Epson 4870 ICE misalignment (infrared channel offset)

[Jan]

Yesterday my Epson 4870 stopped removing dust. I use Epson Scan with
ICE turned on. I was scanning 35mm slides.

This happened once before (end of june 2004 I participated in a thread
in this newsgroup entitled "Vuescan 8.01, Epson 4870, RGBI infrared
channel offset"). My problem then disappeared after having cleaned
with more care than usual both the bottom and top glass plates, and
especially the calibration area. Since then I did another few hundreds
of scans with ICE working fine.

So yesterday I again cleaned "with more care than usual" the
calibration area... but this time the problem persists. I looked into
the calibration area very closely and no dust or scratches are
present. It all looks OK.

I then made a test scan with Vuescan (Vuescan allows to actually
visualise the infrared scan) and saw that the infrared channel detects
the dust particles with a vertical offset of less than a millimeter (I
don't express the distance in pixels, since the offset seems to remain
the same for the different scan resolutions I tried).
The offset seems to be the same whether scanning with Vuescan or with
Epson Scan. Epson Scan doesn't show the infrared channel, but upon
looking verly closely it was visible that some corrections had taken
place with a little offset from the actual dust particles.
This excludes a software problem - it also proves that the IR lamp and
sensors are OK.

I took off the upper scanner lid to examine the upper glass plate
calibration area again. The next scan surprised me as the offset had
now become a horizontal offset ?!?!
Fiddling and testing went on, and a bit later the offset became a
vertical offset again...
I managed to reduce the offset a bit when I lifted one side of the
upper lid about one millimeter during the scan, but I'm not sure if I
can consistently reproduce this. During all my experiments I even got
a test scan with no offset, but the next scan was bad again.

Other things I tried without any effect :
- uninstall and reinstall the Epson Scan software (I was hoping that
some parameter corruption could have caused an offset)
- scanning at different resolutions (the physical distance seems to
remain the same between scanning resolutions - though the physical
distance was not always the same throughout my different tests)
- instead of putting the slide holder in the upper right corner, I put
it in the upper left or lower left corner. No effect either.
- scanning without the slide holder (just put the framed slides on the
glass plates and removed the holder before scanning)

Anybody experienced the same problem ?
Any suggestions before I take my scanner in for repair ?
Thanks for reading me, even more thanks for your suggestions.

Jan

 

[Jan]

Further testing showed the following :

- the offset is now always horizontal again (yesterday's last tests it
was vertical)
- the offset is significantly bigger towards the left and right edges
of the scanner, whilst in right the middle (vertical axis from top to
bottom) all is fine.

Still weird, but there seems to be some kind of logic in it. The IR
records the dust spots farther away from the vertical center axis than
they really are.

All testing done with Vuescan 8.0.6, but as already mentioned the
problem is the same with Epson Scan.

Jan

 

[-]

As you know, the scanner makes two scans. I am guessing that when the
scanner resets itself to start the second scan, it does not return to the
exact starting point? That would explain the offset

Doug

 

[Kennedy McEwen]

Further testing showed the following :

- the offset is now always horizontal again (yesterday's last tests it
was vertical)

What do you mean by horizontal and vertical in this context? Not
everyone has the scanner mounted the same way on their desk. Is the
offset along the short or long edge of the scanner?

- the offset is significantly bigger towards the left and right edges
of the scanner, whilst in right the middle (vertical axis from top to
bottom) all is fine.

Still weird, but there seems to be some kind of logic in it. The IR
records the dust spots farther away from the vertical center axis than
they really are.

My guess is that in your internal forays you have moved something on the
optical head or the transport mechanism. An offset along the long edge,
even a variable offset would indicate a transport problem and/or
inconsistency - perhaps a damaged belt or dirt on the position feedback
sensor. An offset along the short edge, particularly a variable one,
would indicate movement of the optic or the IR source/sensor relative to
the visible one.

It may be that these are simply not fixed well enough and moving the
scanner around to open it has misaligned them. Was the scanner moved
before the problem appeared?

 

[Jan]

Doug,

that would explain a vertical offset, not a horizontal one.

What puzzles me even more than the offset itself, is the fact that
once it's a vertical offset and on other occasions it's a horizontal
offset. It's never been skewed (touch wood ?).
During a series of test scans it doesn't change. I haven't figured out
if it changes after switching the scanner off and back on, or whether
it changes when I switch between Epson Scan and Vuescan to do the
testing.

By the way, the last test scan I did last night had a vertical offset
again.

I'm drilling down into this problem for two reasons :
- pure interest in trying to figure out how things work
- trying to document the problem as precisely as possible, so that I
can provide a very clear problem description when/if I turn my scanner
in for repair.

Jan

 

[Jan]

Further testing showed the following :

What do you mean by horizontal and vertical in this context? Not
everyone has the scanner mounted the same way on their desk. Is the
offset along the short or long edge of the scanner?

"vertical" = along the long edge
"horizontal = along the short edge

My guess is that in your internal forays you have moved something on the
optical head or the transport mechanism. An offset along the long edge,
even a variable offset would indicate a transport problem and/or
inconsistency - perhaps a damaged belt or dirt on the position feedback
sensor. An offset along the short edge, particularly a variable one,
would indicate movement of the optic or the IR source/sensor relative to
the visible one.

It may be that these are simply not fixed well enough and moving the
scanner around to open it has misaligned them. Was the scanner moved
before the problem appeared?

The fact that I have both "long edge" offsets and "short edge" offsets
but never at the same time puzzles me.
Both optics and some position sensor should have a problem then - but
not at the same time.
As the scanhead is (literally) embedded in a black box, no visual
detection is possible from the outside.

The scanner hasn't been opened or tampered with, it hasn't been
transported, it's always been handled carefully (unless another family
member bashed it by accident).

 

[Brad Nishida]

Is it possible there is enough airflow and/or vibration to move the print
between the IR and the regular scans?

 

[=?ISO-8859-1?Q?Herv=E9_JOLLY?=]

Hi.
I never got the ICE function to work properly, and if I read the IR
channel this is because of an horizontal offset. Do I have to send my
scanner to repair like you think for yours ?

 

[Jan]

Hervé JOLLY wrote :

I never got the ICE function to work properly, and if I read the IR
channel this is because of an horizontal offset. Do I have to send my
scanner to repair like you think for yours ?

I've had splendid results with ICE : no dust left at all, fungus and
other "animals" automatically removed from 40+ year old slides. The
only thing ICE doesn't handle is Newton rings (they don't block the IR
light).

I also described this misalignment problem in
http://www.experts-exchange.com. I got one reaction, suggesting to
send it in for repair.

That's what I'm going to do by the end of next week (I absolutely need
to do some scans first, be it without ICE...). I'll post some feedback
when I get it back.

Jan

 

[Jan]

What I had feared has happened... the problem just disappeared
again...

I needed to do some reflective 600dpi scans last night, before sending
my scanner in for repair. After that I did a last test scan of a slide
with ICE, to make sure the problem was still present.

Well... it vanished into thin air, ICE alignment is no longer a
problem (about 50 slide scans done since then).

Funny to say maybe, but I would have preferred this problem to be
persistent. I have no clue.