Not much left here. The DIP package may simply be more fault tolerant on startup or you may be damaging the chip when you solder it. The start up oscillation seems reasonable and using pull downs on startup couldn’t hurt as a test. Always a chance of a bad batch of chips. It would be painful but if you have a DIP socket you could try wiring that on and try the DIP package.
Did you get the chips from a reputable source? They may not even be the correct device if from ebay or aliexpress.
I ordered the ICs from Mouser, all other components from Digkey
Maybe so, but then why wouldn’t Microchip mention this difference between DIP and SSOP in their datasheet?
I was wondering if the chips might be bad, but I had the same issue on 2 dspic33ev256gm002 and 1 dspic33ev256gm102. I wouldn’t think they would manufacture different parts in the same batch. Maybe I’m very unlucky and received 2 separate bad batches of chips…unlikely.
I don’t know if there are adapters to go the other way but you could use one of these and drop it on your breadboard. Might tell you something.
This is a good idea. I ordered some breakout adapters on Amazon and I’ll test my next uC on a breadboard maybe by Monday.
I received my protoboard in the mail today and tested my SSOP uC on a breadboard. I only built the “minimum recommended” circuit in the datasheet, plus the programmer circuit.
When I first applied power, my power supply’s displayed voltage jumped around quite a bit between 3 and 5 volts before settling at 5V. Excessive current was not drawn. I was able to identify the target device and program it successfully.
So I think that means my PCB has an error somewhere. Or maybe the bypass capacitors are bad. I don’t know, I feel as if a dozen different people have reviewed my schematic/layout, and no one has found anything that resembles an error.
I used several dsPic’s a decade ago but all were THT. Never had a problem after discovering it was very picky about capacitance; Especially with respect to the XTAL. Also, unwieldy capacitance at power supply and LCD. Needed to dial it in differently on the PCB versus the Breadboard.
Perhaps of no help but. Screenshot of schem section…
Very interesting. I’m not using an external crystal on mine. I think there’s just something wrong with my PCB. Maybe it’s my clearances between contacts on the SSOP footprint.
I didn’t want to do this but I had no choice. I soldered 30 AWG wire to my 0603 and 1206 capacitors, then I soldered the other end to some male headers and plugged the caps into my uC #4 breadboard. They appear to work just fine (as in they aren’t getting destroyed when they receive power). So that’s another thing to eliminate from the list of potential suspects.
Any other ideas?
I think you answered in your original post, but I want to make certain: Have you powered up a bare board with no parts mounted?
- Use a current limited supply. This can be done with a good bench supply or
a cheap supply and a power resistor,
a cheap supply with an e-load in series to limit the current.
As a practical matter, I find that 1 Amp and 5V are good settings. So if you use a power resistor, use something like 5 ohms and 10 watts rating to limit the current into a short at 1 Amp.
If your resistor heats too much, its value (and the resulting current) can drift as it heats and make this process more difficult to compare readings as you work.
With the board powered up, use a good digital multimeter (set to a sensitive voltage scale) to probe your traces and vias and find the millivolt level voltage drops. The polarity of these voltage drops indicate current flow through pcb resistance and should point to the direction of current flow.
This way of measuring is much better than simply using an ohm meter. That can work in theory but is very difficult to measure a few milliohms.
This method might work with a populated board but I think it is much better on a bare one. I have used this method successfully more than once.
With an LM317 an a 1 Ohm resistor, you can make an 1.25A current source or sink, depending on how you connect it.
If you use a 1.25 Ohm resistor and calibrate it to be very near to 1A, then it is a simply yet useful addition to your toolbox for measuring small resistances. Each mV measured is a mOhm, and because your current sources is a separate circuit it’s also very easy to make 4-terminal measurements.
Calibration can be done with your DMM in current mode and adding blobs of solder on a piece of resistance wire. Just tinker a bit until you have 1.00A.
Related to finding shorts:
Small audio amp (Such as TDA2050) and a function generator to make a relatively high current AC current, which you can pick up with an inductor (or cassette tape head), amplify a bit and either measure or use headphones to follow the (short) circuit current through your PCB.
It also seems to be getting time to not exclude the “weird” stuff, but there is too much to make some list, and it’s all suggestions.
Could it be that your power supply has overshoot during trurn on?
How careful are you generally with ESD precautions?
1 Like
That is a great idea, but be careful to heat sink the LM317. With a few volts differential and 1 Amp, several watts of heat will cook even a TO-220 version LM317 unless it is on a good heatsink.
My other comment is that for locating voltage drops (not precise resistance measurements) to locate a short, I want to have a stable current but a precise current is less important.
1 Like
Never mind about my last post with the capacitors, they aren’t working fine. After leaving power applied for a few minutes because I forgot to shut off the supply, the limited current was reached and the voltage dropped low. Also, I experience the same thing when I have the through hole caps plugged in, I just didn’t realize it yesterday.
Now my 4th uC, which is on a protoboard mounted to a breadboard, is also toasted. I saw pins 27/28 glow up and now my programmer can’t identify the target device anymore.
So there’s another clue, I don’t think it’s my PCB’s fault anymore. I only have 2 uCs remaining, maybe 3 if #3 is still good.
I don’t think I ever powered up a bare board, but I did Ohm it out. I like this idea, but if you read my latest post, I wasn’t even using my PCB for the 4th uC and it still burnt up.
For uC #1, I used an iPhone wall wart. For uC #2, I used the programmer as the power source.
For uC #3 and #4, I used my current limited supply. Any time I use this supply I make sure to power it on first before attaching the leads. This is a cheap supply I got off Amazon, and I realize it may be unsafe to use, but it is not the reason why all 4 uCs got toasted. The reason why I say it is unsafe is because even though I set a current limit of say 20mA, there are instances where the display reads something much higher like 100mA.
I’m pretty careful with ESD precautions. I wore a wrist strap connected to a wall outlet and ESD safe mat when I was soldering components. If I want to touch my circuit I will discharge myself on something metal/grounded first, usually my PC case (sounds like a bad idea but pretty sure it’s a safe method).
And in which of the terminals did you plug it?
(Sorry, could not resist).
I once had a loose connection in the Earth lead of the mains wiring. This was not only a silent fault, but capacitive coupling between the other wires can be small enough to not feel any tingling, while still enough to fry electronics, or render your ESD mat useless.
And for what it’s worth, you have my sympathy. I’ve also been in similar situations where the cause of some fault in electronics eluded me for weeks, and fortunately the last time this happened was quite some years ago.
So, this is a commercial board connected to your breadboard the same way as your THT IC and giving the same result as if soldered to your Kicad board.
As I see it:
Something is causing a massive short inside your IC resulting in 27/28 glowing which I find hard to believe because of the construction of ICs.
or, something is causing tracking, in time(seconds), on the boards, resulting in a short between 27/28 and possibly other pins.
Are you sure you are not using an active flux? Some will cause PC boards to become conductive and exhibit a very low resistance and there is not much space between 27 & 28.
As a possible test, strip all the components from one of your previously used Kicad boards (one that had glowing pins 27/28) and check the resistance between the power and earth tracks.
Must have been the right one as his eyes didn’t glow like 27/28.
1 Like
I use this RA flux and solder. I bought both of these specifically for this project. I’ve never used separate flux before because I’ve never needed to. Again, this is my first PCB and since I used surface mount parts I knew I needed flux. Are you telling me that because my flux is active, I may be shorting things even though I cleaned with alcohol?
Maybe.
Sort of running out of ideas 
Easiest way to tell is remove all the bits from one of the dud boards then check the resistance between the tracks that went to 27&28.
If there is no resistance, connect your power supply to the empty board, put a blob of flux over 27&28, switch on power, and watch. If you get sizzling and/or smoke your flux is conducting…not good.
1 Like