I ran into a bug with my DSP (ADSP-2191) system when I tried to transition it from the hardware development environment to have it boot from EEPROM. I have two external memories on my system: a dual-port SRAM (DPRAM) and a standard SRAM.
It turns out that although the SRAM works perfectly immediately after power-up, the DPRAM requires a "bootup/startup" time of ~700 ms before it will work properly. Then it performs flawlessly afterwards.
I'm using the IDT 70V25 (full p/n IDT70V25L25PF). I've probed the /CE, /WR, /RD, address, & data lines on the chip and all of them are asserted properly as soon as the DSP's /RESET goes high. We assembled the PCB ourselves but the fact that the board performs flawlessly after700 ms would indicate that this is not a PCB bug. I've checked a few key pins for strange transient voltages at 700 ms but didn't find any.
Obviously, I could just insert a 1 s delay in my DSP code before I boot the DSP, but I'd rather fix the real problem or else find documentation indicating that a hardware chip doesn't have a ~0 boot-up time, and then I could know how long to make my delay.
Anyone out there have experience with "slow-booting" chips? Any ideas? It took me a week just to find this bug and we're supposed to be going into production ASAP. Thanks, Todd