next topic »

[on:]

While I was trying out the example project "AnalogInOutSerial", the pot was fully turned to the 3V3 end of its range, and the wiper wire came out of A0 socket pin and brushed against the Native USB shield (around the USB connector). I thought nothing of it because it was an extremely low impedance ground. That's what the shield is for.

I reconnected the wire, and for some reason the LED stayed completely off. I checked the wiring, it all looked right, but the connector felt warm. I brushed my finger across the ARM and it was smoking hot!

The board is now utterly dead. When DC power is applied (7V), the regulators overheat and the ARM chip smokes within 10 seconds. Disconnecting the DC jack and plugging in the Native or Programming USB ports faults my USB powered hub. So this is a dead parrot.

Here's a lesson for anyone playing with the Due : DO NOT, UNDER ANY CIRCUMSTANCES, ALLOW ANY CURRENT TO FLOW THROUGH THE USB BACKSHELLS!

Looking at the circuit, the only thing I can see that would possibly cause this is if the USB ground was not low impedance enough, AND the input voltage dropped enough when the short occurred that the regulators went out of regulation, in which case, the D+/D- active dampers Z1-Z5, may have conducted the full 3V3 to the Data lines for just an instant (but why would 3.3V be a problem? Only makes sense if the regulators were shorted) and deep-fried the ARM.

In case anyone thinks I might have mistakenly connected the pot to the 5V rail, I was particularly careful not to. I cut a set of SIL pins to suit the connector, and specifically pulled out the 5V pin so there was no possibility of accidentally soldering the pot wire to the 5V pin, because the 5V pin is not there.

This is really frustrating. I can't believe an experimental board could fry so simply, with a single loose wire. Unbelievable.

I'd be happy to return the Due to the designers for fault analysis. Hey, it's $60, but I have plenty of PICs that I can use in the meantime.

So do keep your wires clear of the board at all times. And as I said, don't, under any circumstances, short anything to anything!

[Reply #1 on:]

Are you sure the wire didn't brush against the polyfuse right next to the USB port? It's at 5v and is fairly exposed.

[Reply #2 on:]

No, I didn't think of the polyfuse. Thanks for the suggestion, I appreciate it!

And I believe I have lied to you all. I hadn't touched the board after it fried, so this morning when I went in to the morgue, I noticed something I missed yesterday...

I had a lead connected to the barrel of the DC jack as a low-impedance ground for my circuit (pulsing a small detector coil). Today, I noticed that the lead was right next to the USB shield (shield as in the metal jacket around the PCB connector, not shield as in an Arduino shield! Or a hoplite, either!). It is entirely possible that this big, heavy ground wire could have connected with the USB parts. It would have been from the top, not the bottom (I'm paranoid about stuff getting under any of my boards, so I have a non-conductive antistatic foam dam around the board's underside), so the fuse issue may actually be what went wrong...

Now that I've seen that potential hazard, I can't claim to know for sure what the hell happened. Sigh.

I contacted the local vendor, who, like most other vendors I've spoken to this morning, are completely out of stock of Dues. Which is good, I guess - they're obviously going to be tremendously popular! But it makes it frustrating for an Australian developer with only Paypal.

Still, at least knowing about the polyfuse gives me some incentive to break out the USB microscope and look for clues. Heck, I've got nothing else to do!

But thanks for reading everyone, I appreciate the virtual support!

[Reply #3 on:]

so I have a non-conductive antistatic foam

That sounds like an oxymoron, like jumbo shrimp, or Federal Service. How does a non-conductive antistatic device work anyway? 

Lefty

[Reply #4 on:]

so I have a non-conductive antistatic foam

That sounds like an oxymoron, like jumbo shrimp, or Federal Service. How does a non-conductive antistatic device work anyway? 

Lefty

...or Army Intelligence, or analogue logic...

I meant low-conductivity foam. It's not a pure insulator, so static doesn't build up on it. But at about 20Mohms/cm, it doesn't affect CMOS logic circuits like the 20k/cm stuff that you get from electronic shops does, if you accidentally get some beneath your connectors. Now that's another story altogether... Sigh.