From: Spiro Trikaliotis (trik-news_at_gmx.de)
Date: 2003-07-10 08:54:17
Hello,
On Thu, Jul 10, 2003 at 01:12:27PM +0800, ncoplin@orbeng.com wrote:
> >Yes, but due to the nature of USB I doubt that it will be able to fulfill
> >the timing specs needed by SC and similar transfer software.
>
> Maybe not with USB1.x, but maybe with USB2.0 which is 12Mbps?
>
> A typical LPT port write takes between 1us and 2.5us (from the range of PCs
> I've benchmarked). If a USB converter didnot have too mich overhead (not
> sure what these PIC, etc chips have), it might be in the ball park....
> mighten it?
With USB 1.1, the time is divided into frames of 1 µs. The USB host controller
polls every slave which wants to communicate with it.
On the other hand, the host can send "interrupt polls" which are
prescheduled. In effect, the hosqt can send such an interrupt poll every
1 µs, so it should be able to have a cycle time of 2 µs when communicating
with an USB slave (1 µs for each direction), which should be sufficient for
a serial communication thinking about the fact that the 1541 has a cycle
time of 1 µs for the processor itself.
I think the bigger problem would be the latency involved herein. I'm not
sure if Windows or Linux would be able to handle this.
Spiro.
Message was sent through the cbm-hackers mailing list
Archive generated by hypermail pre-2.1.8.