[SOLVED] My PCB for my dspic33 is shorted between power and ground, and I don't know how

EDIT by hermit because this is a long thread. Solution found in this post. Much thanks to jmk!

I recently designed my first PCB. I designed it with KiCad and ordered from Osh Park. Basically, I made a development board for the microcontroller that I normally use on a breadboard, a dspic33ev256gm102. On the PCB, the pins are broken out to a bunch of male headers. I have LEDs and switches that aren’t hardwired to the microcontroller, they are just available if I want to wire wrap to them. The dspic33ev256gm102 was on backorder so I purchased some dspic33ev256gm002 instead, which shares the same datasheet and pinout, just missing CAN functionality as I understood it. It is an SSOP package.

I checked the PCB for shorts before populating it. I populated my PCB with all of the parts, cleaned the board with 91% alcohol, and checked for shorts again using a multimeter. I let it dry for about 24 hours and tried programming it.
As soon as I plugged it in to its 5V power source, I heard fizzing. The programmer could not detect the device and the microcontroller IC got hot. I unplugged it and checked for shorts again. This time I found a short between power and ground. I am confused how this happened since it wasn’t there before. I looked at my board under a magnifying glass and could not find any solder bridges.

I had a spare copy of my PCB and spare parts, so I decided to try again. This time, I only populated the microcontroller, its bypass capacitors, its Vcap capacitor, and a couple resistors needed for the programmer. I also soldered some wires onto the board for power and the pickit3. I checked for shorts on this new board and did not find any. I plugged it in to power and it started smoking. I unplugged it and checked for shorts again, and what do you know, power and ground were shorted again.
I attached a picture of my schematic. Could somebody please review it and tell me if I made a bad mistake somewhere? I can provide the KiCad project files as well if you want to see my layout and BOM.

If there had been a short between the power lines on the board as received, and you did test for this, then the uC wouldn’t have got hot. I think you wired up the uC wrong and killed it then the power lines were shorted by the dead uC. Check the pinouts of the uC against your schematic carefully.


Hi kenyapcomau. The uC is an SSOP package and it has a small indented circle in one of the corners. I am treating this corner pin as pin 1, and I double checked my schematic’s pinout and my layout several times to make sure I didn’t mount the IC wrong. I attached a picture showing the pinout.

And here is a picture of my uC in the layoutuC_layout

I’m having a hard time figuering out which info you post is relevant, and which not…
Schematic is low resolution, etc which makes it hard to judge whether the schematic is ok.
Then I thought of:

If pin 27 is GNDREF, and pin 28 is VCC, and if the DRC check is OK, then it’s unlikely that the generated gerbers have the short in them.

Have you checked the DRC?
Pcbnew / Inspect / Design Rules Check / Run DRC
You can also use the “highlight net” functions both in Pcbnew and Eeschema for verification.

Other common errors are:

  • Soldering IC’s upside down because a dimple in the mold is mistaken for a “pin 1 mark”.
  • Shorts between pins during soldering.
  • Damaged capacitors. Ceramic capacitors break relatively easy because they are brittle, Tantalums get damaged by heat easily.
  • Faulty components fom Ali / Ebay etc.

After re-reading your post…

Mounting the IC upside down does seem one of the more likely causes.
When you orient the IC in such a way that the text is normally readable, then pin 1 is (almost?) always in the lower left corner. In the screenshot from the datasheet pinout, this is not so!

1 Like

Any chance you’ve powered the uC backwards? If the uC is on the reverse side from the rest of the components, did you remember to flip the footprint?


I was going to ask about how they were powering this thing up. Voltage regulator? Direct wall wart?

1 Like

What sort of polarized caps are you using?
Sometimes the positive side is marked, sometimes the negative side, and it’s not always unambiguous.

Sometimes it helps to put a few amps through a short and then look for hot spots on the PCB.
A FLIR camera would be nice, but a bit of Isopropanol also works. It evaporates a lot faster on hot spots.

I have a micro-USB breakout adapter that I solder onto J10 in my schematic. I use an iPhone 5V supply.

I actually do not use any polarized caps, despite my schematic showing one for C2. I’m using ceramic capacitors only, which I don’t believe can be polarized. I like your idea with the alcohol, I guess I could try that on one of my shorted boards.

Another thing though, on my second attempt (second board), when I applied power I saw glowing and smoke coming from pins 27 and 28. Does that tell me that the short was located literally in that exact location? Or maybe that C4 (located next to that pair of pins) got shorted?

I have the uC on the front of the board so I don’t think that’s it. Question though, because this was my first time using KiCad and maybe I did something wrong. For the components that I did put on the back, all I did was press the F key to put them on the back side of the PCB, does this handle the flipping of the footprint as well?

Seeing that flash is important, I believe.
Can you post a photo of pcb 2?

The failed boards are not valuable for further testing because of internal device damage.

If you build #3, put a >1k resistor in series with the +5. Then measure current and voltage at Vcc pins. Compare to a diode rather than the cpu.

My guess is something is wrong with the design or layout and the body diode is conducting.

1 Like

paulvdh, sorry about all text, just trying to provide as much detail as possible to get this solved. And, sorry about the resolution, I had to resize the image to upload it here.

Yes, I ran the DRC during the design process and today as well. I took advantage of the highlight nets function and took another look at my gerbers. All looks ok to me…

This being an SSOP, there’s a chance I made a solder bridge somewhere, but before I applied power to the board I checked for shorts and there weren’t any. I got the uC from Mouser, and I got all the other parts from Digikey, so hopefully the parts are authentic.

I am confused about the pin 1 dimple and soldering upside down idea. I thought the indented circle always indicates pin 1?

Is this really a 5v part? Not 3.3?

The iPad wall wart may not necessarily be putting out 5v. Measure it with no load. It could be much higher.

1 Like

F for flip is the normal way to move a footprint from front to back and back again.
On the top (Red) your dsPIC numbering is anti clockwise, if you flip it to the back you look “through” the PCB and the numbering will be clockwise.

One of the advantages of a current limited lab supply (with meters) is that you can often prevent the smoke from escaping in cases like this.

What did you use to orient the IC and make sure it is not soldered backwards?

Yep, this is a strong indication that much more current went through that spot then should have.

Another easy to make silly mistake:
Are you sure you did not reverse the power supply itself?

Here is a picture of PCB #2

I use the same wall wart for a breadboard which uses the same uC in a DIP package.