??? 12/05/08 19:39 Read: times |
#160657 - Not enough information for a diagnosis Responding to: ???'s previous message |
Are you lusing the DLP-USB245M?
How many values are you accumulating in the FIFO? How deep is the FIFO? You're aware that the access time of the FIFO at the target side is 100 ns, right? How are you managing the FIFO depth? How many values do you have to accumulate at a time, and how are they timed? Do all your values fit the FIFO? You realize that the PC processes the USB channel when it wants to, and not when YOU want it to do that. If your pixel count exceeds the FIFO depth, you have to have some way to capture the entire pixel array synchronously with the data source (CCD) and buffer it locally, then transfer it to the FIFO, and only then will it be transferred to the PC. The FIFO depth is greater in one direction than the other, though I don't recall which is 128 bytes and which is 385 bytes. from the DLP-USB245M user manual: TXE# - When high, the FIFO’s 385 byte transmit buffer is full, or busy storing the last byte written. Do not attempt to write data to the transmit buffer when TXE# is high. Your query indicates you're losing "300-400 pixel" which suggests that there may be a FIFO overflow/management problem. In my application, I saved 512 kBytes in local memory, then transferred the data, one FIFOd-full at a time until the local buffer was exhausted. The same procedure was used to transfer data in the other direction. I had someone else write the PC-interface software, so please don't ask me about that. The vendor-supplied driver was used. RE |
Topic | Author | Date |
CCD Reader Project Problem! | 01/01/70 00:00 | |
Faster PC? | 01/01/70 00:00 | |
P4 | 01/01/70 00:00 | |
Re:P4 | 01/01/70 00:00 | |
Any other USB device on the bus? | 01/01/70 00:00 | |
Missing data even for true raw rate requirement | 01/01/70 00:00 | |
Not enough information for a diagnosis | 01/01/70 00:00 | |
Buffer size on the PC device driver![]() | 01/01/70 00:00 |