Strictly necessary cookies are on by default and cannot be turned off. Functional, Performance and Tracking/targeting/sharing cookies can be turned on below based on your preferences (this banner will remain available for you to accept cookies). You may change your cookie settings by deleting cookies from your browser. Then this banner will appear again. You can learn more details about cookies HERE.
Strictly necessary (always on)
Functional, Performance and Tracking/targeting/sharing (default off)
I have a small home-made board with a CY7C68014A-56PVXC on it. It plugs into my Digilent Spartan-3 FPGA board. I configure EP2 as described in the TRM (fig.9.28) like this: CPUCS = bmCLKSPD1; // 48MHz SYNCDELAY(); IFCONFIG = 0xA3; // drive IFCLK with internal 30MHz clock, use synchronous FIFOs. SYNCDELAY(); REVCTL = 0x03; SYNCDELAY(); EP2CFG = 0xA2; SYNCDELAY(); FIFORESET = 0x80; SYNCDELAY(); FIFORESET = 0x82; SYNCDELAY(); FIFORESET = 0x00; SYNCDELAY(); OUTPKTEND = 0x82; SYNCDELAY(); OUTPKTEND = 0x82; SYNCDELAY(); EP2FIFOCFG = 0x10; SYNCDELAY(); The FPGA drives FIFOADR0 and FIFOADR1 low, and asserts SLRD and SLOE always (i.e it's always ready to accept data when data is available). I then tried a number of scenarios: 1) FPGA counts IFCLK rising edges for which FLAGC says "not empty". If the host sends 1024 bytes, the FPGA shows a count of 1024. So far so good. 2) FPGA calculates a 16-bit checksum of incoming data, using the same timing as in (1). If the host sends 1024 bytes, the FPGA shows a checksum consistent with it receiving 1024 bytes, comprising: 256 copies of byte 0 256 copies of byte 1 256 copies of byte 512 256 copies of byte 513 3) FPGA counts transitions (i.e where the current byte differs from the last), using the same timing as in (1). If the host sends 1024 bytes, all zero except buf=1, buf = buf = 2, buf = 3, the FPGA displays 0x3FF. So it looks like for each 512 byte chunk, the FX2LP chip is sending the correct number of bytes to the FPGA, but it is just sending byte 0 alternating with byte 1, and throwing away everything else. It's as if the 9-bit offset which the slave FIFO uses to read from the endpoint buffer is being masked with 0x0001 somehow so rather than incrementing properly it just goes 0,1,0,1,0,1,0,1. The result is that irrespective of the number of 512-byte chunks sent, the checksum from (2) above comes out as 256 times the sum of the first pair of bytes in each chunk. So my question is, am I merely doing something silly (missing register initialisations?) or have I damaged my FX2LP in some way? I tried the exact same experiments with EP6, with exactly the same results. The other functions of the FX2LP which I have tested (e.g custom EP0 commands, EEPROM & RAM loads) appear to be working fine. Any help gratefully received! - Chris Edit: PS I can correctly read back USB bulk transfers to the FX2LP, so I suspect it's something wrong with my slave FIFO config.