[sane-devel] Trouble with Primax 4800 Direct

[Gerhard Jaeger]

Hi,

On Sunday 23 October 2005 17:50, Daniel Klär wrote:
> Hello,
> 
> > as the pround owner of this masterpiece of scanning technology
> > ;-) I'd like to report some problems I have with it and the sane
> > backend plustek_pp. The problems are hard to desribe since the
> > effect is different every time I try. As a start the color mode
> > scans are never correct. Red, green and blue seem to be shifted
> > against each other. The shift is the larger the higher the
> > resolution is. Sometimes I can scan in Color with 150 dpi, but
> > normally either the driver tries to drive the lamp outside the
> > scanner or xsane crashes (e.g.after preview scan) or the scanner
> > simply doesn't scan. After I switched the PC and Scanner off
> > completely (unplug power supply) with some luck I can do some
> > simple scans.
> 
> That's what I started to write last weekend, when I was frustrated
> by unreproducable errors. I was using xsane as root and the sane
> version is 1.0.15 (as shipped with Debian Sarge). Today morning I
> had some time to investigate the problem a bit more detailed using
> scanimage as root.
> After some hours of testing, reading debug output and rebooting I
> found the mistake, that caused a lot of problems. In
> /etc/sane.d/plustek_pp.conf I got the following lines (some lines
> removed):
> > [direct]
> > device 0x378
> > option mov 2 
> > [direct]
> > device parport0
> After changing into
> > [direct]
> > device parport0
> > option mov 2
> a lot of things started to function, at least for the root account.

well, I might should comment out the first section...

> # scanimage -L
> |device `plustek_pp:parport0' is a Plustek Primax 4800 Direct
> |parallel port flatbed scanner
> |device `net:localhost:plustek_pp:parport0' is a Plustek Primax 4800
> |Direct parallel port flatbed scanner
> Looks good and saned is running too.
> 
> # scanimage --help -d plustek_pp:parport0
> The output contains the following lines:
> |  --mode Lineart|Halftone|Gray|Color [Color]
> |  --resolution 50..300dpi [50]
> |  --x 0..215mm [126]
> |  --y 0..297mm [76.21]
> If not specified otherwise the scanner scans with the settings in
> brackets. I can scan with other settings, but I have to specify
> always all the four parameters. Xscanimage also shows these
> preferences and crashes immediately (simply disappears from screen)
> when I try to change it.
> I thought the values in brackets are updated, when a scan with other
> settings is performed? How can I set the defaults?

No way so far. I'm somewhat surprised, that Xscanimage will crash - need to
test that.

> A scan like that
> # scanimage -v -d plustek_pp:parport0 -x 210mm -y 148mm --mode Color
> # --resolution 150dpi > test.pnm
> usually works with several values for -x,-y and -mode , but with
> --resolution 300dpi the lamp tries to leave to scanner with some
> ugly noise. Well, that's a constrain I can accept.

Those scanner always do this ugly noise during calibration :(

> If I use the device "net:localhost:plustek_pp:parport0" instead the
> colormode doesn't work properly. It scans and I can recognize the
> image, but the colors are totally wrong. The proportions of the
> image are o.k.
> It becomes worse when I try the same with a non-root account. The
> color problem is the same, but additionally the image is jolted. If
> I use Gray instead of Color (as normal user) it becomes crazy: The
> left half and the right half of the image are obviously exchanged.

That's because you're loosing data. They ger lost on their way to 
the application! It's been explaind a thousand times in the meantime:
- These old devices have more or less no buffers which are able to 
  keep data while the driver could not fetch them. Also no backtracking
  is possible. That's why the driver in general should run in kernelspace,
  polling the data and refuse the OS from scheduling - what we do not want to.

> Any idea what I might have configured wrong? If scanimage produces
> different output via saned, it isn't an issue of the backend???

Other timing conditions - no way for you to fix that. 

> Finally there's still the problem that after one or two hours system
> uptime the scanner stops working at all. Scanimage -V or -L still
> works, everything else fails. After shutdown, power off and boot
> (reboot alone doesn't help) it works again.

This is quite strange. What's the debug-output like, when this happens?

> Since the scanner is very old and the changelog of the later version
> of plustek_pp doesn't contain any notice concerning "Primax", I
> think/believe the error is not caused by the outdated version I
> use. Using the current Knoppix-DVD 4.02 scanimage -L puts out a
> blinking cursor, that's all. I can (as far this is possible) exclude
> a hardware defect, because running the same PC with Win2k the
> scanner runs smoothly over several hours in all supported modes and
> resolutions.

Are you saying Win2K? Are you shure? It thought there ain't any drivers
available for this OS, because the driver will completely freeze the system
during scans!

Ciao,
Gerhard