Great, that's a little better.

Theoretically, the maximum for HID is 64K bytes/sec. But I've never been able to actually get the full bandwidth. With 340 both ways that's 680 pkts/sec or about 43.5K, getting closer.

> ... what i understood with the 10ms delay is that i have 10ms maximum to pass a usbservice ( to avoid a disconnection)

Yes, that's true. With 1ms polling, you have to Service USB more often. But, if the USBSERVICE is being handled with USB Interrupts then that never becomes a problem, because it will be serviced whenever it's needed, instead of whenever the program gets around to it. This can also improve the overall speed by servicing things faster.

Here's a little program I wrote to monitor the speed of USB transfers.

http://www.DarrelTaylor.com/files/HIDmonitor.exe

It can only see what's coming from the PIC. Some day I'll figure out how to intercept the transmit side. But it does work well as it is. There's no installation required, just a single .exe file. Place it anywhere you want on the harddrive.

If you set-up the PIC to transmit packets as fast as possible without waiting to receive one from the PC, it can isolate which side of things may be slowing it down.

Let me know if you want to switch to USB Interrupts.