Sorry, you need to enable JavaScript to visit this website.

Zedboard resets upon connecting USB or Ethernet

Unsolved
12 posts / 0 new
segfault's picture
segfault
Junior(0)
Zedboard resets upon connecting USB or Ethernet

Hi

I have some serious looking problems with my Zedboards. It seems as if plugging in an USB or Ethernet cable immediately resets the board. This is visible by the blue "DONE" LED going to the off state. This looks like some ESD issue.

Is this a known problem and will there be a future revision of the PCB which avoids it?

Best regards,
Ben.

PS: I'm using the power supply which came with the ZEDboard.

ctammann's picture
ctammann
Moderator(0)
Hi,

Hi,

You say you've seen the board reset with both USB and Ethernet? If it were just USB perhaps there could be a grounding issue through the USB port since the included supply is not grounded. That wouldn't be true for Ethernet though. We haven't seen this issue before. Try connecting a ground directly to the shield on both the Ethernet port and the USB port when you connect the interface. If it's an ESD issue, grounding the shields directly to ground away from the board should keep the board from resetting. If it still resets there may be another issue.
Thanks
Chris

zedhed's picture
zedhed
Moderator(22)
RE: Zedboard resets upon connecting USB or Ethernet

Hi segfault,

In addition to what Chris mentioned above, when connecting external cables to ZedBoard also be mindful of avoiding contact with jumper header JP13 which is connected to Zynq PS_SRST_B pin.

Regards,

-Kevin

befi's picture
befi
Junior(0)
Hi

Hi

This seems to be a serious design issue of the Zedboard. If being slightly charged, the board resets immediately upon touching the ethernet shield, the usb shield, basically everywhere. And the worst is that it even happens by touching the screws in the mounting holes of the board. As our board is mounted inside a metal housing, touching the housing resets the board. And the screws in the mouting holes are the only connection from the board to the housing.

I would love to hear an official statement about this issue. It happens on all of my 5 boards and is so obvious that I'm almost certain the problem is known. Any news on an upcoming board revision would be very much apprechiated.

Best regards,
Ben.

fletch's picture
fletch
Moderator(15)
More details, please

Hi Ben,
  I'm sorry for the issues you are seeing.  I cannot reproduce this reset problem.  I have a Linux design booting from the SD Card that also configures the PL.  I have tried hot-plugging the USB-UART, USB-JTAG, and the USB host.  I have also hot-plugged ethernet without a single reset.  I have also tried touching all of the connector shields.
  Can you give us more details about your setup?
  There has been some previous internal discussion amongst the Avnet/Digilent teams regarding the connection of connector shields.  We had a problem on an early prototype where the SD Card CD signal would never go low.  The root cause was that the SD cage shield was connected through a 1M-ohm resistor to GND, and the design of the cage tied the CD line to the shield; thus the shield needed to be directly grounded.
  This prompted a discussion as to why a 1M-ohm resistor is used to connect the Shield to GND rather than 0-ohm, which is the case for all other connectors other than the SD Card cage (see R281 and signal name SHIELD on the ZedBoard schematics).  The comment from the designer was, "We connect the shield pins to a shield ring in order to shunt any ESD/EMI directly to the ground source for the board (in this case the barrel connector).  This way any disruptive signals do not flow through ground under sensitive components."
  If you want to try an experiment, try replacing R281 with a wire or zero-ohm resistor to see if this makes a difference.
Bryan

befi's picture
befi
Junior(0)
Hi Fletch

Hi Fletch

Thank you for your answer.

Here is a video of the problem: http://www.youtube.com/watch?v=DBXPmXOtZ3A

I rub my back at my chair and afterwards I touch the ethernet connector. You'll see the board is resetting by the blue "DONE" LED going off. This also happens when I'm touching ground or the mounting screws (although they need a slightly higher charge due to insolation).

As you see on the video, the board is connected to nothing but the power supply delivered with it. It runs the linux demo which was already on the SD card.

I was not able to change the behavior by setting the jumpers to different positions. However, for completeness I list the jumper settings seen in the video:

JP13: open
JP1: open
JP3: open
JP2: open
JP6: closed
JP21: open
JP12: open
JP11: GND
JP10: 3V3
JP9: 3V3
JP8: GND
JP7: GND
J18: 1V8
All unpopulated jumpers are unchanged

I doubt a bit if replacing R281 with a wire would help as this problem also occurs by touching GND directly. But I will try it nonetheless.

If you need more information or want me to try something different, please don't hesitate to ask!

Best regards,
Ben.

fletch's picture
fletch
Moderator(15)
The ZedBoard engineering

The ZedBoard engineering group has reproduced the issue when hot-plugging earthed equipment (monitor) to the ZED. We are still investigating the possible causes and solutions.

As I work-around, we encourage people to make the necessary connections before turning the board on.
 
Bryan
 
 

thobie's picture
thobie
Junior(0)
Cheap'o power-supply

Got a Zedboard delivered yesterday, and noticed the behavior immediately. Or actually, noticed first the electric shocks the board gives when using ESD grounding strap.

The board was supplied with "Powertron Electronics Corp." power supply, which don't have protective earth connector - just the phase and neutral plugs. However, it uses some kind of EMI/mains filter ( like seen in: http://www.intemc.net/info/UploadPic/2010-2/201026181827903.jpg ) against protective ground, and doesn't isolate the outputs from that virtual ground.

Now, when there's not protective ground connector, the virtual ground which is connected to phase and neutral with caps, starts floating at half of the mains voltage. I was able to measure the ~85V from Zedboard's connector grounds to the protective earth. And you can feel it when touching the board and some grounded metal (for ex. your laptop). *zap* There's just the charge of the capacitors, not continuous current to harm any living.

Now, what do you think the FPGA will like about that ~60-100V when you manage to touch the I/O-pins when grounded - or connect some cable where some of the I/O-pins connect before the shielding/ground? And who's responsible of the broken boards?

If not able to manufacture those power supplies within regulations without the EMI-filters, would be nice to even use then network plug with protective earth, and fully isolate the secondary from mains ground/supply!

Next on the shopping list is a proper linear isolating powersupply for these to not break the board... =F

thobie's picture
thobie
Junior(0)
...

And of course there's some other design flaw related to withstand these ESD / voltage spikes on the grounding of the board, cause it's rebooting every time. Shouldn't do that, but shouldn't either have that shitty power supply which will end just fried boards and people getting the board to reboot every time they connect grounded cable.

Just my 5 cents...

thobie's picture
thobie
Junior(0)
...

Did some research, and the connection between primary and secondary of the supply is actually from a cap between the sides to get the supply through EMI tests (to close a current path for common mode current which flows through the flyback transformers coils because of the capacitance between them vs. switching frequency).

The cap size is about 2nF so a small <1mA is produced, and the voltage was in the range of ~85volts with a 230V supply measured from the outputs of the supply to the safety ground.

pawelcern's picture
pawelcern
Junior(0)
Research...

I solved the problem by shorting GND and SHIELD nets (R281 or C385). The problem doesn't happen anymore, even with original PSU. By the way, can you please say the reason why parts R281 / C385 are there?

sbobrowicz's picture
sbobrowicz
Junior(0)
ZedBoard Reset Issues

When the ZedBoard is "on" and powered from the included supply there is no reference to earth ground (the system is floating). When a device that references/connects its shield pins to earth-ground is attached using a shielded cable (like HDMI), a reference/connection to earth-ground is established and the ZedBoard may be reset as the board ground stabilizes. The only way to solve this is to not introduce a new ground reference while the board is on and operating. If hot-plugging is required for your application, a reference to earth-ground can be established prior to powering on the board by either attaching it to a grounded device, or by using a three-prong power supply that provides a ground reference.

It is also possible for the ZedBoard to reset in response to ESD events, but this is a separate issue from the hot-plugging reset described above. All of the ZedBoard’s cable connectors (HDMI, USB, Ethernet) have their shields connected to a "shield ring" within the PCB that is then connected to board ground through a 1Mohm resistor and ESD-suppression cap. We have observed that replacing the 1Mohm resistor with a shunt decreases the likelihood of an ESD event causing a board reset. If you are experiencing excessive ESD related resets in your environment, then you have the option of shunting this resistor (R281 on Revision C and Revision D) to suppress the problem.

Sam Bobrowicz
Digilent Inc.