[sane-devel] Traffic analysis - CanoScan D1250U2

Jens Herrmann sane at bikelab.org
Sun Aug 19 21:51:25 UTC 2007

I have a CanoScan D1250U2 for some years know, a model which is still
unsupported although on the market since 2002. One of the reasons for
this, I suppose is that there is absolutely no support from Canon and
everything has to be guessed from the USB traffic of the Windows driver.
I did log some traffic of my CanoScan D1250U2 with UsbSnoop, recently
and would like you to give some comments on it.

Shortly after the initial descriptors there is a huge number of 0x8000
and sometimes 0x2000 bulk transfers in the log which carry the same
pattern of data again and again. Some of them are filled with 0x00
entirely, some begin with 0x00e2, change to 0x0022 and later to 0x0002.
Others consist of 32 Byte groups with values ascending, all absolutely
regular without any exeptions. Does anyone have an idea of what this is

Then at 1/3 of the log a different sort of data appears which looks like

00002500: ce cb cc cd ca d8 c8 cb d0 cb c8 d4 c8 c7 d6 cc
00002510: c5 d0 cc c6 cf ca cb cc c7 d3 cf c9 cc d8 d3 c9
00002520: d7 cf c9 d3 ca cc cf ca c8 d2 d3 c9 d5 cf cf ce
00002530: cf cf d4 d1 d0 d1 cf c6 d6 e0 cc cd ff d0 d1 ce
00002540: cd cf d5 ce d6 ab cd d5 d3 cd d5 d2 ca d6 d1 d2
00002550: d1 cc cd d5 cf cb d6 cf d1 d0 d0 d3 cd d5 ce cf

The average values go up and down here, so I assume this is for
calibration. Also the values of address + n*3 correlate pretty closely,
so I assume they represent a color each.
My question here is: Which algorithm could be used here to determine if
calibration can be finished or not?


p.s It highly likely that the controller used in the D1250U2 is the
EZ-USB FX2 from Cypress Semiconductor. The D1250U2 was announced in
December 2001 and Cypress had the first USB 2.0 controller available in
August 2001
This would also be consistent with the package on the board
(http://jens-herrmann.gmxhome.de/IC2.JPG). The EZ-USB FX2 is available
as a 128 TQFP.

More information about the sane-devel mailing list