OK, here are the dirty details: I'm stuck with a FTDI USB1.1 solution because there are no (cheap) USB2.0 solutions that can be mounted on a bread-board (one can either get the surface-mount chip and do the PCB/soldering oneself -- which I wasn't prepared to do just yet --, or get an expensive [90 euros or so] module from Germany based on Cypress' E-Z-Usb).
So yeah, I got the USB245M module from http://www.dlpdesign.com/usb/ (it's based on FTDI's USB1.1-capable FT245BM; it has the chip and some minimal logic around it, and you get access to some of the chip's inputs/outputs via the board's legs (that can be plugged into a breadboard). The page claims 8Mbps (1MB/s), which is around the theoretical limit of USB1.1, but I only got 60% of that (could be the USB host on my computer, dunno). Anyway, ideally I would have used an USB2.0 "high-speed"-capable chip for an order of magnitude bettwer performance, but that will have to wait (see reasons above -- moreover, my current USB host isn't USB2.0 capable).
The FTDI is mainly a parallel-to-USB interface (driven from the parallel side, with a small internal FIFO). The Cypress E-Z-Usb (USB2.0 high-speed-capable) has more than one interface (it can interface with a microcontroller as well as with external logic via a "slave FIFO" mode, gives access to all USB channels whereas the FTDI hides all channels but one). On the Cypress I would have used the slave-FIFO for data and the uC interface for control (start/stop, change sampling frequency, etc.). On the FTDI I will be stuck with data only. For now I just made a test circuit with some counters to check throughput and data integrity (actually only 3 out of 8 data lines are connected on the breadboard if you look closely -- I got impatient 😉 ). The next step will be to try to get an FPGA or PLD in to get a state machine to interface with a larger external FIFO (RAM-based). I see no problems with maintaining the 600kHz throughput. When that works I'll release schematics (and maybe even start looking into the Cypress E-Z-Usb variant... but let's do things in order). 😀
V.
