[sane-devel] USB - scanner - devel

Jaeger, Gerhard gerhard@gjaeger.de
Sun, 22 Sep 2002 09:12:47 +0200

On Sunday, 22. September 2002 02:28, Yann E. MORIN wrote:
> OK. In dmesg, the device identifies itself as VID/PID:0x4a9/0x220D, that is
> the same as a N670U. So the chip inside ought to be the same (or
> compatible). After some digging, I came up with these results: N676U scans
> at 48 bits, and LM9832 does only 42 bits. So it should be a LM9833. I got
> the datasheet from National Semiconductors, but I doubt it'll be of
> interest...

The only difference is the ADC inside with it's 16bit resolution, this should
be no problem so far...

> Next, dmesg would spit that the device is claimed by no driver. In fact,
> linux USB scanner driver hasn't got this VID/PID pair as a recognized
> device. Patched the driver so that my device is recognized, works OK. Will
> send the PID/VID pair to the maintainer.


> Next, I tried to run the merlin test. Well, it segfaults when searching for
> the scanner : the pointer for the device appears to be a wild pointer, as
> gdb says "Cannot acces memory at address 0x1031". Any clue? (Attached is
> the output of my gdb session.) OTOH, files look to be a bit 'old', as last
> version is dated february `02.

No problems here with that test program! But you should run this one 
as root user.

> I'm currently digging in plustek* and trying hard to understand what's
> goin' on in there. From what I see, adding a descriptor {"VID/PID",
> DCapsDef, HWDef, "name"} for the device ought to be enough.
> What I don't get is how to 'compute' the data to fill the structures
> DCapsDef and HWDef : is it some kind of black magic, found by luck or by
> chance? Or is it data extracted from some datasheet one have only by
> paying?

No black magic at all. You have to know what goes into which register 
at what time, so you'll need the National datasheets and documentations.
The other thing I currently use as "knowledge base" is the merlin

One thing: you won't get far with the current plustek backend even if
you use some rather good DCapsDef and HWDef entries, as the current
version has some problems with the 670U. In the meantime this has been
fixed and I got some picture data out of the device not really usable but
pixels so far ;-)

I'll tell it at this place when the stuff is available...