[sane-devel] Scanmaker 4 i/o errors

[abel deuring]

Steven Hillis wrote:

> [microtek2] scsi_sense_handler: fd=3, sense=0xb768d060 arg=(nil)
> [microtek2] dump_area: SenseBuffer
> [microtek2]      0: f000090000000216 0000000060000000    ........ ....`...
> [microtek2]     16: 00004c616d702066 61696c7572          ..Lamp f ailur
> [microtek2] scsi_sense_handler: SENSE KEY (0x09), ASC (0x60), ASCQ (0x00)
> [microtek2] scsi_sense_handler: info: 'Lamp failure'
> [microtek2] scsi_sense_handler: Lamp failure
> [microtek2] scsi_read_image_status: 'Error during device I/O'
> [microtek2] scsi_wait_for_image: 'Error during device I/O'
> 
> then stops.
> 
> Ok, so, lamp failure, that sounds pretty self explanatory... but the
> preview still works, and the last time I used it in Windows (admittedly
> about a year ago) everything ran fine.
> 
> Now, I'm not so much looking for someone else to solve my problem as
> much as informing me of a couple things.
> 
> First, should the scanner still be capable of a preview scan if the lamp
> is failing? I've opened the thing up and everything *looks* to be
> working correctly (all the lights are still working).
> 
> Second, will it damage the scanner in any way if I remove the failure
> check and force it to scan anyway, assuming the lamp actually is going out?

well, I'm not an expert for Microtek scanners or the microtec2
backend, but around line 3950 in microtec2.c the values of
ms->fastscan and ms->quality are set depending on the value of the
Sane option OPT_PREVIEW, and around line 4560 ms->fastscan and
ms->quality are used to set some bits in the data of the SET WINDOW
command. So it seems that these two parameters, or one of them, tell
the scanner to be more picky regarding the "light quality". Before
starting a scan, most scanners calibrate the lamp and the CCD
sensor, typically by moving the scan head to positions where they
"see" a white and a black area, so that they can measure the CCD
output for these colours.

If this fails, because there are maybe too large deviations in the
signal returned by the CCD over the scan width, or if the CCD signal
for "white" is too low, the scanner's firmware may complain.

So, if scan quality is not that important for you, you could try to
constantly set ms->fastscan and/or ms->quality resp. the
corresponding bits in the SET WINDOW data to the values for the
Preview mode and see, if this helps.

Abel

PS: Some time ago, I had a similar question for the Sharp backend: I
am not sure about all details (only "indirect communication" between
me and the end user), but a scanner complained about failed
calibration in transparency scan mode. I believe that the user
insisted to cover the calibration area with the quite large film to
be scanned. In this case it helped to simply ignore the calibration
error returned by the scanner, similar to what you suggested to do
with your scanner.