2
\$\begingroup\$

as part of an open-source design I'm working on, I decided to include a "heartbeat" LED instead of just a power LED. The idea is that it powers on as soon as the main SBC is powered (it's an Orange Pi Zero 3), and then I can also control it through an i/o pin, which I use to reflect the CPU load as a "heartbeat" (beating faster or slower).

To accomplish this, I'm using an AND gate, tying one input to the SBC's debug UART Tx pin (which is set high around 1 second after power is applied), and the other pin to a regular i/o pin. Both pins have an external pull (up in the case of the regular pin, and down in the case of the Tx pin), both of them 10k ohms. What I liked about this approach is that if I do a software "shutdown" of the SBC, the Tx pin goes low, effectively turning the LED off to signal that the whole gadget can be powered off safely.

Here's the schematic, which at least to me, looks like nothing out of the ordinary.

Schematic

I've already blown 3 of these AND gates, part number is SN74LVC1G08DBVR-EV, from EVVO semi.

It's powered by the SBC itself, through its provided 3.3VDC rail. Because of how simple this is, I can't think of anything other than the parts being defective.

  1. The first chip worked just fine for a couple of weeks. It was powered by the 5V rail, which should not be a problem, as the outputs of the SBC are 3.3V. However, when it failed, it also blew up the SBC (I guess the gate shorted its inputs and applied the 5VDC to the SBC pins?)
  2. Then I replaced the chip (still using 5V) and even without inserting the next SBC, it blew up right away on power up. I really have no explanation for this...
  3. Then I switched to the SBC's 3.3V rail. This resulted in the next AND gate working fine for a couple hours, and then blowing up. This time, the SBC was unaffected, though.

What am I doing wrong? Should I just try the same gate from a more reputable manufacturer such as TI?

\$\endgroup\$
6
  • \$\begingroup\$ Both DEBUG_TX and HEARTBEAT is 3.3 V? \$\endgroup\$ Commented Sep 9, 2024 at 13:18
  • \$\begingroup\$ Yes, they're both from the SBC, which has it's own internal 3.3V regulator. All pins from the SBC are 3.3V. \$\endgroup\$ Commented Sep 9, 2024 at 14:25
  • \$\begingroup\$ How are DEBUG_TX and HEARTBEAT connected to your gate: are they direct on a PCB or coming in over a cable? If you're plugging/unplugging a debug cable that connects directly to your gate pins that's a potential ESD event that can destroy anything it touches. \$\endgroup\$ Commented Sep 9, 2024 at 18:18
  • \$\begingroup\$ It's PCB traces, I don't touch them or use the scope on them either. \$\endgroup\$ Commented Sep 10, 2024 at 15:06
  • \$\begingroup\$ One think to note here, is that you cannot power this chip with 5 V and use 3V3 logic. The VIH is 0.7 * VDD, which means 3.5 V. Using 3V3 logic means that the inputs are either low or in a metastate, which can destroy the chip. \$\endgroup\$ Commented Oct 21, 2024 at 13:48

1 Answer 1

Reset to default
3
\$\begingroup\$

My guess is that their input protection diodes get shorted to either rail due to an overload.

Try adding a 10kΩ series protection resistor on DEBUG_TX and HEARTBEAT lines, directly at the gate's input - after the pull up/pull down. It should not be necessary, but then this "should" be working, so it'd be cheap insurance.

Should I just try the same gate from a more reputable manufacturer such as TI?

Yes.

But: If you got the EVVO gates from a mainstream distributor like Digi-Key, then presumably they are genuine. And, also presumably, EVVO is using a decent fab and follow good manufacturing practices. It'd be rather hard to make unreliable gates if they stay within mainstream industry practices. IF.

If you got the gates from Amazon or eBay or the likes, they may be from the discard bin at the packaging line, or may not be EVVO at all... Maybe you could get someone to decap them and compare the die to real EVVO gates.

  1. [...] it blew up right away on power up

Replace the blown gate from your exiting "dubious" stock. Then attach a a 4V Zener right across the supply pins of the gate. Apply and remove the power from the board many times. If the gate survives that, then you could have had an overvoltage on the power rail. You should be able to see that overvoltage even with the Zener inserted. Connect an oscilloscope probe between 0V and +3.3V and reconnect the power, using a single trigger.

\$\endgroup\$
2
  • \$\begingroup\$ Got the parts from DigiKey, so they must be genuine. I'll do the 4V zener to be sure, but the current 3.3V are supplied by the SBC itself, through its own internal regulator. I'll also add a 5.1V zener to the 5V SMPS, which I did based on aoz1284PI and is kind of tricky, maybe that could be the cause. However, all other chips in this board (pcf8523, pcf8574, adc128d818, and the SBC itself) show no signs of any voltage spikes or anything like that. What I'll also do is leave my scope running with a trigger to see if I see any voltage spikes. Thank you! \$\endgroup\$ Commented Sep 9, 2024 at 14:29
  • \$\begingroup\$ @Diego If the parts are from DigiKey, then they may be just overly sensitive parts by design. I'd switch to TI/ST. If that will work fine, then perhaps TI/ST parts are less sensitive to whatever is causing the EVVO part to fail. So the EVVO part may be "OK" by some measure, just not robust enough. The damage you describe, with parts blowing up on initial power-up (among other times), sounds like a hard latch-up. With series resistors, latchup will not be likely. If EVVO will still blow-up with series resistors and a Zener clamp on supplies, then I'd not use EVVO any more. \$\endgroup\$ Commented Sep 9, 2024 at 19:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.