(Solved) LPC-Link 2 Apparently burning out JTAG connections

[lpcware]

Content originally posted in LPCWare by xenthan on Fri Apr 08 07:50:45 MST 2016
I have had 4 chips simply stop being recognized by the debugger (both JTAG and SWD).  I've done 3 different designs now using the LPC4357 (both quad flat and BGA) and the LPC4337.  They have all been programmed using LPCXpresso and I have used 3 different LPC-Link 2  debuggers.  After maybe 3-5 days of programming (4-5 hours per day), the device simply stops being recognized and I am no longer able to program the chip.  I've tested the debugger on another board and it still functions fine.  The program maintains its last state (LEDs blinking, other ICs on the board being initialized and functioning), so the processor itself isn't getting scorched.  Power is supplied via a separate LDO and a wall wart, so I'm not pulling too much current through the  header.  I have a 10k pull-down on the TCK pin and everything else is pulled up internally (except for the TDO line), so I don't believe that's the issue.  ISP pins aren't broken out so I can't check things that way (lesson learned). 

My boards pull a little over 1A total, so maybe there's some issues there when the LPC-Link is attached?  The only connection to my computer's ground is through the LPC-Link's USB port, but I would think that excessive current through that would affect the debugger, not the board itself.  Any ideas?

Edit:
Further testing is showing that the TDI pin seems to be damaged (800 ohms to VDD, 300 to GND) on each board.  Seeing as this is a dedicated input, I'm leaning more and more towards something strange happen with the LPC-Link2.

Solved:
I believe the problem is that there is an issue with using the LPC-Link2 as the sole ground to the PC.  I'm not sure why the LPC43xx chips are the ones that get burned out and not the programmer, but after adding a second ground via a USB cable, the problem has gone away.  I'm pretty sure the potential between the AC/DC converter ground and the PC ground was causing an inrush of some kind that was blowing up the chips.  Still not sure why the LPC-LINK2 is still functioning (would rather have the programmer burn out, not the processor), but I can work around that.  The "Logic Levels and Ground" section of  https://www.lpcware.com/content/faq/lpcxpresso/debug-design hints that the ground could be the issue, but there's no mention of it burning up the actual target board. 

[lpcware]

bump

[lpcware]

Content originally posted in LPCWare by mch0 on Sun Apr 10 06:07:46 MST 2016
I have managed to kill one target as well (just JTAG/SWD) over maybe one or two weeks.
As in your case the remainder of the chip (target) is not affected, I can still reprogram it by using a bootloader. Just JTAG/SWD isn't operational ony more.
However, in may case I am 99% sure that it was my fault: I did not protect the targets's JTAG port against beeing driven by the probe while not powered.
Normally this condiction is avoided by using Vtarget to turn off any signals from the probe, but I omitted that on my probe (hence my fault) since I knew both the target an the probe are working at the same VDDIO.

If your setup also inadvertently can enter that "mode" then I suspect that you overloaded the input protection circuit of TDI and it died after some time due to excessive stress.

Since I can't change the current batch of targets (i.e. cannot add theVtarget pin) I inserted a 2k2 resistor in series for the debug lines. This limits the current that can still flow if I forget to power up the target but does not affect debugging (I have very short lines). As a secondary fix I added external schottky diodes connecting the debug lines to VDD on the target HW. These should help to protect the internal diodes, too.

As of now I have not lost any more targets of that batch and I have done A  LOT of debugging with several pairs of target/probes now.

Mike

[lpcware]

Content originally posted in LPCWare by LessThanZero on Sun Apr 10 00:53:59 MST 2016

Perhaps you can share the debug console log so we can see what, if anything, the debugger reports. Also, if you haven't brought out the ISP line you still have an option with the LPC-Link2. Read this:

ISP Reset over debug


Regards,
LessThanZero