Forum Discussion

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

SSN Dependent upon individual device?

First of all, thanks for all of your previous help. Without it, I wouldn't be where I am with this, my first FPGA project.

I have a question about SSN. I've got a design which hopes to end up with each EP3C16Q240 switching 128 channels of 1MHz PWM. On my PCB I have 12 copies of the exact same circuit design, and of the first 6 that I've soldered, 4 seem to work perfectly with all 128 channels switching, while 2 did not. Of the two that didn't one unfortunately burned up before I could change the configuration, but the remaining one is working fine now with only one channel switching.

My question is simply to ask if this seems normal, that the SSN behavior is so dependent upon the individual device? And if so (when I can actually get more chips), can I get by with simply testing devices until I find a batch that are capable of handling the number of SSO (sort of like finding a tolerance band of resistors or a batch of CPUs that can handle overclocking)?

In the longer run I plan to redesign the board when I have more time to use BGAs which as I understand will be able to handle the SSO issues more easily, but in the meantime I have two PCBs which I'd like to get working.

Thanks for any ideas or thoughts.

-JNS.

2 Replies

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

    I don't know if the reason for different behaviour you experienced is actually in FPGA part variations or more in electrical differences between "identical" blocks, e. g. regarding clock distribution or supply decoupling.

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

    I wouldn't rule anything out, but everything is really as close to identical as I can imagine. Each chip has its own set of linear voltage regulators, decoupling caps, and oscillator and the layout of each block is copied and pasted from the original. Even the length of the supply feeds of the regulators to the main power bus is identical. That bus is a .5" wide trace so I doubt being a few inches nearer to or farther from the source on that would be significant. And to that end, actually the chip farthest from the main source has been one of the reliable ones.

    I've swapped out enough regulators to pretty much rule out any problems there. If I had a few extra chips things would be much easier to test. For future reference, is this sort of stock issue typical of Altera products?

    Thanks,

    -JNS.