Mega-R3, Fried Usb controller chip?

Development Tools IDE 1.x
1

I believe I've fried the usb controllers on three mega-r3's - I could use some advice about this, here's the scenario and - could this lead to frying something on the megas?

On a mega shield I have a 12vdc-5vdc dc-dc convertor which is connected to the gnd and +5vdc pins on the mega's header. The convertor works just fine, and powers the arduino just fine. All is well at this point. While the dc-dc convertor was powering the mega, I also got into the habit of plugging in the usb cable to upload new software, or communicate with the mega while running. At that point I'm supplying both usb power, AND the dc-dc convertors power. And it works, for awhile. Sometime later I notice that the board is no longer talking to me, and neither the pc or the mac can see the usb connection anymore. Here today, gone tomorrow. Those usb ports are simply gone.

When the second board died I noticed it right away, and picked up the mega/shield stack and could feel that it was slightly warm on the underside of the mega board about where the usb jack and controller chip are. And I don't think that area should ever get warm.

Now, even though both boards will not communicate usb anymore, I know the processors are running because they're still toggling the i/o pins of the test program I'd last uploaded. Both the megas onboard 5v and 3.3 volt regulators are still working fine (I measured them), and the boards (without any shields or my dc-dc convertor) can be powered up from either usb power, or the barrel jack for an external supply.

So, when using both 5vdc from an external source AND usb power at the same time, can anybody think of a reason why that would harm the usb controller chip and NOT the processor? They're both powered from the same 5v supply! - And, why would it take time, we're talking a half hour or more perhaps to self-destruct? I would think that doing something really gnarly in power supply land would kill things sooner rather than later...

Anybody have any thoughts on the matter? (and thanks).
Kurt

2

All may not be lost.
What is your host platform? MS win, Mac Linux?
Have you tried rebooting the host?

3

OTOH, all may be lost. Read this thread.
http://arduino.cc/forum/index.php/topic,82046.0.html

4

I still find it hard to believe. Anyone try to develop a theory why this would be?

5

My main comment is that by wiring the output of the dc/dc converter directly to the arduino's 5vdc bus and then plugging in the usb cable you are 'hard-wiring' two different +5vdc voltage sources together from two different regulators. This is never considered a good engineering practice, although many do so anyway, often with no problems reported. Note that the arduino board uses a mosfet to isolate the on-board +5vdc regulator's output voltage from the USB +5vdc voltage, and thought that was important enough to add that isolation circuitry.

I suspect your dc/dc converter doesn't play well with another +5vdc voltage source wired to it's output. Now as to why the serial converter chip is more sensitive to whatever fault condition it's being subjected to, I don't have a clue. But I have no doubt that the root cause is based on the hard-wiring of two independent and active voltage sources being wired together.

As the song says, sometimes you are the windshield, sometimes your the bug.

Lefty

6

I have new, not good information to report about this problem and I've posted the results here:

http://arduino.cc/forum/index.php/topic,82046.0.html

It's beginning to look like either a design flaw, or bad batch of mega rev 3's to me, as a mega rev2 is working just fine.

@Lefty - "sometimes you are the windshield, sometimes your the bug." - Arrrggghhh, another quote comes to mind as well "Have you ever considered that your purpose in life might be to serve as an example to others?" (grin)