1
\$\begingroup\$

The PCB I am working on somehow went terribly wrong. As soon as I power up the circuit, MCU(MSP430F6720) starts to quickly overheat and in one case literally smoked. I described this in TI's forum here, but in a general sense, what leads this overheat/short of MCU to happen and how can I troubleshoot the issue?

My understanding is: somehow, either by design or manufacturing of PCB, a short circuit was created inside MCU: 1) Some pin is illegally connected to power supply rail (VCC); 2) Some pin is illegally connected to ground; 3) Power supply voltage exceeds the maximum rating of MCU. Did I miss something here?

Funny thing is, the PCB had been working just fine till I changed the power supply circuit from AC adapter + alkaline battery to AC adapter + lithium ion battery. So now I am kind of back to square one. Any input would be greatly appreciated.

Edit: below is the power supply circuit I used in the problematic PCB. Because I only changed this part of circuit from previous revision, I believe the problem is in this part. The circuit consists of the following three parts:

1) LDO: xc6227, this IC is also used in previous revisions. 2) Lithium Ion charging IC: MCP73844. This circuit has been used in my other PCBs. 3) On/Off push button IC: MAX16054. This is used to turn on/off the power supply.

Update: the problem turned out to be feeding voltage of lithium battery into the analog input. The new PCB with this problem fixed works just fine. Thanks goes to everyone.

enter image description here

\$\endgroup\$
  • \$\begingroup\$ If you want any help you are gonna need to post a schematic and/or a layout. If the uC smokes you can toss it, if you let the magic smoke out the chip can't work anymore. \$\endgroup\$ – Vladimir Cravero Apr 2 '16 at 16:05
  • \$\begingroup\$ Thanks for the comment, I just added the power supply circuit in which I believe the problem is. Also on a side note, I already throw away 5 of them after painfully hand solder those little SMDs. What a pain. LOL \$\endgroup\$ – Zhiyong Li Apr 2 '16 at 16:09
  • \$\begingroup\$ That schematic doesn't help much. One IC is unnamed, (you could also usefully link to relevant datasheets), there's no obviously 3V3 rail (is it VCC?) and we don't know what power you're connecting to the MSP or on what pins. (I also recommend measurisg some voltages, especially on the MSP430's power rails). The Li battery can produce 4.2V, I hope you aren't connecting it directly to the MSP. \$\endgroup\$ – Brian Drummond Apr 2 '16 at 16:14
  • \$\begingroup\$ My bad. IC1 is MCP73844, a lithium charging IC. Unfortunately I did connect BATT_P(3.7 ~ 4.2V) to analog input pin on MSP430, but the problem persists after I severed that trace. Also the 3V3 rail is labelled as VCC on the pin3 of IC4. When MCU is removed from board. Reading on 3V3 rail is correct and switching between battery and AC adapter seems work fine. \$\endgroup\$ – Zhiyong Li Apr 2 '16 at 16:20
  • 2
    \$\begingroup\$ The previous version that gave you no trouble would be useful for a difference analysis - post that as well if you can. \$\endgroup\$ – Peter Smith Apr 2 '16 at 16:37
4
\$\begingroup\$

Things that could go wrong:

  • PCB manufacturing error (especially with DIY technology): you should check the integrity of the traces before soldering the components, for example with a DMM in continuity checking mode (the one with the buzzer),

  • PCB soldering error: if it's not factory-soldered (wave or reflow), you should learn to properly solder by hand (using flux); you may check the integrity of the soldering joints (again, with a DMM in continuity checking mode), or you may simply reheat the suspicious points,

  • PCB design error: incorrect IC pinouts (human mistakes, bad data sheets, incompatible replacement products),

  • component orientation error: inserting a DIP IC with the opposite orientation is such a common error that it deserves a separate bullet point - for a 74xx-series IC this means a negative power supply - usually a very fast destruction...

These were the easier errors, though. The design errors are usually harder to find. Just a handful selection of common errors:

  • not considering the Absolute Maximum values in the data sheet for the pin voltages: for most of the CMOS IC pins, letting a pin above Vcc or below GND usually makes the intrinsic p-n junction forward-biased and that could cause large currents flowing - therefore connecting circuits with separately switched power supplies can be tricky,

  • not considering the Absolute Maximum values in the data sheet for the pin output current: driving an impedance too small, either because the pin is simply not strong enough for the task, or because a capacitive load is driven with a high frequency signal without a proper series resistor,

  • letting a MOSFET gate practically unconnected could let it wander into a half-way open position, causing a large dissipation within the MOSFET (this can be created by letting an MCU input unconnected, for example by connecting it only to another MCU output which happens to be in high-impedance state because the MCU's RESET pin is asserted or because the code is being debugged and stopped before configuring it correctly),

  • configuring multiple push-pull outputs "against" each other (connecting multiple outputs together without making sure the outputs are driven by the same signal, or making sure only one of them is enabled at a time),

  • improper strength for power MOSFET gate driving, considering the frequency and the load involved, leading to the MOSFET spending a large time in transitional region, and causing a large dissipation,

  • driving complementer transistor pairs (PNP & NPN or P & N MOSFET) without proper dead time (this results in both transistors forwarding at the same time, a "shoot-through" situation).

(The last two examples are not really applicable for MCUs but for MOSFETs but they are still worth mentioning here.)

\$\endgroup\$
  • \$\begingroup\$ Thanks for such a comprehensive answer, this is exactly I was looking for. \$\endgroup\$ – Zhiyong Li Apr 3 '16 at 17:26
0
\$\begingroup\$

These kinds of questions, unfortunately, should not be posted on Stack Exchange.

From my experience, I would recommend soldering up another PCB, as your current one is damaged. It is possible that the board was manufactured incorrectly and a short-circuit occurred. I have seen this with my own eyes before after examining a board with a microscope. However, this is unlikely if you are not working with an extremely complex PCB.

If possible, solder up the PCB one IC (integrated circuit at a time). I would solder up the power related chips first and then put a dummy load (a lot of resistors in parallel) and make sure that is operating properly. Then, I might put on the microcontroller. Alternatively, you could put on the microcontroller and power the PCB from a bench supply -- this way, you would know that you don't have any power problems.

\$\endgroup\$
  • \$\begingroup\$ unfortunately I have already soldered 5 of them. They all had the same issue: MCU got fried in less than a minute. I have also tried other methods you mentioned. Soldering IC one by one. and taking out them one by one. So far no luck. I am simply trying to get some high level input. If this offends you. I am sorry. Have a good day. \$\endgroup\$ – Zhiyong Li Apr 2 '16 at 19:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.