hp3300backend-devel — Development of SANE backend for HP ScanJet 3300/3400/4300 scanners

[Hp3300backend-devel] HP3300/HP3400/HP4300 work-in-progress

[Bertrik S. <be...@zo...>]

Hi all,

On http://sourceforge.net/projects/hp3300backend we are working
on a SANE backend for the Hewlett-Packard ScanJet 3300C / 3400C and
4300C models of scanners.

We already have a prototype SANE backend for these scanners,
although only the HP3300 is properly supported (the HP3400C and
HP4300C have their vertical resolution fixed at 300 lpi).

Could this project be added as work-in-progress in the PROJECTS file?

Kinds regards,
Bertrik Sikken

[Hp3300backend-devel] README file

[Ricardo M. <the...@so...>]

I'm working on a very detailed README file (it's about 40% complete - 
started 3 days ago). The testtool.txt file in the testtool package 
already states that there is a problem when the module is loaded.

Please, keep in touch (and tell us if it worked).

Ricardo Martins

Re: [Hp3300backend-devel] I can't make it work

[Ricardo M. <the...@so...>]

NetNoise wrote:

> I try the new code whit my hp3300c , but I cannot make it work ; it gave me
> this error message:
> 
> 
> [root@localhost scan20011216]# ./hp3300 lamp
> bus 001 dev 002 : 0x03F0-0x0205
> Matching USB device 0x03F0-0x0205 ... found HP3300C
> Unable to claim interface
> bus 001 dev 001 : 0x0000-0x0000
> Matching USB device 0x0000-0x0000 ... nothing found
> LibUsbExit: nothing to close
> [root@localhost scan20011216]#
> 
> Any one knows why?
> 


I'm terribly sorry for the delay (I spent this afternoon rehearsing with 
my band).

This is probably caused by the fact that libusb doesn't work very well 
when the kernel module scanner.o is loaded.

To see if the module is loaded try this:

[shell]# lsmod

If there is a module named scanner, unload it like this

[shell]# rmmod scanner


Check if it was unloaded (by doing lsmod again).
If it was unloaded try using the testtool.
If the module is still loaded, do rmmod scanner until it is unloaded.

That should do the trick.

NOTE: Always make sure the scanner module is unloaded (a good method is 
by not even compiling it in the kernel).

Ricardo Martins

[Hp3300backend-devel] I can't make it work

[NetNoise <net...@ti...>]

I try the new code whit my hp3300c , but I cannot make it work ; it gave me
this error message:


[root@localhost scan20011216]# ./hp3300 lamp
bus 001 dev 002 : 0x03F0-0x0205
Matching USB device 0x03F0-0x0205 ... found HP3300C
Unable to claim interface
bus 001 dev 001 : 0x0000-0x0000
Matching USB device 0x0000-0x0000 ... nothing found
LibUsbExit: nothing to close
[root@localhost scan20011216]#

Any one knows why?

NetNoise

Re: [Hp3300backend-devel] My 3400c experiences

[Bertrik S. <be...@zo...>]

Hi Richard,

Thanks for the info.

Richard Varney wrote:
>         Thank you Bertrick for your hard work, its the first time I've ever had any
> life out of my 3400c in Linux :-).
> 
> My experiences with the test tool are as follows:
> 
> 1) Trying to use libusb didn't work, the only message I got was "LibUsbExit:
> nothing to close"

This probably means that USB was already claimed by the usbscanner
driver, but you figured it out OK.

<snip 2 - 4>

> 5) For calibibration I first used "./hp3300 calib2" as recommended in the
> readme file:
> Creating calibration file
> _iBytesPerLine = 3975
> _iLinesPerCircBuf = 3
> Allocated 11925 bytes for circular buffer
> _iLinesPerXferBuf = 15
> Xfer block size = 59625
> buffer level =  18, <reading 59625 bytes>, buffer level =  15
> Black level (6,6,7), White level (94,67,58)
> 
> When attempting a test scan with "./hp3300 scan 1325 1700 test.pnm" the
> colours were really messed up.

On second thought the 'calib2' option does not work well on
the HP3400/4300; the black calibration rectangle inside the scanner
used for determining the black level is not scanned in because
of the vertical resolution issue.
The current calibration routines still need a lot of work,
although they work surprisingly well given their simplicity.

> 6) I then tryed "./hp3300 calib" which yielded:
> Creating calibration file
> _iBytesPerLine = 3975
> _iLinesPerCircBuf = 6
> Allocated 23850 bytes for circular buffer
> _iLinesPerXferBuf = 15
> Xfer block size = 59625
> buffer level =  24, <reading 59625 bytes>, buffer level =  21
> Black level 0 0 0, White level 243 246 201
> 
> This gave a much better scan result. Although the perspective was all messed
> up as expected, the colours looked fine.

This looks mostly OK, although the white level is a bit high,
suggesting that the scanner might be overexposed during calibration.
Actually a complete calibration needs two cycles: one to roughly
set the gain and offset of the analog electronics and another
cycle to finetune the 'digital' gain and offset.
Currently the analog gain and offset is hardcoded and seems a bit
too high for your scanner.

> If there is any further testing you would like me to do, please let me know,

Your help is welcome with the vertical resolution issue.

Bertrik

[Hp3300backend-devel] Re: HP3400 using EPP/libieee1284

[Bertrik S. <be...@zo...>]

Bob Jones wrote:
> I just gave Bertrik's new update a run on my 3400, and it scans nicely using
> the USB interface.

Thanks, nice to know it still works ... :)

> On a lark, I decided to give it a try using the libieee1284 interface since
> the Win-drivers work OK in Win4Lin, but now it refuses to compile.  I did
> some checking in the ieee1284.h and hp3300c_xfer.c and there seems to be a
> structure mismatch.  In ieee1284.h, the external functions ieee_epp_* have
> the structure (struct parport *port, int flags, const char *buffers, size_t
> len). In hp3300c_xfer.c, these same functions have the structure, (port,
> buffer data, len 1).
> 
> Has anybody gotten the ieee1284 interface for testtool to compile and work?
> If so, what version of libieee1284 and were any mods needed in hp3300c_xfer?

I have not received any reports of anybody who tried it.
I own a HP3300 (which has no parallel port connection) myself so
I have also not tested it myself.

When I wrote the ieee1284 support, I used the very first version of
it (0.0.0) and it compiled OK. It is very possible that the
interface of libieee1284 has changed since version 0.0.0 .
The biggest difference seems to be the 'flags' parameter.

If you would like to look into this, you're welcome ofcourse.

Bertrik

Re: [Hp3300backend-devel] HP3400/4300 vertical resolution issue test

[Bertrik S. <be...@zo...>]

I just released a new source zip at sourceforge for this test,
because it may not be trivial to get the latest version from CVS.

To do the tests, you should change line 433 of hp3300c.c:
    _ConvertMotorTable(abData0000, abMotor, sizeof(abData0000), 300);
 and line 438 of hp3300c.c:
    _ConvertMotorTable(abData0400, abMotor, sizeof(abData0400), 300);

I have strong suspicions that table abData0000 contains some kind
of motor profile for moving the scanner head when it is not
scanning (when moving back after a scan for example).
Table abData0400 probably contains a motor profile for moving the
scanner head while it is scanning. The first value of this table
is also equal to the number that I use to set a resolution of
300 lpi on my HP3300.

During initial testing it was discovered that setting both tables
to double their default value resulted in a scan at 600 lpi, which
is also double the default resolution of 300 dpi.

Please try to change the number 300 into 150 or 600, recompile and
do a small scan using (for example):
'./hp3300 scan 300 300 test.pnm'
Do not try any other numbers than 150, 300, 600 yet.
Then check the scanned image to see what the aspect ratio of the
image is. A normal aspect ratio of 1 vertical : 1 horizontal
means 150 lpi, a ratio of 2 vertical : 1 horizontal is 300 dpi
and a ratio of 4 vertical : 1 horizontal is 600 dpi, etc.
Also during scanning and during other movement of the scanner head
try to hear if the scanner makes a different sound or shows any
other odd behaviour.

I propose to number the test cases like this:
1) line 433: number = 300, line 438: number = 300 -> 300 dpi scan
2) line 433: number = 300, line 438: number = 600
3) line 433: number = 600, line 438: number = 300
4) line 433: number = 600, line 438: number = 600 -> 600 dpi scan
5) line 433: number = 300, line 438: number = 150 -> 150 dpi scan?
6) line 433: number = 150, line 438: number = 300
7) line 433: number = 150, line 438: number = 150
8) line 433: number = 600, line 438: number = 150
9) line 433: number = 150, line 438: number = 600
Roughly in order of safety (1 = safest test).

It may be possible that the scanner makes a high pitched noise
(because the stepper motor may be stepped too quickly),
especially when using the number 150 in lines 433 or 438.
In that case, I recommend to quickly turn off the scanner.

So please keep in mind that if you run these tests, you are doing
so at your own risk. If you think it is too risky, then don't
do it!

Good luck testing,
Bertrik

[Hp3300backend-devel] HP3400/4300 vertical resolution issue test

[Bertrik S. <be...@zo...>]

Hi all,

As a lot HP3400/4300 owners have already found out, there is an issue
with the vertical resolution of scans done with these scanners in the
current version of the testtool. While a normal scan is done at
150 dpi both vertically and horizontally on the HP3300, the vertical
resolution is fixed at 300 dpi for the HP3400 and HP4300.

(The HP3400 and HP4300 have a slightly different way of setting the
vertical resolution compared to the HP3300. Instead of a single
value to set the vertical resolution, they have two tables
containing exponential decaying and exponential increasing
series.)

I just committed a new version of the sources to CVS for some
tests to resolve this vertical resolution issue.
The new version will by default still use 300 dpi (so as not to
surprise unsuspecting users), but by uncommenting a few lines,
the test will be enabled.

If you have a HP3400 or HP4300 and have already succeeded in
making it scan using the current version of the testtool and you
like to help testing, just let me know and I'll send some more
detailed instructions for this test.

Bertrik

[Hp3300backend-devel] My 3400c experiences

[Richard V. <ri...@va...>]

Hello All,

	Thank you Bertrick for your hard work, its the first time I've ever had any 
life out of my 3400c in Linux :-).

My experiences with the test tool are as follows:

1) Trying to use libusb didn't work, the only message I got was "LibUsbExit: 
nothing to close"

2) Using the scanner driver worked, "./hp3300" yielded:
VendorId 0x03F0, ProductId 0x0405
Matching USB device 0x03F0-0x0405 ... found HP3400C

3) "./hp3300 dump" yielded:
0000 00 00 00 48 07 00 40 02
0008 00 00 00 00 00 00 00 00
0010 00 00 00 00 00 00 00 00
0018 00 00 00 00 00 00 00 00
0020 00 00 00 00 6A DE DE 22
0028 01 03 79 DE DE DE 00 DE
0030 DE DE DE DE DE DE DE DE
0038 3B 0C 43 01 01 01 00 DE

4) "./hp3300 lamp" did turn the lamp on and off :-)

5) For calibibration I first used "./hp3300 calib2" as recommended in the 
readme file:
Creating calibration file
_iBytesPerLine = 3975
_iLinesPerCircBuf = 3
Allocated 11925 bytes for circular buffer
_iLinesPerXferBuf = 15
Xfer block size = 59625
buffer level =  18, <reading 59625 bytes>, buffer level =  15
Black level (6,6,7), White level (94,67,58)

When attempting a test scan with "./hp3300 scan 1325 1700 test.pnm" the 
colours were really messed up.

6) I then tryed "./hp3300 calib" which yielded:
Creating calibration file
_iBytesPerLine = 3975
_iLinesPerCircBuf = 6
Allocated 23850 bytes for circular buffer
_iLinesPerXferBuf = 15
Xfer block size = 59625
buffer level =  24, <reading 59625 bytes>, buffer level =  21
Black level 0 0 0, White level 243 246 201

This gave a much better scan result. Although the perspective was all messed 
up as expected, the colours looked fine.

If there is any further testing you would like me to do, please let me know,

Best Regards,

Richard.

Re: [Hp3300backend-devel] compile probs

[Bertrik S. <be...@zo...>]

Matthias Hillmann wrote:
> 
> Am Mittwoch 28 November 2001 20:30 schrieb Bob Jones:
> 
> > I do have one small
> > problem/question:  I've gotten libusb and tried to compile the testtools
> > but it keeps giving errors with usb.h in hp3300c_xfer during make.
> 
> > hp3300c_xfer.c:47:17: usb.h: No such file or directory
> > make: *** [hp3300c_xfer.0] Error 1
> 
> > Anyone know what's going on
> > and how to correct this problem?
> 
> Hi Bob,
> 
> I have no idea why, but since there are only two occurences of 'usb.h' in
> hp3300_xfer.c, I just changed the second one to 'linux/usb.h' (same as the
> first one) and got a scan.

Don't you get compiler warnings (like "no prototype defined" or
something similar) if you do that?

> I am using /dev/usbscanner and the kernel modules.
> 
> Hope, that helps you. I am already working with my SJ3400C. Unfortunately,
> I'm not good in colours. My scans seem to be brighter or have more contrast
> or whatever towards the end of the scan-area, so I put smaller pics there and
> scan a4.

Try typing ./hp3300 lamp before scanning and then wait a minute.
It could be that with your scan the lamp had not warmed up completely.
As the scanner was scanning the lamp was warming up, so that the
bottom of the scan finally looked brighter.

Bertrik

Re: [Hp3300backend-devel] compile probs

[Matthias H. <ma...@we...>]

Am Mittwoch 28 November 2001 20:30 schrieb Bob Jones:

> I do have one small
> problem/question:  I've gotten libusb and tried to compile the testtools
> but it keeps giving errors with usb.h in hp3300c_xfer during make.  

> hp3300c_xfer.c:47:17: usb.h: No such file or directory 
> make: *** [hp3300c_xfer.0] Error 1 

> Anyone know what's going on
> and how to correct this problem? 

Hi Bob, 

I have no idea why, but since there are only two occurences of 'usb.h' in 
hp3300_xfer.c, I just changed the second one to 'linux/usb.h' (same as the 
first one) and got a scan.

I am using /dev/usbscanner and the kernel modules.

Hope, that helps you. I am already working with my SJ3400C. Unfortunately, 
I'm not good in colours. My scans seem to be brighter or have more contrast 
or whatever towards the end of the scan-area, so I put smaller pics there and 
scan a4.

Greetings, Matthias Hillmann

[Hp3300backend-devel] compile probs

[Bob J. <bob...@ya...>]

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <title></title>
</head>
<body>
Hi,<br>
<br>
I just posted the following on the sourceforge forum:<br>
<br>
--------------<br>
Greetings, <br>
<br>
I'm not really a code hacker, but having a SJ3400C, I would be most interested
in helping as much as possible.  I do have one small problem/question:  I've
gotten libusb and tried to compile the testtools but it keeps giving errors
with usb.h in hp3300c_xfer during make.   <br>
<br>
<tt> hp3300c_xfer.c:47:17: usb.h: No such file or directory <br>
make: *** [hp3300c_xfer.0] Error 1 <br>
</tt><font face="Helvetica, foo, sans-serif">Anyone know what's going on
and how to correct this problem? </font><br>
<br>
<br>
I'm running RedHat 7.2, the latest libusb from sourceforge, and 2.4.14 kernel
(patched for Win4Lin and ext3) with usb, usb devfs, and usbscanner support
enabled (scanner is compiled as a module).  When I load the scanner module
and run sane-find-scanner (sane v. 1.0.6), it finds the scanner fine, but
without a backend that supports it, I'm out of luck. <br>
<br>
FWIW - the scanner works in Win4Lin in ECP mode, but in a 64Mb virtual machine,
its excruciatingly slow - hence the desire to get it working in Linux natively.
<br>
<br>
Thanks in advance. <br>
Bob J                        <br>
</body>
</html>


_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com

[Hp3300backend-devel] hp4300c

[Jacques C. <Jac...@wa...>]

Hello,
i own a hp4300c scanner.
Thanks to the Bertrik's works, my scanner showed me something
under Linux.
To get a non-streched image, i modified all the values
of Params.iDpi and Params.iLpi to 300.
And i can get a image doing hp3300 scan 2480 3425 test.pnm (not more !).
The only problem which remains is that the scanning is
beginning "too early".
I get on the left and on the top bands which are not useful,
being out of the page. I tried several modifications without success.
Somebody has an idea ?
Thank you and best regards.

Jim
------
Jac...@wa... f6cwo