0
\$\begingroup\$

I have an "interesting" circuit failure. The circuit basically looks like

schematic

simulate this circuit – Schematic created using CircuitLab

The left branch works perfectly fine. The right branch turns on and initially works, but after a few seconds flickers unpredictably and then eventually goes out. When this happens, R2 drops most of the voltage and D3 and D4 basically fail short. Turning off the circuit, waiting a bit and trying again produces the same behaviour.

In its failure mode, D3 and D4 always fail together, and D1 and D2 always keep working.

Aside from the LEDs not working, this is a bad thing to happen because U1 is a port driver on a PIC that is only rated to drive up to 100mA. For details on this port see the PIC16F1773 datasheet sections 11.3.8, register 11-17, and electrical limits in 36.1.

When disconnected from the circuit, D3 and D4 present the exact same voltage on a diode tester that new diodes do - a little north of 1.8V, and produce the same dim light that good diodes do when tested.

Why could D3 and D4 be failing-short at the exact same time, and why is it unpredictable? Is it due to some kind of thermal runaway? These LEDs are supposed to be able to handle 50mA and had not been driven up to that current - until of course they fail short, at which point they see more.

Edit: Mystery solved.

\$\endgroup\$
5
  • 3
    \$\begingroup\$ If over-current is somehow to blame, then one somehow going to a low forward voltage would immediately increase the current taking out the other too. But it's mistaken to say that they are supposed to be able to handle 50 mA - that's an absolute maximum rating, not an operating condition - the actual continuous recommendation is 30 mA or less or talk to the apps engineers. \$\endgroup\$ Commented Dec 26, 2020 at 21:58
  • \$\begingroup\$ If you are using a single port pin to sink 100 mA, then I want to see the exact PIC datasheet and where, exactly, you see that figure. (I hope it's not an absolute maximum specification.) I'd also like to see the absolute maximum for the port, as well. But I suspect your LEDs on that side are already damaged (from either an earlier event or something that happened while testing things out) and that they no longer function as they should. Try replacing them. If that still exhibits the same problem then look at the wiring. \$\endgroup\$
    – jonk
    Commented Dec 26, 2020 at 22:20
  • \$\begingroup\$ @jonk added; HIDRV was pretty surprising to me, too. \$\endgroup\$
    – Reinderien
    Commented Dec 26, 2020 at 22:25
  • 1
    \$\begingroup\$ @Reinderien It is amazing. Since you only are using one of these, not both, that's also good. I see that the absolute maximum specification for the entire device is 250 mA! Do keep that in mind, as well. If you decide to use both high current pins you will already be too close to the absolute max for my comfort, for example. \$\endgroup\$
    – jonk
    Commented Dec 26, 2020 at 22:32
  • \$\begingroup\$ @jonk look at the wiring turns out to have been the key. The problem was a marginal solder bridge that would short or open depending on board heat. This is probably the best possible outcome, because it diverted current away from the MCU rather than frying it. \$\endgroup\$
    – Reinderien
    Commented Dec 30, 2020 at 12:34

1 Answer 1

1
\$\begingroup\$

If the circuit is really wired as drawn, and if D3 and D4 consistently fail while D1 and D2 do not, then I suspect your resistors. It may be that the resistors are not the same value, and the branch with the lower resistor value will see more current.

Isolate the resistors from the rest of the circuit and measure their values. If you can't do that, measure the voltage across the LEDs when they are illuminated. If the voltages on D3 and D4 are significantly higher then the value of R2 is too small.

\$\endgroup\$
1
  • \$\begingroup\$ Good guesses, but it turns out to have been a marginal solder bridge. \$\endgroup\$
    – Reinderien
    Commented Dec 30, 2020 at 12:33

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