Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
17 years ago

Cyclone 3 problem: Holding nConfig low

I am having a really weird problem with a really simple board. Using an EP3C25E144C8N, active serial configuration mode with an EPCS16, MSEL[2:0] = 010. I can successfully erase and program the configuration device, but the FPGA is a different story. It appears to be pulling nConfig low on its own! I have a 10k pullup to +3.3v, and nothing else on that line! I have even lifted that pin on the part and added an external pullup, to no avail. I have also tried shorting this pin to +3.3v directly, but the FPGA still will not program (I see no activity on the config DCLK, and it is obvious that my code has not been loaded). Also, there seem to be 3 different behaviors of the circuit, which seem to be selected at random on powerup:

1) nConfig gets pulled up by 3.3v as expected, then gets pulled to 0v as soon as the 1.2v core supply turns on. Additionally, all I/O pins seem to be driving ground in this mode.

2) nConfig gets pulled up by 3.3v as in the first case, has a downward spike when the core supply turns on, but then stays at 3.3v for ~25ms, before dropping to 1.8v. (no idea where this is coming from, all IO banks are +3.3v). IO pins seem to be driven at about 1.6v in this mode.

3) nConfig gets pulled up by 3.3v as before, and holds there for 100-200ms, before dropping to 1.8v.

I have tried powering MSEL[1] from both 3.3 and 2.5v. I have tried shorting nConfig to ground until all the supplies are operating. My power supplies are all active within 10ms. I have at least double checked all the power, ground, and configuration pins. This is the second device that I have had problems with...I and my team are all baffled by this...If I have to order another $40 part, I'd at least like to understand why the first two died!! Has anyone seen this before, or have any suggestions? Thanks!

6 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    17 years ago

    I have a simple answer to your verbose description: nConfig is an input pin according to the Cyclone III device handbook (unconditionally, as with all other FPGA). If you observe it pulled low, something is wrong with your hardware.

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    17 years ago

    Thanks, FvM...This is what I feared...but now the question is, what could I have possibly done to the device to cause an input to act as a pulldown!

    Oh, and sorry for being so wordy in my initial post...I figured it was better to provide all the information I had, rather than keep people guessing :)

    I have attached my schematic, if anyone sees an error that would cause this failure mode for the FPGA...please let me know!
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    17 years ago

    Did you electrically connect the exposed pad? Also, there may be a problem with non-monotonous rise of the 3.3V supply (drops after the 1.2 and 2.5V supply have reached their valid levels), see the respective comments in the device manual. If these points are O.K., nConfig may be simply shorted somewhere.

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    17 years ago

    Thanks FvM! Actually, you and Altera support both pointed me in the same direction today...I had not connected the exposed pad! The note on what to do with it was footnote# 7 at the bottom of page 9 on the pinout listing...not the most prominent location for it, and I obviously missed it :-/

    I'm going to pull the part tomorrow and try attaching some tiny wires to the pad to ground it, and verify that this is in fact the issue before I respin the board.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    17 years ago

    You're not the first one to miss this important information, according to the forum discussion. Altera may want to display it in a more obvious way in the future, I think.

    The exposed pad baseplate may be 3 or 4 mils in thickness, enough for carefully milling through the PCB with a handtool without damaging the device, I think.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    17 years ago

    FvM, I was able to successfully mill a 1/16" hole through the PCB and attach a tiny wire to ground that pad...and it works now! :D