Fail to make own STM32 board

[sragen]

Posted on August 23, 2011 at 05:28

Dear all,

Sorry, I am a newbie on STM32 microcontroller.Yesterday I have made my own STM32F103RBT6 board. I use L1117-3.3 as my 3,3 volt regulator. The external components are Crystalls, capacitors for crystall, supply capacitors, reset circuit (using R and C), and resistor pull down for boot 0 and boot 1 pin.When I apply 5 volt voltage to the regulator, firstly its is OK it only draws current about 4mA. But several second after that the current increase over 200mA. and then I check the VDD and VSS has been short Circuit. And the chip has been damaged.Thus I remove the chip and measure the VDD-VSS voltage, and it is 3.3volt. I've recheck my PCB, and I can't found any mistakes. I've already burn 5 chips last night, and I don not know whats wrong with the circuit.If the there any suggestions  for my problem will be very appreciated.Thank for advance.Best Regard,Eko Henfri. #make-own-stm32-board

[drzile86]

Posted on August 24, 2011 at 13:52

Hi, 

I have looked at your schematic, everything is OK, except you don't need pull up resistor on the NRST pin, STM32F1xx MCU has it embedded. Look at the AN2586. Maybe this will solve your problem.Regards,Dragan

[John F.]

Posted on August 24, 2011 at 14:52

Bypass your regulator and use a +3.3V supply. If you have analog supply and VBatt supply make sure these are connected. Still have a problem?

Make another PCB and solder only the passive components and decoupling and crystal etc. What current is drawn? Now insulate the other pads with tape and solder only the power pins of the microcontroller. What current is drawn? Go round the pins one by one with a voltmeter and see if that gives you any clue.If all else fails, post the schematic you ''copied'' and your own (with pin numbers) and we can look at that. It just isn't meant to be this hard. You are the only person to have this kind of problem - must be your power supply or your hardware (PCB).

[sragen]

Posted on August 24, 2011 at 19:29

Thanks John, I had connected VBAT to VDD, and the problem is solved on my STM32F103C8T6, but it did not work on my STM32F103RBT6 chips that I've posted earlier. Maybe my chip already damaged when I solder the chip. But I am not sure why VBAT if not connected to VDD can cause the chip's current increased (since in the manual it just a recommendation, not an obligation which can damage the chip).

Now I am waiting for the new chip from ebay that I've ordered two weeks ago. I will post again the result after I get the new one. Maybe it takes several days from now, since I lived in Indonesia where the chip is not available in local store.By the way thanks for all suggestions and help guys. I really appreciate it.Best Regards,Eko Henfri.

[raptorhal2]

Posted on August 24, 2011 at 20:54

I have successfully checked out my 103C8T6 and 103RBT6 PCB prototypes with VBAT not connected. The backup registers and RTC contents aren't preserved during a power cycle, but that is to be expected.

Unless you started with defective chips last time, I think your current plan will release more magic smoke with the new chips. Before you try the new chips, do the following: - Make absolutely sure you have Pin 1 in the correct PCB corner. You stated earlier, Pin 1 was the lower right, but the chip dot marker on your schematic correctly shows it upper right. That could be just how the diagram was oriented in your mind at the time, but check it out. - Have someone take a good look at the board. - Record the chip version and manufacturing date markings on the RBT6 chips. Contact your regional ST representative and ask him/her if they discovered that they had released some defective chips to the market. The answer is probably ''no'', but if ''yes'', then there is some confidence that the new chips will work. Cheers, Hal

[sragen]

Posted on August 24, 2011 at 22:17

Thanks Hal for the suggestion.

From the orientation of PCB picture I've posted earlier, it is correct that number 1 pin is on the upper right corner side. What I mean in the lower left corner is, the dot (pin number 1) of STM32 chip is located in lower left corner where I can read the text printed on the chip (ARM logo, series, etc).Currently my suspect is my soldering method who make the chip damaged, even though I just wonder how fragile this chip from the temperature during soldering. As your suggestion, I already asked my friend to check the PCB, and he can not find where the problem is.I can not contact the seller for the chips that I've got directly, since I bought them through my friend in South Korea. But I think they will not sell a defective product to the consumer, but bad binning chip (low quality silicon die or package) maybe possible thus make it more fragile from hot temperature. But it is purely my humble opinion and prediction in which I can not assure weather it is correct or wrong.If you have a trusted seller who sell a good quality STM32 chip, I think it will be useful for me. As for now I already ordered the chip from ebay, which I did not know yet what the chip quality is.Best regards,Eko Henfri.

[raptorhal2]

Posted on August 25, 2011 at 01:50

An electronics supplier will provide the chips in a protective foil sealed package. It maintains low humidity and protects from ESD damage. I order my chips from www.mouser.com. Other electronics suppliers may be closer to you.

Cheers, Hal

[infoinfo989]

Posted on August 25, 2011 at 06:53

Your PCB is very simple and the chips are fairly robust. I hand-solder them without any problems. Assuming you're taking reasonable precautions (you're not holding onto a tesla coil with one hand while soldering your board with the other) I would imagine you should be ok.

However your PCB is little more than a breakout board, and I presume it's plugged into something (like a base-board) that gives it power and perhaps has other things hooked up. Have you doubled-checked that base-board? Pulled out everything that's not strictly necessary for an initial power-up? Confirmed that base-board doesn't have a diode or transistor backwards or anything like that? If you have LEDs, check the series resistor of the LEDs are a large enough value? Obviously I'm just guessing here, but clearly something's going on, and I know from my own experiences, when I blame the chip, 95% of the time I'm wrong. It's almost always not a problem with the chip. Beyond that, I second the idea of checking all the pads on a board that doesn't have a chip soldered down yet. Power it up and make sure every pad for the IC has the voltage on it you expect. Good luck!

[sragen]

Posted on August 25, 2011 at 12:06

Thanks Frank for the response.

You're correct. My PCB actually just a breakout board that will be used with the main board. It was very very simple test, very basic. Yesterday I did not connected the board to anything else except 5 Volt supply to the regulator.  The other pins on the chip except VDDs, VSSs, VDDA, VSSA, Oscillator , BOOT 0, BOOT 1 and NRST are left unconnected.I already checked the output voltage from the regulator was just fine, 3.3 Volt. I could not measure the other pins just yet, since the chip will goes to short circuit in just several seconds. I already removed the chip from the PCB, powered it up and then measured each pin's voltage to ensure there is no voltage more than 3.3 volt goes to them.  Then I soldered the new chip on it, tested it, and the same situation occurred. As for the chip orientation, I am sure that the chip was already soldered in the correct direction.I think the chip has been damaged during the handling process is very low possibility, since I lived in the tropical country in where electric static is not a big issue. Until now I still have no clear clue what is going on except may soldering method as the culprit.Best Regards,Eko Henfri.

[donald2]

Posted on August 26, 2011 at 17:15

From the time period involved I initially suspected that the processor was trying to run the built-in boot code, eventually switching to driving a pin that was connected to a power rail.  But I haven't spotted anything in the schematic (e.g. disconnected Boot0 pin) that would trigger a ROM or RAM boot.

My experience is that the chip packages are pretty resistant to soldering abuse.  I didn't have time to a make a stencil for our first board.  Despite knowing that always leads to too much solder paste, my ''bare minimum'' hand application put on too much. It took several reflows, removing the chip with a hot air gun, wicking off the solder, debugging and a bunch of hand soldering to get all of the pins connected.  (That included a bunch of other component removals as the OSC pins). By the time I had finished the pretty purple solder mask had turned an ugly brown from the heat, but the chip still worked 100%. Have you tried applying a low voltage (1.5V) to the board?   You might be able to tell if the power spikes as the chip comes out of reset, or if there is a constant current drain.

[jj2]

Posted on August 27, 2011 at 21:53

Really this is unpleasant - perhaps something ''easy'' has been missed?

Both your 5V supply and 3V3 regulator have (we assume) remain unchanged during the repeated (now mass) burial of STM32 devices.  Is it not in the realm of possibility that - especially during power ''turn-on'' - a voltage transient eludes or someway/somehow bypasses your regulator?  If not an ''over-voltage'' - brief ''reverse polarity'' (even < -1V) could also be a cause.  Does your regulator have some ''minimum'' output current to meet its ''regulation'' spec?  Powering only the MCU may heighten the chance for ''turn-on'' transient propagation.  ''Should'' this suggestion warrant investigation - I'd ''pull'' the  existing 3V3 regulator and find a superior 3V3 source.  To eliminate ''start-up'' transients from this new supply - you may consider using a FET and ''bounce-free'' separate circuit to apply the 3V3.  (you may need some capacitive load to meet VDD rise specs - if any)  It is also worthwhile to recheck your reset circuit - insure that it's correct. My 0.02 USD rides on ''transient voltage'' pass-thru (heightened by too light a load) being your culprit.  A scope with storage - triggered upon the turn-on voltage - should be easily able to capture such an event. Should these suggestions fail - I would then raise my wager to 0.05 USD - and challenge your understanding of ''pin 1's'' real location and/or direction of pin sequencing.