Yes I change the resolution of the picture in my messages! sorry! none unconnected wire in the design , of course I have checked before making the board
I designed this board as a datalogger for my weather meteo station
yes the silkscreen is present
Do you think it is the chip itself or maybe error is present in the design? After checking the design during one week I do not always see the error... and why a shurtcut in my supply 
I would like a link to the eagle files - or an uploaded .zip file here in the forum. The immediately readable images are important, but the “source code” surely won’t hurt.
I looks as though after the board's ground plane has been flood filled, the SAMD21G17A's GND net isn't connected to anything (on the left), also the VDDCORE and VDDIN capacitors' GND is also isolated (on the right):
Then you will find shematic and board file that you can open with eagle...
You have well see Martin, unfortunately in the file you will see that you have a via connected to GND (but not visible in the jpg image) , idem the GND fo the capacitors (VDD and VDDCORE) are also connected to GND with via (not visible)
if you have eagle you will beable to see them... Anyway with help of DRC , I have none unconnected wire... then for me it is ok for the layout. I assume that problem has to be present in the schematic
Are we supposed to be checking the board layout with the SAMD21 48-pin TQFP package or the QFN?
Normally it is the same wiring concerning the chip (SAMD21G) but for information it is the QFN package which has been used.
I believe that 17 or 18 is just the size of the memory...
@edouard56899 It's just that your earlier PCB image shows the QFN package, while your Eagle screenshot (with the GND net highlighted) shows a TQFP.
My mistake, now you point it out, I can see the vias on the PCB connecting the GND planes.
@MartinL I make an error in the different version of my board, big headache with this project!
you see something from your side? I believe problem is coming from the inside of the chip (but not completely sure because I do not have this chip with me to change it on the board)
@edouard56899 Maybe it's just pixelation, the only other obvious thing I can see at the moment is the clearance between AREF and GND appears to be quite close:
hello guys
do you find something wrong with the eagle files?
I think it is the chip which is burnt... I have no idea
thanks again
I will done an update with other chips...
Hi,
What are you using as a power supply?
Does it have current limit control?
What current did the faulty board draw?
Can you remove a controller from the PCB and power it up, and measure the voltages at each of the controller pins, cross checking with your schematic.
Also check voltages at strategic points such as the power and gnd pins of the other ICs.
Tom.... 
PS, You are in troubleshooting mode now and must form some sort of strategy on how to find the fault and fix your problem.
Going through the most obvious visual inspection:
Have you checked that the QFN's orientation is correct? Or for that matter any other polarized components, as sometimes it can just be something as simple as a diode mounted the wrong way.
Are the SAMD21 leads in contact with the board? If the solder paste mask aperture for the external pad is too large, (assuming your soldering the pad), this can sometimes result in the QFN package floating on a ball of solder preventing the leads from making contact with the board.
Hello
Yes Idea is to check the resistivity between 3.3V and GND for a chip alone. Yes QFN orientation is correct...even if the pad is too large and 3.3V pin do not touch the board , the microcontroler does not work properly but there would be none shutcut...To have shurtcut , it means that 3.3V and GND are linked by a component...
current for samd is really too much , I have for example a measurement of 1A for 3V !!! and it is the only chip which is hot on my board, then it is the chip in faulty... If remove it , everything is ok on the board and voltage from my converter is ok...
I will see with the company which has made the assembly, maybe it will be possible as proposed to solder only the microcontroler , capacitor and quartz on a new board just to check if it is yet ok like that.
I propose to tell you the final word next month after checking every thing on the board but I would like also to thank you for your help in this forum
If you think whatever , not to hesitate to write me!
I looked at the EAGLE .brd file (unfortunately, I can't look at the .sch, because you have a newer version of EAGLE than I do.) I don't see anything specifically WRONG, but there are some things that are worrisome.
Worst, I think, is your 3.3V power trace, which is quite thin and wanders all over the board. I don't see how that would cause the problems that you're actually seeing, but it's worrisome.
A somewhat similar design I have (for SAMC21) looks like this (and it works!)
Also a lack of bypass caps..
Tom...
hello
@TomGeorge Normally every caps are mounted closed to the chip. you means decoupling capacitors?
I am ok with @westfw , 3.3V is not really well traced but because current in the board for 3.3V is really weak (less than 100mA) it is not normally a problem , no need to make a fat power trace. And unfortunatelly for me , I can not increase the size of board. But If I have enougth time I will make an other design to have a beautiful layout as your!
Then ok no problem concerning the shematic and layout, then it means that maybe chip on them could be the problem
I will try to fix it and I will tell you
Both...
Tom..
This is thinking only in the DC paradigm. It isn't just about providing power to different devices, it's also about minimizing the supply voltage AC coupling between different devices.
Also, 100mA isn't weak. It's significant.
Regards troubleshooting, approach it looking for general symptoms first. Excessive supply current, unexpected heating or warming of parts, etc. Check simple digital signals, like clock. Make sure the chip has working code it it - the fuses have to be set correctly for the clock options and so forth.
This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.
