T4 3.3V rail short to GND

Status
Not open for further replies.

guzu

Well-known member
Hello everyone.

I have a Teensy 4.0 that I have been using for a hobby project. All worked fine until yesterday when I wanted to power the board with the usb cable and nothing happened.

After some digging, I found out thet the 3.3V rail of the Teensy board is shorted to ground. (0 Ohm tesistance) And the board pulls 400 mA from the usb.

I want to ask if you know of some components that would fail and make the short? Did you notice during testing that some capacitors failed sooner than other parts? (For example) I would like it not to be the chip. :D
Or do you know a way to find out what part failed?

I would like to repair the board just for the fun of it. I have 2 more boards on the way from the shop.

Thank you very much for your input.
 
Since there is no point in opening a new thread I will just write an update in this one.

So after burning 3 x Teensy 4.0 I would really like to know what am I doing wrong. Yesterday the 3rd Teensy 4.0 just got the same problem described in the first post. It is a short between the 3V3 and GND rails. When I power up the Teensy with 5V on the VIN pin, the multimeter shows only ~3.0V on the VIN pin. After a few seconds the Teensy starts to “hiss” and the voltage goes up to ~4.2V on the VIN pin that is connected directly to the 5V input.

I get the 5V input from an 11.1V battery pack (3 x 3.7V 18650 cells in series) that goes through a LDO (AMS1117-5.0). Yesterday when the Teensy burned out the voltage in the powerpack was only ~6.5V. Since my 2nd Teensy got fried I made it so that the Teensy does not power any other peripherals via the 3V3 pin (I thought that the first one burned because of the current draw of the connected peripherals). Now the Teensy was powered from the 5V LDO, the peripherals were connected to another 3.3V LDO, and the Teensy only got input or send output on the sensors etc (not powering anything). I know now that the current draw was not the problem because the Teensy burned out exactly at startup all 3 times and everything else works fine + even if I connected the peripherals, the measured current draw does not go over the 250mA max stated for the Teensy 4.0 board.

Things I observed:
1. The current draw is not the problem (the Teensy does not power anything, they have separate power sources).
2. The boards did not show signs of burning out (overheating, “hissing” etc.). They would just not turn on and then the 3V3 short to ground would be present.
3. The LDO seems to give constant 5V (maybe max 5.03V)
4. Now (after burn out) the MKL02Z32VFG4 chip gets hot if I let the board powered for ~10 sec.
5. Now, when powering the board (the burned out ones) with 5V on the VIN pin, the multimeter shows ~3.0V for about 4 seconds and ~4.2V after on the VIN pin (the voltage changes when the board starts “hissing”).
6. I think it is a power source problem but cannot find the problem.
7. The problem (not working and "hissing") persists even if I power the burned boards from the 3V3 pin.

I attached a schematic and the PCB layout I am using. (The top copper layer is GND and the bottom copper layer is 3V3.) Could anyone point me to what I am doing wrong? I want to upgrade to the Teensy 4.1 (need a few more input pins) but I really do not want to burn any more boards. I have another device that works just fine but the Teensy 4.0 there is powered through the 3V3 pin using a 3.7V LiPo source through a 3.3V LOD (SP6205EM5-L-3.3/TR). You can see both the boards in the link below.

Here is a link where you can see more about my project.

SCHEMATIC Sch.png

LAYOUT Layout1.jpg

PCB TOP Top.png

BOTTOM Bottom.png

Could I replace something on the burned boards so that they work again?

I know a lot of you are now busy with the Teensy 4.1 launch but I could really use your help here. I can provide any other information you would need regarding my setup.

Thank you very much.
 
Last edited:
As a place to start, see https://www.pjrc.com/teensy/schematic.html
In particular the layout at the top left corner where there is the 500ma poly fuse and the regulator.
This is not a hard cutoff device but should warm up as it approaches the current limit and constrain the draw. So 400mA is an odd result since it should not be doing much at that level, and the connected regulator should also be clamping above that level. Though a possible case may be the regulator cycling and and out trying and failing to drive the short circuit and the average output is 400mA.

The most likely casualty is the main microproccessor unfortunately, either because it somehow got more than 3.3V on the power in or because one of the input pins got above 3.3V. given there is a regulator upstream and a 5V switching power supply and one case of the defect happened with discharged batteries blowing the regulator due going above 5.5V in is unlikely. Have looked at your circuit and not seeing anything that would drive a pin above 3.3V. A test if you can remove the teensy from one and see if any of the pina on the PCB are above 3.3V.

Another avenue to test is the 3.3V supply on the board - drawing too much 3.3V from a teensy4 just causes the onboard reg to brown out and/or fail so does not normally harm the main microprocessor, it just stops working because the regulator is undervoltage, but if you have a seperate 3.3V supply and it can get higher than that for any reason than any related Teensy pins will be in overvoltage condition, which is a known way to blow the chip, though my personal experience has only resulted in open rather than short circuits.

If you are curious you could try trying to heat gun the main CPU off one of the dead boards to see if the regulator kicks back in but it does not help overmuch since replacing the MIMX is not going to be cost effective or straightforward.
 
A test if you can remove the teensy from one and see if any of the pina on the PCB are above 3.3V.
...
If you are curious you could try trying to heat gun the main CPU off one of the dead boards to see if the regulator kicks back in but it does not help overmuch since replacing the MIMX is not going to be cost effective or straightforward.

Thank you very much GremlinWrangler. I will try your suggestions and will post back the results.
 
So I did a little testing today to see what is shorted on my burned Teensy 4s.
It turns out that is indeed the MIMX chip :( Once I desoldered the chip, the short was gone. So no hope of repairing them.

I also checked the voltage on the other pins on the PCB that go into the Teensy to see the voltage. Only SCK and MOSI have 3.34V. Is that too much?

Also, I noticed that I have voltage on the 3V3 pin of the Teensy (with the Teensy unplugged from the PCB) if the Teensy Audio Shield is plugged into the slots on the PCB. When I unplug it, the voltage on the 3V3 pin is 0V. Could that be a problem? Where could the 3.3V come from on that pin from the Audio Shield?

Thank you for your input.
 
I have 9 pull up resistors (4.7k to 3.3V) and 3 pull down resistors (4.7k to GND) on the PCB. After analyzing the Audio Adapter (Rev C) schematic I noticed that the SCL, SDA, and the 2 CS pins (memory and SD card) also have pull up resistors mounted on the Audio Adapter. Could the fact that I have also mounted on my PCB 4.7k resistors to all of those 4 pins be a problem? Or could the total number of pull up resistors be a problem that causes the Teensy to burn?
Also, could it be a problem that when I power the Teensy, I also power all the other components on PCB at the same time? (The Teensy is powered from the same LDO that powers the other components on the board.)
I want to mention that every time the board stopped working, it did not happen immediately after I put a new Teensy on the PCB. The boards worked just fine for a few weeks (2-3 days of usage/week).

Any help would be appreciated. Thanks.
 
Last edited:
The pull up resistors should be fine as long as they are to 3.3V or close to it, having too many of them can cause strange I2C behavior but that is not the current problem unless you have somehow put a .3ohm resistor in there. I would not suspect it of nuking an entire teensy but have had trouble in the last mixing pins up and trying to drive an input as an output. for something like a button this menas all is fine until you output is being driven high AND the buttons is pressed and suddenly the pin is short circuited. So a double check of your pin functions may be wise, but do not believe this is your problem since I'd have expected a dead pin and 'random' resets to be the observed problem rather than a completly dead microcontroller.

So things go back to the power supply, and the input pins. If I'm understanding things you are running the Teensy and all connected hardware off a 3.3V supply? If this supply is not stable under varying load this may have bearing on things, will need to review your current schemtic but as a way ahead I'd suggest:
Run hardware off your 3.3V supply
Run Teensy off 5V and onboard reg, isolate Teensy 3.3V from system 3.3V
Look at adding 1k resistors in series with inputs that may have question marks on the value (so the battery voltage sense for example) which limits the potential fault current.

None of this is pinning down the why here though, just trying to avoid it being a problem again.


Re the I2C pullups, there are some advantages in making sure the total resistance to +ve is the target range so you can probably remove the pullups if it makes sense to do so and leave the ones on the audio board to do the job. Main hassle comes when you have half a dozen I2C devices all with pullups and things get problematic.
 
Having re-read the schematic, is there any possibility at all that that 5V in the IDE connector can get shorted into a Teensy signal by cables flexing or moving etc, or can it come into contact with anything else with power at the far end? Come to that, how long is the cable? 10s to 100s of meter cables can be problems in many ways without actually touching anything else just by acting as antennas in various ways.
 
for something like a button this means all is fine until you output is being driven high AND the buttons is pressed and suddenly the pin is short circuited.
I don't think this is the case because the fault appeared exactly when the board was powering up. No button was pressed. Still I will double check the functions.

So things go back to the power supply, and the input pins. If I'm understanding things you are running the Teensy and all connected hardware off a 3.3V supply? If this supply is not stable under varying load this may have bearing on things, will need to review your current schemtic but as a way ahead I'd suggest:
Run hardware off your 3.3V supply
Run Teensy off 5V and onboard reg, isolate Teensy 3.3V from system 3.3V
Look at adding 1k resistors in series with inputs that may have question marks on the value (so the battery voltage sense for example) which limits the potential fault current.
This is exactly what I have done except for the 1k resistor on the input lines. The 3rd time the Teensy was burned, it was not powering anything else. The battery was connected to the 5V LDO and the output of the LDO was connected to the VIN pin. Just like in the schematic. Still I will look at adding the 1k resistor to the questionable inputs.


Having re-read the schematic, is there any possibility at all that that 5V in the IDE connector can get shorted into a Teensy signal by cables flexing or moving etc, or can it come into contact with anything else with power at the far end? Come to that, how long is the cable? 10s to 100s of meter cables can be problems in many ways without actually touching anything else just by acting as antennas in various ways.
The cable connected to the IDE is a CAT5e internet cable, ~40cm long.
You can see how the cable is connected here, at around 0:21.
At the far end the 5V is connected directly in a internet socket.
I will triple check again to see if the 5V could have shorted on the IDE connector or on the far end of the CAT5e cable. But if the 5V shorted to some other pin, shouldn't other components also fail?

Could the problem be the 3.3V returning somehow to the 3V3 pin of the Teensy from the Audio Adapter? Interacting in some way I do not understand with the power supply? Why is the Adapter behaving like this? If I unplug the Audio Adapter, there is no voltage on the 3V3 pin. When I plug the Audio Adapter back in, 3.3V is present on the 3V3 pin of the Teensy.

Thank you very much for the information. I really appreciate you taking the time to help me.
 
The 3.3V from the audio adapter is odd, since it does not have a 3.3V regulator of it's own. For linear regulators it is normally ok to have them in parallel but given all the signs point to a stray voltage somewhere it bears investigation in case you have something achieving resonance or something equally exciting and going out of tolerance (3.7 V from https://www.pjrc.com/teensy/IMXRT1060CEC_rev0_1.pdf)
 
The 3.3V from the audio adapter is odd, since it does not have a 3.3V regulator of it's own.
The 3.3V from the Audio Adapter I think comes because the Audio Adapter also has a pull up resistor for the MEM CS pin and I also have a pull up resistor for that pin (this resistor is the only component still connected to Teensy 3V3 pin because of pcb constraints - will change the traces in the next version).
Voltage path: 3.3V from my LDO > 3.3V Audio Adapter pin > Audio Adapter pull up resistor for the MEM CS > MEM CS trace to my Teensy > MEM CS pull up resistor on my PCB > Teensy 3.3V rail.

Could this have burned the Teensy? Some sort of voltage loop?
 
Is there any other voltage regulator on the other boards?

Hi Frank. Yes there is one more LDO on the other PCB connected to the PCB in the first post.
Here is the schematic for it.
Steering PCB.jpg

I know it is a stupid design, but here is how the power goes from one PCB to the other through the ON/OFF switch (S1):
PCB1 battery+ > 9V-IN on the RJ2 (Steering PCB) > 9V-IN S1 Switch (Steering PCB) > 9V-OUT S1 Switch (Steering PCB) > 9V IN (IDE Connector on PCB1) > AMS1117-5.0 LDO on PCB 1 > Now the 5V from the LDO go to the Steering PCB RJ2 +5V > +5V S1 Switch > SP6205EM5-L-3.3/TR LDO > components on the Steering PCB (work at 3.3V).

Could the problem be some inductance from the Battery+ wire and the +5V wire going through the same CAT5e internet cable with 4 other signal wires the to the Steering PCB?

On the new PCB for the Teensy 4.1 the battery+ will go to the Steering PCB and it will be regulated to 3.3V directly.
 
Last edited:
Is there any other voltage regulator on the other boards?

As I mentioned in the post above, there is one more LDO on a board connected to the main PCB. Do you think this has something to do with why the Teensys getting fried?

Thanks.
 
Sorry to write in such an old thread but I just wanted to say that I think the problem was that in the IDC connector where all the sensor wires go, there were 3.3V and 5V wires. When working with the PCB somehow the 5V wires must have touched the other pins that are not 5V tolerant (either inside the IDC connector or the ends of the wires touched each other). I think this is why my Teensy boards got fried. Now I have a redesigned PCB (with the 5V power supply separated) and everything works well.
Just in case someone else encounters this problem. :)

Thank you all for the great input.
 
Status
Not open for further replies.
Back
Top