2x Teensy 4.1 3v3 is shorted to GND

[synaptics]

I purchased a couple of Teensy 4.1's several months back for a project that ended up delayed; lacking an unused FTDI adapter I used the pair as 3v3 UARTs for a pair of routers using the following code:

unsigned long baud = 19200;
const int reset_pin = 4;
const int led_pin = 13;  // 11=Teensy 2.0, 6=Teensy 1.0, 16=Benito
const int led_on = HIGH;
const int led_off = LOW;

void setup()
{
  pinMode(led_pin, OUTPUT);
  digitalWrite(led_pin, led_off);
  digitalWrite(reset_pin, HIGH);
  pinMode(reset_pin, OUTPUT);
  Serial.begin(baud); // USB, communication to PC or Mac
  Serial1.begin(baud); // UART, communication to teensy
}

long led_on_time=0;

void loop()
{
  unsigned char c, dtr;
  static unsigned char prev_dtr = 0;

  if (Serial.available()) {
    c = Serial.read();
    Serial1.write(c);
    digitalWrite(led_pin, led_on);
    led_on_time = millis();
    return;
  }
  if (Serial1.available()) {
    c = Serial1.read();
    Serial.write(c);
    digitalWrite(led_pin, led_on);
    led_on_time = millis();
    return;
  }
  dtr = Serial.dtr();
  if (dtr && !prev_dtr) {
    digitalWrite(reset_pin, LOW);
    delayMicroseconds(250);
    digitalWrite(reset_pin, HIGH);
  }
  prev_dtr = dtr;
  if (millis() - led_on_time > 3) {
    digitalWrite(led_pin, led_off);
  }
  if (Serial.baud() != baud) {
    baud = Serial.baud();
    Serial1.begin(baud);
  }
}

I only hooked up the ground and RX/TX pairs and it worked fine for several days. I went to manage one of them a few days ago and one of the Teensy 4.1's was dead. It would not button reset to blinky, and the programming light never even flashed when the button was held down. What appears to be the LDO (u4) was too hot to touch (though the CPU never even gets warm), and it appears the 3v3 is shorted to ground (0.06 volts between 3v3 and GND when powered over USB, 26ish ohms between 3v3 and ground when disconnected and left probing for several second). There are no solder bridges or metal flakes anywhere because I never soldered any of the pins, I just wrapped thin wire through the holes and ran some heat shrink and then hot glue over them to hold it all in place well enough for UART work (was hoping to keep them unsoldered so I could use them for the previously mentioned project). I checked the second Teensy and determined that one was dead too, same 3v3 to GND issue. I have verified the routers were at 3v3 signaling level, and I have never connected any voltage sources other than a USB cable to the Teensy (e.g. no 3v3 after power off from the router to damage it). I checked the forums here for similar issues and it seems that the LDO has been the issue in the past, but I have no idea how it might have rx'd backcurrent as I never connected 3v3 or 5v, unless it was somehow being driven by the voltage on the UART pins when disconnected from USB (though I thought all the pullups/downs were disconnected when powered off), and the UART is coming from the SOC on the router which had no issues, so I don't know how many amps that could realistically deliver before dying itself if it was trying to drive the Teensy.
I'm happy to send the units back in if it would help, but I'm guessing that's not going to be very useful in diagnosing the cause of the failures. I'm also assuming I'm now out $63, as it doesn't make much business sense to send replacements for this kind of hardware.
If anyone has any ideas on how to get the units back up and running (or any idea what could have caused this) that would be amazing, thanks.

[PaulS]

The LDO (U4) is apparently in current-limiting mode and most likely not dead. So your Teensy 4.1's might still be OK.
My suggestion would be to remove the hot glue, the heat shrink and the thin wire you added to the board. And then closely inspect the board for irregularities and check if it's still shorted. My hunch is that one the thin wires is causing an (almost invisible) short.

Paul

[synaptics]

Sorry, I should have clarified. I had confirmed the wires were not shorted after disconnecting them from the router, and I also fully removed the wires as part of my examination of the Teensy's. The wires were solid (not stranded) and came out cleanly when I untwisted them with no flakes from the PTH plating or similar, so I'm confident no broken strands are hiding under a BGA chip or similar. The Teensy's were doing the things as described above with the wires removed and only plugged into a USB cable (with all four wires present/not a "charging only" cable).

[PaulS]

Hmm, pity. Could there have been voltage spikes on the RX/TX lines?
Since the boards seem to be damaged already, you could try one more thing [provided you have the equipment]: externally power the Teensy on the 3v3 pin using a current-limiting powersupply, feed around 3.3/26 =~125mA and check with a thermal camera [e.g. FLIR] what component or part of the board gets warm.

Paul

[rcarr]

Leaving unpowered Teensys's connected to powered routers could be detrimental in two ways. Serial lines idle high, so the Teensy was likely powered via the RX I/O pin when not connected to your laptop. Then powering up the Teensy with voltage already present on an I/O pin could result in CMOS latchup. I do not know if the Teensy is sensitive to CMOS latchup which is why I said could be in the first sentence. But the correct way to power everything up would be to connect the laptop to the Teensy and only then connect the serial cable to the router.

[defragster]

Paul@PJRC has noted a few times the pins should NOT be held high when the Teensy is off.

Though, during Hardware Beta for T_4.0 two units were set side by side with all 7 UART Serial connected: GND and 14 Rx&Tx pins each between. UART Serial definition is that when transfer stops, each of the 14 pins on both Teensy units are held high at 3.3V. Tx gives full light on LED and the Rx gets less current and results in a dimmer LED.

This was a software test for the UART - running them all feeding data through them all in sequence at speed and verifying the transfer integrity.

In the process at some time either one Teensy or the other will always be out of sync being powered on or off while the other is holding all 14 lines high. Especially if one Teensy is recoded to another task and perhaps powered down, while the other stays powered running the Serial test with active lines.

Indeed, in Beta this caused a problem {It was partly (vam)powered as if an RTC battery was attached, and it held the On/Off button state} and required Paul to add a white wire to resolve while an altered PCB was redone. Since then both T_4.0 and T_4.1 tolerate this situation without any part of the Teensy being powered up.

So, two pins if held at only 3.3V should be tolerated. In fact the Beta problem took 3 UART pairs (or more) IIRC to cause the issue.

[sicco]

Was the 'Beta testing' with the old or the new LDO as hardware that was tested?
(ref "Teensy 4.1 manufactured after March 2022 has TLV75733P (U4) replaced by NCV8186AMN330TAG " at the bottom of https://www.pjrc.com/store/teensy41.html)

[PaulStoffregen]

I believe the beta test Defragster mentioned was the period from December 2018 to August 2019, before Teensy 4.0 was released.

The first ~6 months were before NXP released the RT1062 chip, so we had hardware built with RT1052 chips to get started.

Teensy 4.1 came in April 2020.

All the Teensy 4.0 and 4.1 boards made before early 2022 had the TLV75733P chip. It used to be easy to buy that chip, and pretty much all chips. When the chip shortages were on the horizon in early 2021, Robin bought up all the TLV75733P chips we could. They lasted until early 2022, but eventually ran out. As far as I know, it's still impossible to buy any substantial quantity, other than scalpers who want more than the price of a Teensy board just for that chip!

[PaulStoffregen]

As for what went wrong that killed your hardware, I don't know and really can't know from only this info.

I can say the thing we've seen over and over that instantly kills a Teensy 4 is accidentally touching 5V power to the 3.3V power line. Even a slight accidental touch of a wire for the briefest moment means instant death for the main processor.

With all boards, a simple thing you can do that greatly improves odds of survival from mishaps is adding resistors in series with the signals. 1K is a good starting point... low enough for many high speed signals to work, but high enough to limit current into the pin to a (hopefully) safe amount. Difficult to say if this would have helped, but it's something to perhaps keep in mind for the future.

[defragster]

I believe the beta test Defragster mentioned was the period from December 2018 to August 2019, before Teensy 4.0 was released.
...

Indeed by 'Beta' I was talking about the Pre-Release hardware beta.

Paul - I boxed two T_4.1's back to PJRC - not sure what they might show if you have looked at them - been a while delivered 11/20/2021 - short note/post on 'test' forum - just added a note.