short circuit somewhere killing all my teensies?

Reviewing the board files and the schematic, I think the power supply to the Teensy is fundamentally not correct.
In the schematic, the Vin pin is labeled "+3.3V". That 3V3 comes from a regulator mounted on JP13.
3V3 is too low to power the Teensy. That's perhaps why Teensy's 3V3 regulator output shows 0.12V?
And as @jmarsh stated above: "VIN is meant to be 5V and it should be the only way the Teensy is powered - do not attempt to feed in 3.3v to the Teensy's 3.3v pins". As far as I can tell from the schematic and board files, you don't use the Teensy 3V3 output.

Is it possible for you to remove the regulator from JP13 and then power the Teensy over USB only?

Perhaps superfluous but if you power the Teensy over USB, you will see the USB voltage on the Vin pin as well as on the VUSB pin.

View attachment 33980

Paul
Funny enough, I did try that. Teensy is reading 0.056V between 3.3V and GND when on USB power and not connected to anything.
 
We did try the 3.3V pin to ground as suggested in other posts about problems like these, but we got a reading of 0.12V. This hovers between what was described as a short in metallic connections and a short in the main processor from what I've read, so I have no clue what to make of that.

0.12V is indeed unusual.

Metal shorts conduct very well, so you normally wouldn't see so much voltage unless the metal short is a really extreme situation, like one of the really cheap and extremely thin breadboard wires with longer length, or you have a lot of current flowing.


Teensy is reading 0.056V between 3.3V and GND when on USB power and not connected to anything.

Now that does sound more like a metal short. Normally metal shorts measure very close to zero volts, so seeing something like 56mV really does need to be understood in the context of how much current you believe is likely flowing through the short.

If you can't find and fix the short, at the very least you can label that Teensy as damaged and not waste more of your time on it. Electronics problems are always frustrating, but everything become exponentially more difficult if you end of reusing already damaged parts (let's not talk too much about how I know this....)


Did we just get a bad batch of Teensies? We did order them together.

Unlikely. If it were just 1 bad DOA unit, maybe. But you said they work for 30 minutes then die. Maybe I missed it, but didn't see how many boards we're talking about here? Can you say how many you have and how many have died so far?

As for differences in the hardware that might matter, since release in 2020, there have been 3 different voltage regulators used on Teensy 4.1 due to the 2021-2023 chip shortages. The earliest had Texas Instruments TLV75733P. In late-2022 those ran out and Teensy 4.1 was maybe with OnSemi NCV8186AMN330TAG. In early 2023, those ran out and we switched to TI TLV75533P (which is virtually identical to TLV75733P but shuts off at about 600 mA). Throughout 2023 as previously ordered parts were delivered very late, a mix of the OnSemi and TI parts were used. By late 2023, all Teensy 4.1 were made with the TI TLV75533P part.

We did have some reports that the OnSemi is not as robust as the TI chips. The problems have been reported as a combination of adding a large amount of extra capacitance to the 3.3V power line and also using a 5V power source which acts as a low impedance to GND when switched off. The theory is the charge stored in large capacitors can backflow through the voltage regulator if the 5V power source acts like a short to GND when "off". Whether any of that is relevant to your design, I don't know. I saw you have some capacitors, but they're only small parts. This sort of problem potentially happens when adding a large capacitor like 220uF or 1000uF. Even then, it depends heavily on the characteristics of the power supply as it turns off.

Of course, we've seen Teensy 4.1 killed many other ways. Driving any pin with 12 volts kills it instantly. So does accidentally driving the 3.3V power with 5 volts or more.

Another way to kill Teensy is from "back EMF" of driving motors or solenoids or other large magnetic coils. If current is flowing in the coil, a high voltage spike gets created when the current flow suddently stops. Normally a diode (or combination of diodes) is used in parallel with the coil to give the current a path and prevent the high voltage spike. Sometimes a resistor and capacitor "snubber circuit" is also used, because the diode can take time to react. Those short high voltage spikes are well know to damage Teensy and most other digital chips, though usually it takes time because the pins inside Teensy's main chip do have some limited ESD protection diodes. But those diodes are only meant to handle the relatively minor static discharge that happens when humans handle the electronics. Sometimes they can end up protecting against much more for quite some time, which is good, but also frustrating because failure manifest much later after everything seemed to be running ok.

Whether any of this is useful or applicable to your situation, I really don't know. I still don't even have a clear idea of what other stuff you're connecting. But I want to help, so this info about common problems we've seen before is the best I can do at this point.
 
0.12V is indeed unusual.

Metal shorts conduct very well, so you normally wouldn't see so much voltage unless the metal short is a really extreme situation, like one of the really cheap and extremely thin breadboard wires with longer length, or you have a lot of current flowing.




Now that does sound more like a metal short. Normally metal shorts measure very close to zero volts, so seeing something like 56mV really does need to be understood in the context of how much current you believe is likely flowing through the short.

If you can't find and fix the short, at the very least you can label that Teensy as damaged and not waste more of your time on it. Electronics problems are always frustrating, but everything become exponentially more difficult if you end of reusing already damaged parts (let's not talk too much about how I know this....)




Unlikely. If it were just 1 bad DOA unit, maybe. But you said they work for 30 minutes then die. Maybe I missed it, but didn't see how many boards we're talking about here? Can you say how many you have and how many have died so far?

As for differences in the hardware that might matter, since release in 2020, there have been 3 different voltage regulators used on Teensy 4.1 due to the 2021-2023 chip shortages. The earliest had Texas Instruments TLV75733P. In late-2022 those ran out and Teensy 4.1 was maybe with OnSemi NCV8186AMN330TAG. In early 2023, those ran out and we switched to TI TLV75533P (which is virtually identical to TLV75733P but shuts off at about 600 mA). Throughout 2023 as previously ordered parts were delivered very late, a mix of the OnSemi and TI parts were used. By late 2023, all Teensy 4.1 were made with the TI TLV75533P part.

We did have some reports that the OnSemi is not as robust as the TI chips. The problems have been reported as a combination of adding a large amount of extra capacitance to the 3.3V power line and also using a 5V power source which acts as a low impedance to GND when switched off. The theory is the charge stored in large capacitors can backflow through the voltage regulator if the 5V power source acts like a short to GND when "off". Whether any of that is relevant to your design, I don't know. I saw you have some capacitors, but they're only small parts. This sort of problem potentially happens when adding a large capacitor like 220uF or 1000uF. Even then, it depends heavily on the characteristics of the power supply as it turns off.

Of course, we've seen Teensy 4.1 killed many other ways. Driving any pin with 12 volts kills it instantly. So does accidentally driving the 3.3V power with 5 volts or more.

Another way to kill Teensy is from "back EMF" of driving motors or solenoids or other large magnetic coils. If current is flowing in the coil, a high voltage spike gets created when the current flow suddently stops. Normally a diode (or combination of diodes) is used in parallel with the coil to give the current a path and prevent the high voltage spike. Sometimes a resistor and capacitor "snubber circuit" is also used, because the diode can take time to react. Those short high voltage spikes are well know to damage Teensy and most other digital chips, though usually it takes time because the pins inside Teensy's main chip do have some limited ESD protection diodes. But those diodes are only meant to handle the relatively minor static discharge that happens when humans handle the electronics. Sometimes they can end up protecting against much more for quite some time, which is good, but also frustrating because failure manifest much later after everything seemed to be running ok.

Whether any of this is useful or applicable to your situation, I really don't know. I still don't even have a clear idea of what other stuff you're connecting. But I want to help, so this info about common problems we've seen before is the best I can do at this point.
We had 3 dead boards so far, 1 Teensy 3.5 and 2 Teensy 4.1.

The more we test the less it seems like a back EMF issue. We tested the same motor setup on an Arduino mega 2560 and it ran just fine. Plus, I'm pretty sure there is reverse polarity protection and diodes already on the motor drivers, they are pololu ones and designed for standalone operation after all.

It's also not like we are powering the motors directly from the Teensy, they have a completely seperate power supply.

There are some pretty big caps on the external voltage regulator we are using... I'll look into that and see if that's the issue.

Finally, we tested it with an Arduino (not plugged into the female pin headers that were on the PCB, but instead jumper cabled into the female pins in the shape of a footprint on the PCB). That Arduino worked just fine, so it seems more like a Teensy issue. Another thing we noticed is that once we disconnected the shorting Teensy, it still has that 0.056V reading from 3.3V to GND even though there was nothing connected and it was being powered by a USB cable. The voltage regulator also got pretty hot too.
 
Another thing we noticed is that once we disconnected the shorting Teensy, it still has that 0.056V reading from 3.3V to GND even though there was nothing connected and it was being powered by a USB cable. The voltage regulator also got pretty hot too.
That doesn't show much except that the Teensy has been permanently damaged, the voltage regulator has failed / shorted.
 
0.12V is indeed unusual.

Metal shorts conduct very well, so you normally wouldn't see so much voltage unless the metal short is a really extreme situation, like one of the really cheap and extremely thin breadboard wires with longer length, or you have a lot of current flowing.




Now that does sound more like a metal short. Normally metal shorts measure very close to zero volts, so seeing something like 56mV really does need to be understood in the context of how much current you believe is likely flowing through the short.

If you can't find and fix the short, at the very least you can label that Teensy as damaged and not waste more of your time on it. Electronics problems are always frustrating, but everything become exponentially more difficult if you end of reusing already damaged parts (let's not talk too much about how I know this....)




Unlikely. If it were just 1 bad DOA unit, maybe. But you said they work for 30 minutes then die. Maybe I missed it, but didn't see how many boards we're talking about here? Can you say how many you have and how many have died so far?

As for differences in the hardware that might matter, since release in 2020, there have been 3 different voltage regulators used on Teensy 4.1 due to the 2021-2023 chip shortages. The earliest had Texas Instruments TLV75733P. In late-2022 those ran out and Teensy 4.1 was maybe with OnSemi NCV8186AMN330TAG. In early 2023, those ran out and we switched to TI TLV75533P (which is virtually identical to TLV75733P but shuts off at about 600 mA). Throughout 2023 as previously ordered parts were delivered very late, a mix of the OnSemi and TI parts were used. By late 2023, all Teensy 4.1 were made with the TI TLV75533P part.

We did have some reports that the OnSemi is not as robust as the TI chips. The problems have been reported as a combination of adding a large amount of extra capacitance to the 3.3V power line and also using a 5V power source which acts as a low impedance to GND when switched off. The theory is the charge stored in large capacitors can backflow through the voltage regulator if the 5V power source acts like a short to GND when "off". Whether any of that is relevant to your design, I don't know. I saw you have some capacitors, but they're only small parts. This sort of problem potentially happens when adding a large capacitor like 220uF or 1000uF. Even then, it depends heavily on the characteristics of the power supply as it turns off.

Of course, we've seen Teensy 4.1 killed many other ways. Driving any pin with 12 volts kills it instantly. So does accidentally driving the 3.3V power with 5 volts or more.

Another way to kill Teensy is from "back EMF" of driving motors or solenoids or other large magnetic coils. If current is flowing in the coil, a high voltage spike gets created when the current flow suddently stops. Normally a diode (or combination of diodes) is used in parallel with the coil to give the current a path and prevent the high voltage spike. Sometimes a resistor and capacitor "snubber circuit" is also used, because the diode can take time to react. Those short high voltage spikes are well know to damage Teensy and most other digital chips, though usually it takes time because the pins inside Teensy's main chip do have some limited ESD protection diodes. But those diodes are only meant to handle the relatively minor static discharge that happens when humans handle the electronics. Sometimes they can end up protecting against much more for quite some time, which is good, but also frustrating because failure manifest much later after everything seemed to be running ok.

Whether any of this is useful or applicable to your situation, I really don't know. I still don't even have a clear idea of what other stuff you're connecting. But I want to help, so this info about common problems we've seen before is the best I can do at this point.
So, to close off this series of events, we finally found the fault.

A few things handled it - as suggested, we switched to a 5V power supply. But the most important thing was that we had improperly grounded our motor drivers - apparently the ground for the 12V motor supply was not internally connected to the ground for the operating supply, meaning we had to ground it twice, once for the pinhead and once for the 12V supply (which had its own connector). We did not do the second ground for the pinhead. I suspect that possibly the current backflowed into the teensy through the connections for the motor driver's pwm speed control and digital direction control pins. This was probably exacerbated by the fact that we had a 3.3V supply rather than the proper 3.6V to 5V supply. As for why the teensy's 3.3V to ground still measured 0.056V even after being removed and it was overheating even with nothing else connected, I suspect that the voltage regulator either got fried or something in the chip itself was melted by the heat generated and shorting inside the chip.

Anyways, we swapped it out with another teensy and tested, so far no operating above regular temperatures and everything is running fine and dandy.

And for everyone that helped - thank you all so much, the amount of help was wonderful and I am grateful for the support.
 
Back
Top