Teensy 4.1 keeps dying

[davidus]

Hi everyone,

We are running into a recurring issue where our Teensy 4.1 boards are suffering fatal hardware damage after running for a period. We have destroyed 4-5 boards under the exact same conditions, so this is not a one-off defect.

Background & Setup:

Power: 5VDC supplied via VIN pin through a 24VDC to 5VDC buck converter.

Function: Running firmware that generates square wave triggers via several PWM pins.

Connections: The PWM pins are hooked up directly to an oscilloscope. USB is also connected. Teensy got damaged with VIN/USB power bridge cut as well as intact. It doesn't seem to matter if we cut it or not

I/O: There is no external voltage applied to any pins. We have a few input pins pulled up to 3.3V using 1k resistors, which we have verified and measured multiple times. There is no external input attached to those input pins.

Environment: Home lab on a rubber mat powered from 24VDC to 5VDC buck converter

The Problem:
Everything looks great on the scope for the first few hours. However, after running continuously for about 30 hours, the Teensy dies. It gets physically damaged at the chip level - it becomes completely unresponsive and will not start again after a power cycle.

Troubleshooting so far:
There is no apparent overvoltage or wiring issue that we can see with the setup (image of the setup attached).

Any help is greatly appreciated.

 

[clinker8]

Were all the units connected to the same buck power supply? If so, suspect the power supply. Have you inspected the power supply with a scope before and after the failure?

For what it's worth, you didn't show your setup, only a partial schematic. If you can please post a picture of the setup. Also, what is the part number of the power supply are you using. What's it's overshoot spec? Is the power ever cycled? Did it fail after a cycle, or while continuously running? Does it recover if you press the reboot button?

Hard to make further suggestions without more details.

 

[skpang]

What 24VDC to 5VDC buck converter are you using ? Could be transient from the converter.

Post a photo of your setup.

 

[davidus]

Thank you for your response. This has happened both on a buck converter like this one (LM65645RZTR) as well as from a 5VDC "wall adapter" which is plugged into the micro-USB port. It doesnt seem to matter how I power it.

It fails after continuously running. Basically, you would set it up, let it run, come back ~30 hours later, you already see something is wrong based on the scope not showing the square wave. You try to power cycle the teensy, it no longer comes back. After this point no matter what you trye, the teensy is dead.

 

[kd5rxt-mark]

@davidus: from the picture in your original post, it appears that you are providing 3.3VDC from an external power source (the schematic seems to indicate that the 3.3VDC pin on the Teensy is not connected). Where is the 3.3VDC supply coming from ?? if/since you seem to have multiple power sources, do you have all grounds (Teensy, 5VDC supply, 3.3VDC supply, etc.) connected at a common point ?? What else is on your PCB that may be either connected to your Teensy and/or leaking unintended power into your Teensy by some sneaky path ?? And, just a reminder that the T4.x is not 5VDC tolerant . . . all inputs must not exceed 3.3VDC (looking at all of your T_xx connections). Just guessing at this point !!

Mark J Culross
KD5RXT

 

[Sudo]

Maybe this is a noob suggestion, but if you are not using batteries and are ultimately drawing power from the mains, how about putting your setup behind a UPS (uninterruptible power supply) or at least a surge protector for the next test? Maybe your utility is giving you voltage spikes. Over the years I burned a few computer screens because of that, until I started using a UPS, which totally eliminated the problem.

 

[davidus]

@davidus: from the picture in your original post, it appears that you are providing 3.3VDC from an external power source (the schematic seems to indicate that the 3.3VDC pin on the Teensy is not connected). Where is the 3.3VDC supply coming from ?? if/since you seem to have multiple power sources, do you have all grounds (Teensy, 5VDC supply, 3.3VDC supply, etc.) connected at a common point ?? What else is on your PCB that may be either connected to your Teensy and/or leaking unintended power into your Teensy by some sneaky path ?? And, just a reminder that the T4.x is not 5VDC tolerant . . . all inputs must not exceed 3.3VDC (looking at all of your T_xx connections). Just guessing at this point !!

Mark J Culross
KD5RXT

Hi Mark,

All grounds are common. The 3V3 comes from an LDO which gets power from the 24VDC to 5VDC buck converter which powers the Teensy.

There are two things on the PCB - ethernet socket for the teensy, and then RS422 transmitter which takes the PWM signal from a the teensy pins and turns it into differential signal. This does work as confirmed by a scope, but again, it magically kills the teensy after several hours.

All input pins are disconnected, but are pulled high to 3.3V rail. Output pins generate square waves and put that into RS422 transmitter. That is it. It's a very bare setup.

Thank you, AB3Y

 

[kd5rxt-mark]

@davidus: Teensy is not really made to be fed 3.3VDC from an external supply, and particularly when external 5VDC is supplied as well. It's possible that the on-board 3.3VDC regulator is burning out. Here's a <link> with the detailed info on the power-up sequence required by the IMXRT1062 microcontroller (scroll down to the "Power Up Sequence" section & click on the numbered buttons to see each step in action), followed by the "Power Design Notes" section with more info that may be useful for your board design. Here's another <link> that describes how to possibly "get away with" supplying 3.3VDC externally. Does your board really need 3.3VDC to be externally supplied, or would the 250mA capability of the on-board 3.3VDC regulator be enough for whatever else you're powering with the external 3.3VDC supply ?? Can you post a complete schematic ?? This may allow someone else to see something that could be a cause of the failure(s).

Hope that helps . . .

Mark J Culross
KD5RXT

 

[Fluxanode]

IMHO - In my implementations I power the teensy with a 5v volt onboard switching supply IC circuit. The teensy uses it's built in 3v3 reg only for itself. For the other 3v3 components i have a separate 3v3 regulator. That way I never have to worry about interfering with the teensy's 3v3 power,

 

[davidus]

@kd5rxt-mark , I only power the teensy via 5VDC. All the 3v3 pins on the teensy are floating; nothing is applied to them.

As @Fluxanode said, the 3V3 rail is for other ICs (namely the RS422 transmitters).

Schematic is attached.

 

[skpang]

Have you measured the output voltage on the 5VDC wall adapter ? I've seen some are high as 5.2V

Would really like to see a complete schematic of the board. Can't see the 3.3v regulator at the moment.

If the 24v input is not on and you have the 5v USB connector then you are back feeding the 5v to the LM65645RZTR. Normally you would have a diode such as the BAT54 connected from the LM65645RZTR to the 5v line to prevent back feeding.

 

[davidus]

@skpang, full schematic is attached one post up. The wall adapter was good, 5VDC. We tried several of them as well. We nuked maybe 5 teensies with different setup running the PWM out code - some going through wall adapter, some even powered from Raspberry Pi's USB port, some hooked up to 24VDC->5VDC buck converter. It is a mystery to me.

You can cut a trace on the Teensy to prevent back feeding. Both, cut an uncut, resulted in the Teensy dying over some period of time.

 

[jmarsh]

That's a lot of low resistance pull-ups. The total sinking current may be exceeding the Teensy's limits.

 

[Fluxanode]

Can you do a current check on what the teensy is pulling from 5v?