Forum Discussion

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

migration from EPCQ to EPCQ-A

Hi,

on a custom board we have successfully used an altera Cyclone V (5CEBA2) with an EPCQ32 (ASx4).

Now we had to migrate from the EPCQ32 to the EPCQ32A.

We used Quartus Prime Lite 17.1.0 and 17.1.1 and programmed the EPCQ32A via the Quartus Programmer + USB blaster + SFL.

We followed the migration guideline AN822:

https://www.altera.com/documentation/tfb1498107381358.html

(https://www.altera.com/documentation/tfb1498107381358.html)

After updating the EPCQ in Quartus from EPCQ32 to EPCQ32A, compilation and download of the .jic file to the EPCQ32A are successful, but then the FPGA fails to load the configuration file at startup (CONF_DONE remains low).

We believe the hardware is fine (never had a problem with the EPCQ32).

(MSEL: 10011, nCE to GND, nSTATUS, nCONFIG, CONF_DONE with 10k pull-up to 3V3, six lines between the EPCQ32A and the FPGA (DATA[3..0], nCS, DCLK)

What we noticed in the map file is the following line:

quad-serial configuration device dummy clock cycle: 12

whereas the AN822 states:

--- Quote Start ---

epcq—the dummy clock is configurable with the non-volatile configuration register (nvcr). when the epcq is used with a cyclone® v, arria® v or stratix® v device, the dummy clock is configured to be 4, 10 or 12, depending on the byte-addressing mode and asx1 or asx4 configuration.

however, in epcq-a devices, the dummy clock is fixed at 8 and 6 for fast read and extended quad input fast read respectively. therefore you must regenerate the programming files, such as .pof, .jic, and .rpd.

--- Quote End ---

Is the dummy clock cycle wrong and is it the cause of our problem?

If this is our problem, how can we fix it? (we did not find any place where the dummy clock can be adjusted)

Or is the problem somewhere else? What else can we check?

Thanks for your help,

Regards

13 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    Hi rbatfmi,

    Out of 25 boards with the EPCQ32A , 4 works well (16%), configuration fails on the others.

    We used the SO8 package (AS x4). We soldered 0805 10pF C0G capacitors between pins 2, 3, 5 and GND (pin 4). i.e. between D0, D1, D2 and GND. After this modification, the configuration works on all boards (tested at room temperature, at approx. -10°C and at approx +50°C). We do not like this solution, but we could not find a better one. It is not clear to us what causes the problem and why these capacitors solve the problem. An hypothesis is that the EPCQ-A might have faster edges than the EPCQ devices and that the capacitors increase the rise/fall times. This, in turn, would improve the quality of the signals at the other end (the FPGA side) of the tracks. Other ideas?

    A reply with more details is currently awaiting approval (due to the attached files). I expect it to be published over the following days.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    Adding 33 Ohm 0805 resistors in series to each data line (between the flash pins and the corresponding pads on the PCB) solved our problem with the EPCQ32A.

    On the same board the EPCQ32 works without those resistors.

    Thus, they are clearly not equivalent regarding the PCB design. The EPCQ-A devices are less forgiving.
  • Irionhead's avatar
    Irionhead
    Icon for New Contributor rankNew Contributor

    Hi everybody,

    is there any new knowledge about this problem?

    I testet Winbond W25Q128JVEIQ flashes and got the same problem. I soldered capacitors (10pf 0805) between the Data lines and ground, on some boards this solution work fine but not at all.

    So i changed to Cypress S25FL128SAGNFI01 with this flash device all boards run without capacitors, but not stable.

    Every 10 times the configuration fails.

    Does anybody know a flash device with WSON-8 (8x6mm) package that runs reliable?