SAMD21 MCUs: VDDANA in appears to be shorted to GNDA

Question

I've been designing a custom circuit around the Microchip ATSAMD21E MCU. I thought my circuit design was fine until I soldered it all together and discovered a short between 3.3V and GND. (Needless to say, this caused the voltage regulator to get hot and angry and smoke to appear, and nothing worked).

After several days of debugging, I realized the MCU itself is causing the short. Upon further inspection, it looks like the VDDANA pin (pin 9), which I connect to 3.3V, is shorted to the GND pin (28) internally to the chip. I even took a brand new MCU out of the tray and tested it without any PCB at all, and lo and behold it is shorting GND(28) to VDDANA(9).

Ok so now I don't understand how I'm supposed to power this MCU!

My schematic:

For reference, this is what the data sheet says I'm supposed to do:

Here's the pinout according to the Datasheet:

I bought them direct from the manufacturer... So far every chip I've tested has this problem (4 out of 4) — Joel Spolsky, Feb 19 '22 at 19:54
Wrong pinouts on schematic symbol. Or datasheet. That's pretty bad. — DKNguyen, Feb 19 '22 at 20:00
Ah. I don't think that's the problem in this case... I attached the pinout from the Datasheet. — Joel Spolsky, Feb 19 '22 at 20:02
I'd email the manufacturer just to make sure you're not chasing your tail. You aren't mistaking the wrong corner of the chip for pin 1 are you? — DKNguyen, Feb 19 '22 at 20:10
Post your PCB so that we can check that you have the footprint correct. — user1850479, Feb 19 '22 at 20:15
OMG you're right. They have a new variant of this MCU with a different pinout and that seems to be what I bought. — Joel Spolsky, Feb 19 '22 at 20:20
It’s a weird voltage regulator that smokes with the output shorted. Most sensible ones have overload protection. It’s worth figuring out why it smoked. It shouldn’t have, unless it’s something very simple like a zener reference feeding an emitter follower, or a low-dropout shunt that got the ballast resistor sorely overloaded due to the short. — Kuba hasn't forgotten Monica, Feb 20 '22 at 01:18
It was a MIC5225, and indeed the spec sheet says that does seem to have thermal protection. I don't think it actually SMOKED per se but it was incredibly hot to the touch when I pulled the plug. — Joel Spolsky, Feb 20 '22 at 16:10

score 14 · Answer 1 · edited Feb 19 '22 at 20:57

14

Ok, it turns out that the MCU's I purchased are ATSAMD21E17L -- notice the L, which have a different pinout.

There was nothing about "L" variants in the older datasheet I have been reading, but the newest datasheet describes the -L variant as being "Pinout optimized for Analog and PWM" whatever that means.

This probably explains how I was able to buy these chips in the first place, but CMON MICROCHIP WHY YOU GOTTA MAKE MY LIFE IMPOSSIBLE :( :(

Kudos and thanks to TQQQ and DKNguyen who set me on the right track

edited Feb 19 '22 at 20:57

SamGibson

17,231
5
37
58

answered Feb 19 '22 at 20:23

Joel Spolsky

547
1
5
17

1

Putting PA27 and PA28 (assuming these are analog pins given the name) right next to digital logic (VDDCORE, RESETN, main core GND pin) is rather suboptimal from a noise point of view. I can see why they have the variant, although the rearrangement is unfortunate. – TLW Feb 20 '22 at 07:52
2

It really never occurred to me that there might be a pinout-incompatible variant and that might be my problem. I guess, because I am too new at this, and just assumed if it wasn't working, it had to be my fault :) – Joel Spolsky Feb 20 '22 at 16:06
1

@JoelSpolsky Well, it was - you just assumed the fault was somewhere it wasn't ;) – J... Feb 20 '22 at 19:06

SAMD21 MCUs: VDDANA in appears to be shorted to GNDA

1 Answers1