What do you mean by body diode, is that something internal to the uC IC?
Yes, Iām certain that the power supply connector is oriented correctly
I actually have a current limited supply, and I will use it next time on board #3. I tested the same circuit on my breadboard with the DIP package version of my uC. It only draws about 5 mA on its own when not doing anything. Iāll limit the current on my supply to around 20mA for board #3 so I donāt destroy it again.
I used RA flux paste to hold the IC in place on the PCB. While soldering, I applied pressure with my tweezers so it didnāt move. In KiCad I placed a dot in the silkscreen indicating pin 1, and I aligned the indented circle of the IC with my silkscreen dot.
I came late to this thread so I donāt bother reading through all posts carefully. I just skim through to find two most important pieces of information which can help finding problems:
- The design files (zipped project) itself, if you can share them
- Direct links to datasheets of the important components
But I didnāt find any. I wonder why, because logically they are the ones which are needed to find problems in a design. Usually screenshots are much better than nothing, but often they fail to convey the critical information and the discussion goes back and forth for days while it could have been solved within minutes.
1 Like
Did you use any sort of glue to hold the IC in place before soldering? Or flux?
Must not forget a faulty soldering iron. Is the tip properly earthed?
Itās also me, some parts of my had make it more difficult for me to sort between information and dross, and too much can lead to a temporary overload.
About the pin one mark.
Yes, Pin one is marked with the pin one mark, but some ICās have artifacts from the injection moulding process which can be more pronounced and be mistaken for the pin one mark.
With the orientation in the photograph you posted (again very low resolution, and not sharp on top of that) the text should be upside down.
Body diode:
ICās are etched on a piece of silicon. The bulk of the silicon is (semi) conducting, and forms diodes with the circuit etched on the surface.
This bulk of the silicon is usually connected to GND, because leaving it floating can (and does) lead to unwanted behavior. It can also act as a thyristor. In early ICās this has been problematic. The thyristor was triggered by āground bounceā, where a signal line was very temporarily below GND during switching, the Thyristor gets triggered and shorts the whole IC (and the PCB with everything on it) until power is removed and the thyristor is āresetā. This usually leads to complete destruction of the IC, as it heats up quickly. But if current is limited, then the IC does not heat up too much and can survive. In modern ICās there usually are measures built in to vastly decrease this effect, but it is never completely eliminated. It may be triggered by the loose wires hanging on the board, lack of decoupling caps and the way you apply power to the board.
On a sidenote: Lots of people have complained about cheap MAX7219 (64 LED display driver) clones from Ali / Ebay / China not working. I have a strong suspicion tat these clones are very susceptible to Ground Bounce and the thyristor, but I need to investigate further to be sure.
About the soldering:
I usually solder a few pins with a blob of solder in one corner of the IC. Usually these are all shorted together at that moment. Then carefully verify the IC is straight on the pads, if not, then re-heat the blob and reposition the IC. When the Positioning is correct, start soldering on the opposite side. and only near the end, when the IC is soldered with many pins and wonāt move anymore, I remove the short created with the first blob of solder.
I guess Iām also guilty of long and only vaguely relevant postsā¦
1 Like
I still think you have reverse powered it. Same colour wires for both +5 and 0, surely thatās prone to a mistake. Quick experiment: have you put voltmeter probes on the power pins of the uC of a ruined board and then turned the power on? That will also tell if you have overvolted it.
1 Like
kenyapcomau,
The red wires were just on the second board. On the first board, I had a micro USB breakout adapter. Yes, I have verified with a voltmeter and ohmmeter that my 5V and ground were connected correctly.
I use a Weller WES51 station with a PES51 iron. I was using a new Weller brand tip, the correct series too (ET I think), but this tip is a little wobbly in the iron which I thought was strange.
eelik,
Iād be happy to share my design files. Iāll do it when Iām home from work today. If Iām not mistaken, this site doesnāt let me upload file types besides jpg, txt, or gif. What is the best way to send them, a link to google drive maybe?
It is possible to upload the files directly, the best way would be to zip your whole project folder, that way all your symbols and footprints will be included (doesnāt take lots of space either).
1 Like
A well know caveat and annoyance in version 5: if you use your own global libraries, only footprints are included in the board, symbols are in a cache file of the project, so you must not remove the lib cache file from the project before zipping.
1 Like
In the screenshot of the PCB you posted earlier, you have 2 named and different nets for pins 27 an 28 of the uC, so if I trust you if you say you did DRC then uploading the project will be of little use.
On this forum thereāse a anti spam robot which prevents new users from uploading files or making links. I think you have gained enough āexperienceā by now to have the right to upload files.
Iāve uploaded whole zipped projects several times myself.
The optimum, is to also have the project folder in the zip file, so others donāt have to create an empty directory for the files.
The most important files are:
asdf.pro
asdf.sch
asdf-cache.lib
asdf.kicad_pcb
1 Like
By the way, when testing a new board, I would recommand to always (and I mean always) use a benchtop power supply with programmable overcurrentprotection. Then you can put something like 20mA or even less and check if everything is good without frying a component.
If you do some electronic, a āgoodā (as in one which output voltage and current can be set) benchtop power supply is IMHO the best gift you can do to yourself 
4 Likes
It would be good procedure to have different colored wires for + and Gnd. If you only have red, then cut the lengths much different for + and Gnd. Or put labels on the wires with tape and write ā+ā and āGNDā on them. Anything to help prevent confusion.
For the #3, only attach the + and GND wires to begin. Do some tests with just the minimum.
You can add programming wires later.
I could see a small breakout board of polyswitch fuses in my future too.
2 Likes
I always use an ohmmeter with a known output polarity and voltage to check for power bus shorts before applying power to a board for the first time. This ohmmeter has a high enough output voltage to forward-bias diodes, but low enough maximum current to avoid damaging mis-wired components.
After assembling my most recent design this basic testing showed a short between the 3.3V and 1.2V buses (each capable of supplying 600 mA) to ground. Eventually I found Iād installed an FPGA (144-pin TQFP) rotated 90 degrees. Removing and reinstalling the FPGA properly fixed the short with no damage done.
1 Like
eelik + others,
Here are my project files. I included some files from the custom library that has my schematic symbol. Iāll take any feedback even if itās not the solution, this was my first PCB design.
dspic33ev256gm102.zip (563.0 KB)
Ok, good points and I will try what you said for #3
And here is the datasheet for the uC