Resurrecting busted Teensy 4.1 (or understanding the issue)

Unfortunately, my new Teensy 4.1 broke because of a short circuit from 3.3V to GND. I connected a HX711 load cell amplifier the wrong way around, so VCC was connected to GND and GND to 3.3V (those damn DuPont connectors!). Other than that, there were two GPIOs connected to the HX711 data lines and the whole board was powered exclusively from the 5V pin. This bad connection caused the HX711 chip to short itself, so now if I measure resistance between VCC and GND I get ~10Ω. A busted HX711 is not a big deal, but this also forced the 3.3V Teensy output to ground. This event only lasted 5-10 seconds, because I noticed the over-current protection on my 5V power supply had tripped and the Teensy was not blinking. Nothing was hot to the touch and there was no magic smoke. I removed the HX711 and connected the Teensy to the PC, but no luck. It was dead.

Today I bought a reflow station and removed the NCV8186AMN330TAG 3.3V regulator from the Teensy 4.1, because that's what is likely to have broken. I connected the Teensy to the PC with a USB cable to give it 5V, then I powered the 3.3V line with a regulated power supply (in this order to comply with the power-up sequence suggested here). Sadly, the board didn't power up again. It consumes about 200mA at 3.3V, the MIMXRT1062 chip is fairly warm and the computer detects nothing. I ordered a new NCV8186AMN330TAG from Mouser but my hope is mostly lost for this poor Teensy (unless someone knows what else to do).

Now, however, I would like to understand why the Teensy was killed by a simple short like that, so hopefully it won't happen again - and next time I'll make sure to use JST-XH instead of DuPont. The Teensy 4.1 has a 500mA polyfuse and the voltage regulator is rated for 1A. I don't think the polyfuse protected the board at all because the 5V power supply's current limit was set to 1A, and it was maxing out since the overcurrent/short indicator was on. This is weird, but I have no other explanation. The MIMXRT1062 built-in voltage regulation shouldn't be broken because it's completely independent from the 3.3V line. A short on that line shouldn't even break the main or bootloader MCUs, they just wouldn't power up. Same for the ethernet and USB host chips. Despite all this, the Teensy seems lost forever. If you have any insight on what might have broken, please let me know. I'd like to turn this event into a learning experience for the future.
 
Shorting 5V to GND, or shorting 3.3V to GND, certainly isn't good, but odds are slim those will destroy Teensy 4.1.

However, shorting 5V to 3.3V, even for milliseconds while a connector partially mates, will instantly destroy a Teensy. Any chance that's what might have happened?
 
No that's impossible, I tested the Teensy connected to the 5V power supply without the HX711 and it was working fine. As soon as I connected the bad HX711, it stopped working...
 
This bad connection caused the HX711 chip to short itself, so now if I measure resistance between VCC and GND I get ~10Ω.
If you measure resistance between GND and the I/O pins on the HX711, do you get similar continuity?

If you do, then that GND, having been connected to 5V supply, pulled those Teensy I/O pins to 5V, at the same time when 3.3V was forced to ground. Ouch. Could, perhaps, explain why the Teensy went too.

(But I'm just a hobbyist myself and know nothing.)
 
There are a few hundred kΩ between the data and power pins of the HX711, so those are ok. The load cell amplifier was powered by 3.3V, so even if they were shorted out it would have been "fine"
 
Please keep in mind you're asking us for opinions about what might have gone wrong, but we can't see details. I don't even know what a HX711 is. Tried Google just now and landed on this Sparkfun page. I don't see any connectors, DuPont or otherwise, not to mention how they're wired or what would happen if mated backwards or partially if used for hot plugging.

It's not clear to me if this extra hardware had its own power source or if it was being powered by Teensy.

Usually it's difficult to post-mortem diagnose the root cause of what really caused hardware to fail, even when the dead hardware is sitting on your own test bench where you can see and touch it and perform tests. Over the internet, without any photos, clear diagrams, or even links to the specific hardware used, I hope you can understand how this is really just blind guesswork.
 
Don't worry Paul, I was indeed just asking for an opinion, especially on the fact that the polyfuse didn't protect the Teensy and if shorting the 3.3V line breaks only the regulator. I have no idea either on how the HX711 chip works, but just for reference here's the one I used and its wiring. The pin headers were provided in the bag and I soldered them to the pads. I ordered a new Teensy 4.1 and hopefully I'll be able to get everything back running ASAP. Thanks

10-kg-Cella-di-Carico-Sensore-di-Peso-HX711-Digitale-di-Pesatura-Sensori-Ad-Modulo-per.jpg
a9cb97189c4449f195179d94f703be3e.png

Load Cell Amplifier HX711 Breakout Hookup Guide
 
I can't say what went so wrong, but I can answer this...

and if shorting the 3.3V line breaks only the regulator.

Shorting the 3.3V pin to GND does not break the regulator or cause permanent damage to Teensy 4.1.

I've tested this many times. I tried again just now on a Teensy 4.1 with the NCV8186A regulator and a lab bench power supply (no USB power) set to 5 volts and 1.5 amp current limit connected to VIN & GND. The power supply's overcurrent LED does come on for a brief moment, but then the steady state current settles to about 210 mA. While the NCV8186A regulator is capable of 1 amp, it also has "foldback" current limiting where current is reduced to a much lower amount after overcurrent protection is triggered. I kept 3.3V shorted to GND for several seconds, and I repeated the test about a dozen times. That Teensy 4.1 is still working. Every time I removed the clip lead causing the short, it rebooted and run fine (at least as observed by the LED blinking).

I also tried just now with setting the power supply current limit higher. Even at 10 amp limit (the highest my BK Precision 1747 power supply can go) the Teensy 4.1 is not damaged. I repeated the test many times in rapid repeats, just in case the initial surge of current might cause a problem before the NCV8186A foldback limit kicks in. With the power supply set at 10 amps max, the red overcurrent LED on the power supply does not blink when the touch the clip lead to 3.3V & GND. I did more experiments with the current limit setting and rapidly touching the 3.3V & GND clip lead, to see where the power supply overcurrent LED no longer blinks. At 1.6 amps, it sometimes blinks, sometimes doesn't, and by 1.7 amps I can't get it to blink at all. Maybe I should try connecting a shunt resistor and high bandwidth amplifier to my scope, so I can get a better idea of the actual current, as my power supply's overcurrent LED seems to be a pretty imprecise indication. But I really don't have time for that right now...

I can tell you I've just now shorted this Teensy 4.1's 3.3V & GND pin about a hundred times while fiddling with current current limits, with many of those with the limit set at its 10 amp max, and a few of the tests sustaining the short for over 10 seconds. This poor little Teensy 4.1 is still working fine after all that 3.3V - GND short abuse.

Whatever killed your Teensy 4.1, I can say with pretty good confidence it wasn't a simple a matter of shorting 3.3V to GND.
 
Wow, that's impressive! This definitely means that the HX711 chip has done something very nasty, perhaps through the data lines even if they still have ~300kΩ of resistance to VCC and GND. Fine, rest in peace my Teensy. Thanks for your time Paul
 
Resistance measurements on chips can be tricky to interpret. Almost everything interesting inside a chip is junctions of N & P type semiconductor, not really resistors. Your multimeter forces a test current through the leads and measures the resulting voltage, using ohms law to show you a number if the thing connected between the leads is a real resistor. But with a chip, that test current is passing through some unknown (to people who didn't design the chip) combinations of P-N junctions, so your meter see a voltage that depends on the current in a very non-linear way that isn't anything like a real resistor.

About the HX711, I saw on a Sparkfun guide it can work with power between 2.7 to 5.5 volts. If so, I'd recommend powering it from 3.3V. Teensy 4.1 pins are not 5V tolerant. Running the 5 volt power over connectors and cable is a lot riskier than using the 3.3V power for everything.
 
Yes, I'm powering everything from 3.3V. It'll be a big board with two Teensy 4.1 and a lot of sensors and actuators in a rover for an European university competition. For once I can say I'll be using Teensy to the max!
 
Back
Top