs outside the device do not get darker, despite being part of the same circuit and being powered by exactly the same power supply.
  • The logic functions appear to be being performed correctly, in spite of the brightness weirdness. Most particularly, the digital outputs must be being driven at approximately the right levels.
  • Certain input combinations also affect the brightness even if no LEDs turn on or off. (I presume certain inputs make the gates draw more current, and this is just a different manefestation of the same issue.)
  • When I tested the PCBs individually, the same effect was just barely detectable. But with all 10 PCBs, the effect is pretty drastic.

I am most confused as to how two LEDs connected to the same power rails driven by the same power supply can have such drastically different brightnesses. Is my PCB enclosure possessed, perhaps?


A full schematic would be huge, but here is a rough sketch:

schematic

simulate this circuit – Schematic created using CircuitLab

Hopefully this gives a flavour of what I've built. (I've only drawin 3 PCBs rather than all 10, but in reality there are 10 PCBs connected to the backplane.)

D1 through D5 seem to illuminate normally, however D6 to D14 get dimmer as more of them illuminate. If I turn on, say, D6 + D7 + D8, then they get dimmer, and so does (say) D11 if it's on. But D1 seems unaffected. It's as if there's a current bottleneck somewhere...

\$\endgroup\$
  • 5
    \$\begingroup\$ I'd suspect voltage drops in the power supply wiring - check the voltage on the boards with the dim LEDs. \$\endgroup\$ – Peter Bennett Sep 9 '18 at 17:33
  • 3
    \$\begingroup\$ What voltage is at the beginning and the end of those wires? \$\endgroup\$ – Chupacabras Sep 9 '18 at 19:43
  • 4
    \$\begingroup\$ Measure that! Don't guess. \$\endgroup\$ – Chupacabras Sep 9 '18 at 19:44
  • 3
    \$\begingroup\$ Measure voltages at different points between power source and the diodes. That is the way you'll find where is the voltage drop. \$\endgroup\$ – Chupacabras Sep 9 '18 at 20:18
  • 3
    \$\begingroup\$ So, you have the power going through a breadboard - you didn't mention that before, or show it in the drawing. You obviously have a poor connection in the breadboard, or possibly in the cable to the backplane. Start at one end and measure voltage wherever you can to determine where the problem is. Note that the problem may be only in the ground lead, only in the positive lead, or in both. \$\endgroup\$ – Peter Bennett Sep 9 '18 at 20:59
1
\$\begingroup\$

In my confusion as to why the highly complex PCB assembly wasn't working, I almost completely forgot about the little breadboard in the corner hooking all the power lines together. I mean, it's only connecting a few wires together, right? There's not even any components on it, it's just tying some wires together.

On closer inspection, the part having the problem was connected to the furthest point on the breadboard away from the power supply. So I moved the wires to be right next to the incomming power, and the problem seems to have disappeared. Still not sure why exactly, but it seems to fix my issue.

\$\endgroup\$
  • 1
    \$\begingroup\$ Breadboards don't make very reliable power distribution centers - you might try a barrier terminal block instead. \$\endgroup\$ – Peter Bennett Sep 10 '18 at 21:14

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.