[sane-devel] HP 8270 scanjet support

[m. allan noah]

On 1/25/08, olin.sane.7ia at shivers.mail0.org
<olin.sane.7ia at shivers.mail0.org> wrote:
>    Date: Fri, 25 Jan 2008 17:09:20 -0500
>    From: "m. allan noah" <kitno455 at gmail.com>
>    On Jan 25, 2008 4:36 PM,  <olin.sane.7ia at shivers.mail0.org> wrote:
>
>    it does not immediately hit any of our known chipset tests. that does
>    not mean much however, as we cant test for everything. the next steps
>    are to open the scanner and get some chip info, or perhaps better to
>    take a trace using benoit's usbsnoop on windows, and post that on the
>    web somewhere. perhaps someone here will recognize the protocol.
>
> I am loathe to open the thing up, since I might end up returning it to Newegg.
> It looks like I am stuck using this scanner through a vmware WinXP.

no- if you can get a usb trace under windows, someone might recognise
the protocol.

>    to that end, i would personally buy a refurbished Fujitsu fi-5220C,
>    ($1009 at pcnation.com). Fujitsu has been reasonable about providing
>    documentation to me, and has kept the protocol of their larger
>    machines pretty much the same for years. Something similar can be said
>    for Epson, though they seem to have more new machines with an unknown
>    protocol as of late. HP particularly is quite bad about changing
>    protocol with every single model.
>
> Great, this is a useful bit of intelligence for someone new to the
> scanning-in-linux scene: Fujitsu products are the linux-friendly ones, yes?
> (Perhaps there should be a list on the sane-project.org web pages saying
> which scanners and companies work well with sane.)

well, i dont know that i would sane 'linux-friendly' about Fujitsu, i
just have sources where i can get docs, as opposed to Canon and HP who
have both given me the runaround for months in the past. I am sure
that other sane devels could recommend a company that has given them
some support, i can only speak to the ones i know.

> While I'm writing, I discovered something else I didn't know about sub-$1000
> scanners today. I thought the automatic document-feeders on these flatbed
> scanners would work by feeding the document onto the glass, where it would
> then be scanned in the standard flatbed way, by waving the scanner bar across
> the bed. But that's not what seems to happen. What happens is that the the
> scanner bar sits at the ADF end of the flatbed, and each page of the document
> is fed past the bar in a tight U.

yes- your suggested method would take twice as long, and getting the
paper off the flatbed is hard with variable sized documents, you would
need a huge belt and a long output tray, and the belt would abrade the
glass.

Now, the cheapo ADF mechanism seems
> very unlikely to have the kind of precision registration that the scanner
> bar tracking mechanism would have, so I'm guessing that ADF-scanned documents
> would come out lower quality than documents scanned by laying each sheet down
> on the flatbed by hand.

if you want edge registration, you have to use a flatbed, or you need
to use a scanner with a black background and overscan capability, and
some deskew/cropping software to clean it up.

other than skewing issues, the quality of a good adf will look no
different from the flatbed. just beware of the 'U' being too tight for
the thickness/print of your media.

> Am I right?
>
> If I am, what do you have to pay to get up to the class of machine whose ADF
> works by positioning each page on the flatbed? Is that done?

i've never seen a machine (other than a big copier/scanner combo) that
does this. i prefer the paper path in most adf-only machines, since
the 'U' is often pretty straight. flat-beds with good quality are very
inexpensive. the combination machines just are not as popular, so cost
more. perhaps you would be better off with two machines.

> Is this issue related to the ultrasound-sensor stuff that Fujitsu brags about
> for their scanners, whatever that is?

nope- that is double-feed detection. senses a change in the thickness
of the paper. their less expensive machines use IR instead.

> Is there a place on the web where I could learn more about these issues?

sure- the as-yet unwritten webpage you make which documents your
experiences looking for a machine :)

allan
-- 
"The truth is an offense, but not a sin"