I've killed 3 micro's and I don't know why

[Peeps]

I've put together a fan controller/temp monitor/hardware monitor that drives a 2x40 LCD and displays data collected from the PC over serial such as temperatures, CPU and GPU usage, network activity etc for this custom PC case. I'm using a Pro Micro which is an Arduino Leonardo based breakout board.

My only problem is I've killed 3 Pro Micros and I have no idea why. The first two died during prototyping but the 3rd just died while I plugged a fan into one of the headers while it was running.

Here's a schematic of the Pro Micro PCB:


I removed fuse F1 to disconnect it from USB power since the USB tends to provide power even when the PC is off and it causes issues. Also I'm applying 12V to RAW which runs it through the regulator and steps it down to 5V. VCC is used as an output to supply a regulated 5V to the thermistor.

My board is done on stripboard and is as follows. The blue traces are the stripboard, red are jumper links. The lower right circuit is a buck converter used for driving fans that don't have a discrete PWM input. 




I had just set up a 25khz PWM signal from pin 5, which connects to the PWM2_FAN header. Everything was running and I plugged the fan in and the micro blinked a couple times, the display corrupted and it died. The fan is a small 70x70mm fan, draws 300mA @ 12V. The ATX spec states for PC fans, they want a 25khz PWM signal at 5.25V max 5mA max. Pin 5 measures around 10k to ground like the other pins, so I don't believe I shorted it.

The regulartor on the Pro Micro is now putting 1.8V instead of 5V. I'm going to remove it and see if its damaged or if its being dragged down by the micro.

I'm thinking this is a problem due to the inductance of the fan and my stripboard with its lack of ground plane but I'm not really sure how that all works.


Edit: Just checked and the regulator on the Pro Micro board died but the micro is still okay. I had problems with the regulators on the other ones dying but in those cases the microcontroller IC was also being damaged. Its a 5 pin SMD labeled S8QF which turns up not much data. I'm thinking its getting some kind of voltage spike perhaps? Maybe I should put like a 15V zener on the RAW pin to GND?


Also here's what it looks like if anyone's curious. The case is a modified Leightronix Pro-16 video event controller I had no use for. The micro communicates with a program called Open Hardware Monitor which I modified to send various sensor info to the micro to be displayed. It's my first project involving a microcontroller so everything is new to me and it took a whole bunch of hours to get to this point. From driving the display, to getting the serial data right, to modifying the Open Hardware program has taken months but its pretty satisfying at the same time.

[StillTrying]

A floating GND somewhere ?

[Buriedcode]

I'm not sure what you're doing with that MOSFET and inductor in the lower right of your schem, as its set up as a buck converter, but the output isn't connected to any kind of feedback for regulation.  Its quite a large MOSFET with a large gate charge, and a 3.3 MILIhenry? inductor?  Along with a low current 1N4148 diode which has a max forward current of 200mA and is silicon rather than schottky, totally unsuited for that task.  I won't even comment on the thin traces and layout (SMPS's even low current ones require shorter current paths/loop areas).

Also, how are you powering this?  If its directly from the 12V from the PC then whilst noisy, this line shouldn't go above 15V at any point.  If its the output form that buck converter, then judging by your schem who knows what voltage that could reach.

The fans might generate spikes on their power lines (I don't know what fans you're using) as motors are inductive devices.

As with all circuits, break it down into modules, testing each one individually. What are you trying to achieve with that circuit in the bottom right

edit: typos

[ez24]

   Watching because interesting 

[picandmix]

Hi,

Most folk say the internal pull up for the Reset line is too weak and that can cause all sorts of problems if any noise etc is encountered. Adding a 10k or 4k7 pull up plus a 100nf to 0v are better.

Never a good idea to connect/ disconnect output  loads to a hot circuit.

[Peeps]

I'm not sure what you're doing with that MOSFET and inductor in the lower right of your schem, as its set up as a buck converter, but the output isn't connected to any kind of feedback for regulation.  Its quite a large MOSFET with a large gate charge, and a 3.3 MILIhenry? inductor?  Along with a low current 1N4148 diode which has a max forward current of 200mA and is silicon rather than schottky, totally unsuited for that task.  I won't even comment on the thin traces and layout (SMPS's even low current ones require shorter current paths/loop areas).

Also, how are you powering this?  If its directly from the 12V from the PC then whilst noisy, this line shouldn't go above 15V at any point.  If its the output form that buck converter, then judging by your schem who knows what voltage that could reach.

The fans might generate spikes on their power lines (I don't know what fans you're using) as motors are inductive devices.

As with all circuits, break it down into modules, testing each one individually. What are you trying to achieve with that circuit in the bottom right

edit: typos

Oh that diode is mislabeled. Its a 1N5408. I'm feeding a square wave into the fet driver and vary the duty cycle between 50 and 100% to give me 6v to 12v output. Its basically just a low pass filter  but looks like a buck converter or so I've been told. As far as i can tell it works as intended.  I know the stripboard isnt ideal but I don't want to get a PCB made just for this. Also yes it isbeing supplied by the PC power supply 12V rail.

[SaabFAN]

My guess is that you are creating voltage-spikes on the 12V-Line on top of the 12V-Spikes already coming from the PC, which in turn kill the 5V-Regulator on the board. Combine that with the 5V-Regulator being the cheapest part the manufacturer of the Pro Mini could find and you've found the culprit
In addition to that, there are the motors of the fans, which are adding their own spikes when the coils are being switched.

[coromonadalix]

Maybe putting an line driver ic in the lcd line are you sure it has standard 14 or 16 pins (lcd) pinout  ??

Why 12vdc on the RAW pin ??   doesnt seems right to me ??  second schematic

[Buriedcode]

Oh that diode is mislabeled. Its a 1N5408. I'm feeding a square wave into the fet driver and vary the duty cycle between 50 and 100% to give me 6v to 12v output. Its basically just a low pass filter  but looks like a buck converter or so I've been told. As far as i can tell it works as intended.  I know the stripboard isnt ideal but I don't want to get a PCB made just for this. Also yes it isbeing supplied by the PC power supply 12V rail.

Ok, that makes sense although that is still technically a buck converter Just one without feedback.  But if you've tested it with various duty cycles and it does indeed supply ~6 to~12V then I'll shut up.  I was more wondering if you still have problems with your Arduino if that part of the circuit is removed or inactive.   Although some don't have access to a scope, looking at the 12V line on that board whilst its running will show spikes/dips/noise.

Stripboard is probably fine for this application, as long as that pseudo-buck isn't running at high frequency (>40kHz ish) but depending on the currents involved you might need to use two more more traces because stripboard traces are always thinner than one thinks.

I would add decoupling caps across the power lines of every single fan output, and directy across the "raw" input of the Arduino.  I have destroyed the regulators on these as well from over voltage - it can be rather easy to do with some 12V supplies.  A 15V zener - as you suggested - won't hurt, and would also add extra reverse polarity protection (which isn't really needed as the pro micro has a series diode anyway).

Although you may have mentioned it, I cannot find it - what 12v supply are you using?.  If its from the PC then this has limits on max/min voltage, but can also supply >15A.  This means any shorts or mistakes can send components into the sky.  If it is from a separate 12V supply, then double check this line for noise or spikes.  It might not take much for other circuits powered by that supply to cause spikes on it.

[davep238]

If you are trying to connect a 12 volt fan directly to a micro's pin, you will fry the micro. The micro can't handle more than 20mA at 5 volts. you stated that the fans need 300mA at 12 volts, so you will need a driver transistor (FET) for each fan! BTW, what are you trying to do with the buck converter (U2 + FET + coil)?

[Mr. Scram]

How are the regulators on the boards with the dead micros doing?

[Rasz]

Why 12vdc on the RAW pin ??   doesnt seems right to me ??  second schematic

Voltage on any Pin except RESET and VBUS
with respect to Ground( .............................-0.5V to VCC+0.5V

no wonder his chips are dying

[thm_w]

If you are trying to connect a 12 volt fan directly to a micro's pin, you will fry the micro. The micro can't handle more than 20mA at 5 volts. you stated that the fans need 300mA at 12 volts, so you will need a driver transistor (FET) for each fan! BTW, what are you trying to do with the buck converter (U2 + FET + coil)?

Take a look at the schematic, the micro is not driving any fans directly.. its clearly coming from Vout and ground, which has been explained in the thread.
The micro drives the PWM pin, which is internally pulled up in the fan to either 3.3 or 5V. So that part is OK.

[Peeps]

Why 12vdc on the RAW pin ??   doesnt seems right to me ??  second schematic

Voltage on any Pin except RESET and VBUS
with respect to Ground( .............................-0.5V to VCC+0.5V

no wonder his chips are dying

If you look at the Pro Mico schematic, you can see that RAW is a direct feed into the voltage regulator and I'm using the VCC pin only as an output to be used for a thermistor reference.

How are the regulators on the boards with the dead micros doing?

I lost one...I desoldered it and managed it fell off my desk and bounced into the abyss. I'll check the other one though.

[Peeps]

Oh that diode is mislabeled. Its a 1N5408. I'm feeding a square wave into the fet driver and vary the duty cycle between 50 and 100% to give me 6v to 12v output. Its basically just a low pass filter  but looks like a buck converter or so I've been told. As far as i can tell it works as intended.  I know the stripboard isnt ideal but I don't want to get a PCB made just for this. Also yes it isbeing supplied by the PC power supply 12V rail.

Ok, that makes sense although that is still technically a buck converter Just one without feedback.  But if you've tested it with various duty cycles and it does indeed supply ~6 to~12V then I'll shut up.  I was more wondering if you still have problems with your Arduino if that part of the circuit is removed or inactive.   Although some don't have access to a scope, looking at the 12V line on that board whilst its running will show spikes/dips/noise.

Stripboard is probably fine for this application, as long as that pseudo-buck isn't running at high frequency (>40kHz ish) but depending on the currents involved you might need to use two more more traces because stripboard traces are always thinner than one thinks.

I would add decoupling caps across the power lines of every single fan output, and directy across the "raw" input of the Arduino.  I have destroyed the regulators on these as well from over voltage - it can be rather easy to do with some 12V supplies.  A 15V zener - as you suggested - won't hurt, and would also add extra reverse polarity protection (which isn't really needed as the pro micro has a series diode anyway).

Although you may have mentioned it, I cannot find it - what 12v supply are you using?.  If its from the PC then this has limits on max/min voltage, but can also supply >15A.  This means any shorts or mistakes can send components into the sky.  If it is from a separate 12V supply, then double check this line for noise or spikes.  It might not take much for other circuits powered by that supply to cause spikes on it.

Yep it's the PC's 12V supply. Under testing the PC power supply is only powering this board though. It's not connected to any other PC components. The PWM frequency for the buck converter thing is 25khz.

I had a decoupling cap, 0.1uF on the Micro's RAW pin when I killed the regulator, but none across any of the fan outputs. I'll add those in here.