| ??? 03/09/06 15:31 Read: times |
#111737 - YES YES YES Responding to: ???'s previous message |
Richard Erlacher said:
Access to USB from inside the PC is done at the PC's rate, and not at the rate needed for the data sampling. If you use that USB/FIFO module, it allows a few samples be buffered in the FIFO, but, ... and it's a critical "but," ... if you need to take more samples than what the FIFO will hold, there's no guarantee that the PC will grab the data in time to avoid overflowing the FIFO. YES. FTDI claims 1MByte/s transfer rate, so I'd say it's something sustainable. I would indicate the FIFO overflow on the latch. Richard said:
That's why I say one should use a local RAM buffer. You see, one has to be sampling before the trigger occurs if one intends to acquire the history of the "blessed event." Often, it's useful to have history of the events leading up to the trigger condition, as well as what happens immediately afterward. If you're sampling signal states, a few (< 32K) samples are usually adequate. If you're capturing serial data or looking at the data from an external source, it's often necessary to capture LOTS of it and use the computer to search it for sequences of data afterward. YES. I would use the PC as the buffer and forget about the trigger. It can capture megabytes of data and then look up the trigger event in them. Brute force. I won't say it's too god. I say it's simple and usable for slow events (such as serial port, I2C or similar). And if there would be a high-speed equivalent of the FTDI245, I'd say it might become a usable piece of hardware at an extremely low cost, parts number and dimensions. JW |
