Second Teensy 3.5 Project Failure - please help for heartfelt wish

[Dragonfly]

My dear friends,
I don't have any technical background, nor had I programming lessons. Three years ago I started with a few Arduino projects that are still up and running. I was so impressed by it that I wanted to prepare my own PCB that leaves as many future connections open as possible. For this reason I decided last year to go a step further with Teensy 3.5 in a compact format.

After I shorted the first Teensy 3.5 by mixing up the Ethernet connections (as they are up side down) of VCC and GND I corrected my PCB layout and gave it another try (as the pins of the Teensy are not breadboard friendly I could only try out some functions one after another before).

While building the second layout step by step I have had success powering the board up with different smaller sketches - however that didn't last long - for the second time I shorted the Teensy: my multimeter shows that VCC and GND are connected again. This damn thing was successfully running several times with different sketches to test some connected boards as the INA219/W5500 Ethernet Module/sensors and without any further changes it broke?! I've read in another thread that "high capacitors" might cause this problem, but what on earth means "high"? Most of the capacitors on my PCB stabilize the power output of three LM1117s (2x 3V3 + 1x5V for Teensy). The board will be powered by a 24V solar bank (voltage ranges between 24 and 29V), for this reason the DC-DC Buck Converter on the bottom of the PCB reduces the voltage to around 11 volts for those mentioned LM1117s. It also worked successfully with my lab power supply. Most of the connections shown are simple terminal connectors for future purposes:


Top front view with voltmeter on the left, motion sensor and ST7735 Display


Top view with socket for future nRF24L01 module and pressure sensor on the right


Top back view with INA219 current sensor on the bottom


Top view of the main PCB


Bottom view with DC-DC buck converter and W5500 Ethernet Module (with 2x 220uF capacitors for it). None of the enamelled copper wire from the bottom of the Teensy has a short.


Right now I have to decided wherever I want to risk a third board or go back to Arduino which never fried so far. As my knowledge is limited I am willing to pay somebody that can help me with a hardware and software set-up to get at least started in a save way! In this very moment I don't know if I should throw everything away as this task exceeds my capacity, on the other hand this project for my parents is a heartfelt wish for which I already invested quite a lot of patience and time! Of course I can deliver complete EAGLE files or PDFs of the schematic.


Power for the whole PCB


PCB just for reference

What I primarily intend to do with this board is:
1. Measure the water flow of a garden hose as an interrupt over 24 and 48 hours
2. Measure the rain (rain bucket) as an interrupt over 24 and 48 hours
3. Read two numbers of a website with the Ethernet module and show the measurements (of 1. and 2.) on a simple website
4. Show measurements on an 1.8" Adafruit ST7735 Display
5. Save measurements on SD card
Later on I would like to add some temperature sensors and relays.

I will exclaim in delight if someone has an idea why the Teensy friend again or if an expert than can help out. Thank you!

 

[tonton81]

teensy power input should not be more than 6v max

 

[Dragonfly]

Thank you for your respond tonton81. The Teensy gets power from one LM1117 5V linear regulator. It's power was measured right before I added the Teensy.

24V Battery Bank > Fuse > DC-DC Converter to 11V > 2x LM1117 3.3V (one just for the Ethernet Module and the other for several future sensors) and 1x LM1117 5V just for the Teensy.

 

[Darcy]

Have only glanced at your board breifly and I'm a novice with no real world experience, but it looks like you have a lot of large caps on your board and bare wires running to the underside pads of the Teensy? I'd suggest using pogo-pins on your board to tidy-up. Not sure how long those caps remain charged after power removal, but maybe they can store energy long enough to be a potential shorting hazard when working on the board?
Do you need 11v ? That is a lot of power/heat burning up in the 5v and 3v linear regulators, and a lot more board area with 'dangerous' voltage levels. I would POLOLU 036VF6 buck-regulate to 6V and LDO linear regulate down from there as needed.

 

[Dragonfly]

Thank you for your suggestions Darcy.
Pogo Pins are really useful and would be much easier to e.g. replace the board, however I would not have the space on the PCB under the Teensy. As I've never worked with LM1117 before I read lots of information before designing my PCB: I read that caps are necessary to stabilize the output so every voltage regulator has a 100uF capacitor*. Are those classified as "large" already? Secondly, as long as I don't touch the board for let's say two to five minutes after cutting power, they should be discharged enough to work with the board, right?

I could reduce the juice of the DC-DC Buck Converter down to 6 or 7 volts, that is absolutely right and I should do that for not burning so much heat. What annoys me so much is that I broke again a Teensy without any learning effect. If nobody in this forum finds a serious fault in my schematic, is there any way to protect the Teensy or prevent this from happening?


*One LM1117-3V3 (just for the Ethernet Module) has two additional 470uF caps and the 5V regulator just for the Teensy one 470uF.

 

[Dragonfly]

Is there anybody out there with another idea why the Teensy 3.5 broke? So far, if I got it right, the "lage" capacitors were very likely the problem... so my question is: if I replace them with smaller ones (e.g. 100uF), should I give my PCB with a new Teensy another try?
Thanks

 

[PaulStoffregen]

Is there anybody out there with another idea why the Teensy 3.5 broke?

Maybe something other than the power supply is going wrong?

Looking at your PCB layout image, on the right hand side I see a terminal block with a "12V" label next to "A15", and indeed that pin appears to route directly to the A15 pin without any protection circuitry to prevent an accidental short to 12V from immediately destroying your Teensy. Likewise on the other side, I see 3.3V right next to 5V, and 12V power nearby. Maybe you're not using these pins, or maybe you're always being very careful... but this sort of design where human fumble with loose wires going into terminal blocks is very risky. Human error and simple clumsy mistakes are pretty much inevitable. Running bare microcontroller signals or the sensitive 3.3V power to the same terminal block where humans handle 12V power is just asking for trouble.

I see you have a 2N2222 transistor, but I don't see it on the schematic. I didn't manage to follow the traces and reverse engineer what it might be doing. But I will say, a common problem happens when using switching power off to inductive loads, like relay coils or motors. Usually a bare minimum of a clamping diode is needed to prevent a high voltage spike.

 

[Dragonfly]

Thank you for your response PaulStoffregen, I appreciate your help.

...on the right hand side I see a terminal block with a "12V" label next to "A15", and indeed that pin appears to route directly to the A15 pin without any protection circuitry to prevent an accidental short to 12V from immediately destroying your Teensy.

I triple-checked what you described and I would definitely consider that in future PCB designs. As traces of three different voltages are close to A15 as you mentioned it wasn't very clever to do so in the retrospective. I checked with my multimeter and A15 has not connection to GND, 12V, 5V or 3,3V. Moreover I haven't used any POWER terminals so far (on the left side you mentioned) as I first uploaded some smaller sketches, e.g. for the INA219 module mounted on the bottom of my PCB, the PIR sensor and so on to see if every component works independently, ...the processor broke before I was able to merge several tasks in my own sketch. In short, I wasn't even connecting anything to the terminal blocks, but I did measure voltages on several terminal blocks with my multimeter though!

Likewise on the other side, I see 3.3V right next to 5V, and 12V power nearby. Maybe you're not using these pins, or maybe you're always being very careful... but this sort of design where human fumble with loose wires going into terminal blocks is very risky.

Right, good point. I can exclude this fault this time, however fumbling around is very risky and I will consider this seriously.

I see you have a 2N2222 transistor, but I don't see it on the schematic. I didn't manage to follow the traces and reverse engineer what it might be doing. But I will say, a common problem happens when using switching power off to inductive loads, like relay coils or motors. Usually a bare minimum of a clamping diode is needed to prevent a high voltage spike.

I didn't want to show the whole schematic for clarity and as most pins are just connected to terminal blocks let me add the NPN transistor which has a diode. In several projects I use the same configuration that has been reliable so far. And, as I said, I wasn't even connecting external devices.

On the one hand side I am pretty impressed that nobody found a grave mistake that could have been so stupid to be embarrassing, on the other hand this makes the decision hard to give it another try with the risk of destroying a third Teensy. I am absolutely irresolute right now and feel insecure. If we assume that I didn't touch any of the power pins, could the "large capacitors" be the answer as Darcy mentioned? I added several caps to stabilize the power for the Teensy 3V3 output and each of the three LM1117 voltage regulators.