[Gb-ccb] Single board computers with USB 2.0.

[John Ford]

Randy McCullough writes:

  < snip>

 > John... your parenthetical comment (question) regarding a possible fall
 > back to<br>
 > USB 1.1 with the advent of 4 USB "ports" at the host raises an
 > important point.<br>
 > <br>
 > I've been studying the actual (650 page!) USB 2.0 specification in my
 > spare time<br>
 > over the past few days and have learned some interesting tidbits about
 > both the<br>
 > 2.0 and 1.1 standards...<br>
 > <br>
 > USB 1.1 supports two basic transfer rates... there's "Full Speed" at
 > 12Mbps and<br>
 > "Low Speed" at 1.5Mbps (here, both are expressed in bits (not Bytes)
 > per second).<br>
 > However, the 1.1 standard calls for <u>all</u> transactions to be
 > carried out (factored into)<br>
 > fixed-length "Frames" of 1mS each.   This frame rate is pretty much set
 > in concrete <br>
 > by the standard and carries some implications regarding bus access
 > allocations,<br>
 > issuance of multiple, time-critical command packets, etc.  Since our
 > instrument's<br>
 > current specification calls for a 1mS polling interval, this might
 > create a problem...<br>
 > <br>
 > On the other hand, USB 2.0 operates at a maximum transfer rate of (up
 > to)<br>
 > 480Mbps and utilizes what's referred to as "Micro Frames" which
 > are125uS <br>
 > in length.  It seems to me that this 8X increase in effective frame
 > rate would <br>
 > be
 > highly desirable given our intent to poll the CCB at 1mS intervals...<br>
 > <br>
 > Randy<br>
 > </body>
 > </html>

Or if we had enough buffer space to store several integrations while
it is awaiting transfer, the 1 ms frame rate would work.

I do like the 2.0 solution better.

John