Custom STM32 board bootup problem - Fried MCU? Help!

[stefanskos9]

Posted on April 17, 2013 at 10:11

Hi All,

I have made a custom board with the STM32F405. When I apply power to the board, the MCU is drawing lots of power (like ~200mA+!). I haven't loaded any code onto it, its straight out of the box. I have 2.2uF caps on VCAP_1 and VCAP_2 and they read 3.3V (Vdd rail). I think this is incorrect and tells me the chip is possibly blown? Possibly from ESD during fitment (baked reflow oven). I hope this is an rare and freak occurrence, is there any way of testing if its dead? MCU gets no response from ST-Link SWD, with IAR sometimes returning ''No MCU Device found'', or ''Failed to load flash loader''. Some other potentially useful information for diagnosing: This is my first STM32 board, I mainly copied the schematic from the F4Discovery. I have BOOT0 and BOOT1 on jumpers to change between GND and Vdd. Have tried all boot configurations with these pins. I am using PC13/14/15 for IO. PC14&15 drive LEDs, PC13 drives a PNP and is tied high through a 47kohm resistor and base resistor - is this okay? PH0/1 are not connected NRST is tied to 5V with 47k resistor and 100nF to gnd+tactile switch. All Vdd are decoupled <1cm Vref uses an external reference chip + decoupling Vbatt is connected via jumper to Vdd + decoupling Some of the pins are outputting weird voltages. For instance, with Vdd=3.3, Pin51 (PB12) is outputting 2.5V. I think its dead... Any thoughts? Thanks for any help in advance!!

[Tesla DeLorean]

Posted on April 17, 2013 at 13:41

Sounds like you've got in orientated wrong.

[jj2]

Posted on April 17, 2013 at 16:36

Misplaced chip certainly possible.  If not - have you double/triple checked that indeed all VDD and GND pins are routed correctly - and that solder joints on each/every one are good? 

Too late now - but our group always enforces strict current limiting when powering up new board during early test/verification.  Proper - and consistently applied ESD precautions are ''real'' - must be followed. There do exist relatively inexpensive SMD Proto Assembly shops - more costly than toaster-oven - but serves as a superior means to ''test/verify'' your basic board design.  Once known good - you can then migrate to lesser means - with far better odds of success...

[stefanskos9]

Posted on April 18, 2013 at 04:12

Thanks for the replies guys.

Chip is definitely not misplaced, also double checked pin order is correct and not reversed. Diode/beep setting on multimeter indicates short between Vdd and Vcap, I think this indicative something is wrong with the MCU? The solder is not shorted, and the situation is identical for both Vcap pins. There is also a short between the two Vcap pins, not sure if this is a problem. Have also checked on an unpopulated board that Vcap and Vdd traces are not shorted. Solder joints for power and GND seem okay. I check from a Vdd/GND testpoint to the top of MCU pin (not at the bottom where solder/pad is). SWD pins also seem fine. Another strange indicator is an LED with cathode to PE3 lights up, so the pin is obviously sinking current. I think the MCU should be shipped without any code loaded? Therefore all pins should be disabled by default? Also, I do not have an external OSC. Any further thoughts? I will be trying to desolder the MCU today and check that there is nothing unusual about the board without it. I have ordered some more MCU's with express shipping. Thanks again for your help guys

[stefanskos9]

Posted on April 18, 2013 at 04:25

Just found another interesting thing.

With diode setting on multimeter, positive lead on anode of an LED, negative lead on ANY pin on the MCU (expect the pin the anode is connected to through a resistor), the diode lights up!! Dead dead dead ?

[jj2]

Posted on April 18, 2013 at 05:59

Your meter may be improper for your testing.  (i.e. in diode/beep mode - may produce both excessive voltage and employ no (or minimal) current limiting.)

When the MCU is not under bias (i.e. unpowered) you should not be applying voltage to non VDD pins.  Junctions may be ''back biased'' - large, damaging currents may flow.  (and we've just read how you expanded such treatment - beyond PE3.) Is there any chance that you attempted to program this chip?  Most GPIO should default into inputs - and if PE3 sinks current (when the MCU is properly powered) that is unusual.  (unless PE3 defaults into some non-standard mode) Have you measured the output from your 3V3 regulator?  A spike upon turn-on may wreak havoc on any/all chips in receipt - and likely requires a scope to confirm...

[stefanskos9]

Posted on April 18, 2013 at 06:25

Thanks for your continued help jj.sprague!

Your meter may be improper for your testing.  (i.e. in diode/beep mode - may produce both excessive voltage and employ no (or minimal) current limiting.) Possibly. It is just a cheap multimeter, however connecting straight over an LED causes it to glow quite dimly. I would say its not putting out much voltage, or is current limited - im not sureWhen the MCU is not under bias (i.e. unpowered) you should not be applying voltage to non VDD pins.  Junctions may be ''back biased'' - large, damaging currents may flow.  (and we've just read how you expanded such treatment - beyond PE3.) So your saying the circuit of having the cathode to PE3 is a bad move? Its actually a circuit copied from the F4Discovery board, for the USB power chip fault pin. Except I'm using the 5V rail as the power/pullup source - this is the case with a few pins depending on where power was routed (either 5V or 3V3). If this really is a concern - as the 3v3 rail gets it power from the 5V rail, perhaps there is a delay in power reaching Vdd before a 5V pullup is applied? i.e 5V is applied to non-Vdd pin before MCU is powered - would only be u/mS though. Also FWIW, I only used the diode/beep to check for shorts after I had noticed there was a problem. Perhaps the LED is getting a return path through protection/suppression diodes in the MCU?Is there any chance that you attempted to program this chip? I have tired, but wasn't able to. If any of the code did manage to load, I have not enabled PE3, nor initiated the GPIOE clock.Have you measured the output from your 3V3 regulator?  A spike upon turn-on may wreak havoc on any/all chips in receipt - and likely requires a scope to confirm... Have checked this on scope, there is no spike. I have also used this regulator before to power Vdd directly on the discovery with no problems. Again, thank-you for you ongoing help 🙂

[Tesla DeLorean]

Posted on April 18, 2013 at 06:45

A partial schematic may be helpful, redact as required.

Reported current does seem excessive. VCAP pins should be 1.2V, bonded together wouldn't surprise me. VSSA/VDDA connected to supplies? Used by POR/BOR circuits, VBAT switch. PDR_ON ? PC13,14,15 are somewhat problematic as they are in the BKP/PWR domain, low drive, pull ups here may feed through the ESD diodes. FT (5V tolerant) pins, might float an internal supply ring, not sure the construction there. For non-FT pins higher voltages will feed into the 3V/3V3 ring.

[stefanskos9]

Posted on April 18, 2013 at 07:52

A partial schematic may be helpful, redact as required.

Happy to provide, though the schematic is pretty massive and across many pages. Are there any pins in particular of interest? In the MCU schematic doc all pins are just connected to ports on other pages - apart from power.Reported current does seem excessive. VCAP pins should be 1.2V, bonded together wouldn't surprise me. So the fact the are both shorted to Vdd is indicative of a problem? Perhaps it doesn't mean the chip is blown if the problem is rectifiable - perhaps an internal rail is pulled to Vdd causing the Vcap = Vdd?VSSA/VDDA connected to supplies? Used by POR/BOR circuits, VBAT switch. PDR_ON ? Vdda is decoupled and connected to Vdd within the analog AGND moated fill. Vssa is connected to analog ground (moated AGND fill). Vbat is on a jumper to Vdd with 1uF decoupling - I have tried with a shorting jumper, and without - which is required? Could perhaps monitoring Vbat current give an indication of health? PDR_ON is permanently connected to GNDPC13,14,15 are somewhat problematic as they are in the BKP/PWR domain, low drive, pull ups here may feed through the ESD diodes. FT (5V tolerant) pins, might float an internal supply ring, not sure the construction there. For non-FT pins higher voltages will feed into the 3V/3V3 ring. Should I try disconnecting the 3V3 pullup on PC13? I found it hard to find information on which pins were 5V tolerant. It was my understanding all the of the standard IO pins were? Pins which have 5V pull-up are: PA15 PC3 (input of buffer amp is pulled-up), PC7 PE0, PE1, PE3, PE5, PE6, PE12 NRST (can be disabled with jumper) I have connected the NRST jumper - hope it can be pulled to 5V :o Also, probing around, I found the exposed bits of metal out of the side of the package on the corners are grounded. I didn't even know they were there. Thanks 🙂

[Tesla DeLorean]

Posted on April 18, 2013 at 08:05

I found it hard to find information on which pins were 5V tolerant. It was my understanding all the of the standard IO pins were?

It's in the data sheet in the pin list.

http://www.st.com/st-web-ui/static/active/en/resource/technical/document/datasheet/DM00037051.pdf