Honored Contributor
Honored ContributorHi Andrew,
Can you tell me what kinds of compact flash cards these are? Prior to my posting the software files I ran through some tests using 10-20 types of cards in sizes from 16mb to 1gig on a regression farm. This did reveal a power-up problem, whcih was evident on certain combinations of dev boards & CF cards... ths issue is described in the Nios II errata. To summarize the problem: In "True IDE" mode (as we support), compact flash cards require that the atasel_n input be driven low at power-up. This I/O comes from the FPGA on our dev boards. When the board is powered up, the FPGA can't drive the I/O (FPGA isn't yet configured) and as such the CF card needs to be power-cycled to wake up properly. The CF controller IP & HAL driver do indeed power-cycle the card, however, certain types of CF cards will not power-off when told to do so because of leakage current issues. As a result, to work around the situation, we are reccomending that users tie the atasel_n CF input to ground on their board (or use a strong pull-down resistor). This procedure is detailed in the Nios II 5.0 errata. After performing this modification on my own dev board, the problematic CF cards have worked without any problems, with either the software I posted to Nios forum or the 3rd-party software (eCos, Micrium uC/FS, etc.). However, I am keenly interested in whether there are other CF issues out there. Since this issue is sounding like the same power-up problem, I would reccomend giving this change a try. I have passed the reccomendation for our CF sockets on to our PCB engineers, but unfortunately we cannot modify boards in the field to address this.
