[Fwd: Stalling and banding problems with Umax Astra 610S in SANE 1.0.1]

David Crooke (dave@dcc.vu)
Sun, 10 Oct 1999 03:14:40 -0500

David Crooke wrote:
> Hey guys, nice toys!
> Sorry if this is a known problem, but I couldn't find a "known bugs" list, and
> an archive troll didn't turn up anything quite the same....
> Configuration (more details http://www.dcc.vu/sane/):
> Umax Astra 610S Firmware v1.3
> Linux 2.2.5-15 (generic Red Hat 6.0)
> Tekram DC390 PCI SCSI-2
> No other devices on SCSI bus
> tmscsim.o generic driver, default config (8Mb/sec transfer)
> ASUS P2B-F m/board
> Single Intel Pentium-II
> SANE 1.0.1, out of box build (./configure ; make ; make install)
> Using dll.conf set to "umax" with the default umax.conf
> I have encountered 2 problems with this setup:
> 1. The scanner stalls in the middle of operation
> ------------------------------------------------
> The scanner stalls at two points during a preview, and three during a scan. This
> locks up the application (e.g. xscanimage) which fails to respond to events,
> such as X exposes. In all cases, sending a SCSI reset to the stalled scanner
> causes it to proceed without error, and the application recovers cleanly.
> In this context, I have been sending resets with this Linux command:
> echo "RESET" >/proc/scsi/tmscsim/0
> In preview mode, it moves the head back for perhaps 200ms, pauses for 3-4
> seconds, then moves it forward for about 1 sec, then the scanner stalls for the
> first time. After the reset, it proceeds to do the full scan, and returns the
> head home, but appears not to deliver the last data packet back to the
> application, and a second reset is required.
> In scan mode, it stalls at the same initial point, then stalls again apparently
> just after the first data block (xscanimage brings up the "Receiving RGB
> data..." green progress bar, which stays at 0), and finally stalls right at the
> end, at the same point as the preview.
> The size of a scan does not seem to affect the stall behaviour - I have tried
> scans from 32Kb to over 30Mb. On full width 300dpi scans, the scanner needs to
> pause, back up and resume (rather like a tape streamer) due to the data volume,
> and this works perfectly.
> Does anyone have any thoughts? I'd be happy to collect debug output and send it
> to a developer if they like; failing which I'm going to dive into the code and
> kludge in these resets for my own use, in which case I will report on progress.
> 2. Image has colour banding
> ---------------------------
> The scanned images have a noticeable colour banding across them. The banding
> seems to be periodic, cycling with the number of pixels scanned, and is not
> correlated to the physical image size. The banding seems to have a certain
> colouration about it, e.g. the bands go CMYCMYCMYCMY.... Here is a sample,
> collected with xscanimage, selecting 300 dpi but otherwise using default
> options.
> http://www.dcc.vu/sane/sample.jpg
> (I didn't want to attach it as it's over 70Kb)
> Is this a hardware fault with the scanner, or a software (e.g. gamma) issue? I
> am not in a position to (easily) try the thing under Windows or MacOS. Any
> advice welcome.
> 3. Minor thing
> --------------
> Minor additional question - is there a way to configure what /dev/ devices SANE
> tries? I have /dev/scanner linked to /dev/sga, and xscanimage always brings up a
> pop-up asking me to choose one of the two. I'd like to just have it use one
> configured.
> Best wishes
> Dave
> --
> David Crooke, Austin TX, USA. +1 (512) 795 0805
> Bibendo ergo sum.

David Crooke, Austin TX, USA. +1 (512) 795 0805
Bibendo ergo sum.

Source code, list archive, and docs: http://www.mostang.com/sane/
To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com