USB 2 and large data transfers

[arhi]

iteadstudio had usb2.0 development board (itdb001) that was great (I have one) that has the same chip as saleae logic analyzer but it looks like they do not have it any more on their site :(


EDIT, here it is: http://blog.iteadstudio.com/usb2-0-deve ... d-itdb001/

might be worth a try, there's bunch of samples on itead site too and board is fairly cheap

demo sources:
http://blog.iteadstudio.com/some-demo-c ... cy7c68013/
http://blog.iteadstudio.com/demo-code-for-cy7c68013-2/

[brian]

I had full speed numbers for AVR 8 and Ti's M3 (LUFA/Ti stacks respectively) It was on the order of MB/s for the Ti as I was indexing entire HDs full of data and had 30 min throughput timings. They may be written down somewhere in my pile of stuff if not I probably have the test code somewhere if people care for the FS numbers (also it was mass storage for the Ti test naturally).

The on chip PHY for the USB 2 HS makes LPC18xx the most interesting micro for me so far. Fewer chips is better for everything pretty much... Better still LPC4300, HS PHY + M4, 200 MHz. Sounds pretty good so far (although the coprocessor M0 seems like it might add misery to programming). I haven't used NXP yet I hope the docs are good.

There is a LUFA clone I think for NXP I think or something like it... I might have to find it.

[brian]

I take it back the LPC4300 sounds like the thing to use. The M0 and M4 are more like just having two chips on the same PCB save they are on the same Si. This means I should in principle be able to use the M0 for USB transfer interface and the M4 for each application. I hope I will be able to setup a generic interface and I/O service program on the M0 and then just write application specific code on the M4. If I can get that to work it would open up a lot of possible uses.

[arhi]

I told you it's worth considering :D ... there's also a large community around it .. some of the interesting sites that you probably already found http://lpc4300.com/

[brian]

I ordered a LPC4330-Xplorer development board (only BGA for LPC4300 currently and doing the work of laying out a board to evaluate the chip isn't worth it to me given my limited free time right now), as time allows (a big if). I will try to get the M0 to do all the dirty USB 2 HS stuff and the M4 to just do something very simple with a datastream. If possible with GCC tools only as I did with Ti's chips. If I get that base worked out nicely it would sort of be like your FTDI + uC solution save one placement on the board rather than 2. Downside is only flashless parts are really in channel right now so another placement for the external flash is semi-required (not as big a deal on a FPGA + uC board which basically also needs that sort of thing).

Everything sounds easy but I'm sure there will be a bit of a learning curve to get this one running smoothly. If and when that happens I will post some throughput numbers and the base code too of course.