Nikon LS 9000 scanner – Gradual sharpness decrease during scanning

After purchasing new scanner Nikon Super Coolscan 9000 ED and performing several multipass scans with switched-on ICE cubed function I was shocked by the poor sharpness of the scanned film grain. 

During the course of many tests I have discovered the problem of gradual decreasing sharpness during scanning. Gradual decreasing of the sharpness is caused by overheating of some of the elements inside of the film scanner NIKON LS 9000. 

Follow the grain sharpness, please on following examples.

All the scans of the film grain are scanned in 4000 dpi and publicized in the 100% size. Particular swatches from the left to right matches position on the scanned slide:

0, 14, 28, 42 and 55 millimeters in the direction of the film moving. 

Test1:

This problem is bounded to the switched-on DIGITAL ICE function and includes both modes - normal and fine. 

For example:

4x_superfine_no_ICE....... is trouble free! 

Scanner Nikon LS 9000 is equipped with the glass holder FH-869g. Focal plane of the holder (tested without appropriate plastic mask) is excellent – variation from the
ideal plane is within 3 – 4 focusing steps (focal sharpness tolerance is around
10 steps up-and-down). This problem is not affected by change of the focal plane. I have made number of scans with manually shifted focal plane +- 15 and +- 30 steps, however the effect was only respectively shifted.

The degradation is beginning to appear during second minute of scanning and gradually increases up to approx. 30th minute of continual scanning. During the course of scanning time longer then 30 minutes (8x superfine or 16x superfine) the reached unsharpness remains without change. Manual focus changes before scanning does not compensate this fault: 

1x_normal:



1x_superfine:



2x_superfine:



4x_superfine:

In the situation that a next scan is started within a few seconds after completion of the previous scan, the following scan unsharp already from the beginning. 

Most of low contrast slides without deep shadows and bright highlights are scanned by the “slower mode” (6x6 cm on 4000dpi/4x superfine takes about 26 minutes) - this mode is fully affected by the problem of decreasing sharpness during scanning. It applies to all modes (1x .... 16x). Only small amount of normal density slides, scanned by the faster* mode (6x6 cm on 4000dpi/4x superfine takes about 14 minutes) is a little bit better (but not good). 

* This “Faster mode” in the case of scanning high density or contrast slides is illogical – it is an automatic choice of the scanner firmware that cannot be changed by the user! 

I discovered this problem on three machines (two scanners I tested myself), serial numbers 200473 and 201705 – both were absolutely new.
These unsatisfactory results cause each unmodified scanner to be practically unusable for any serious work with function of the “hardware” dust and scratches removal.

This problem was probably the cause of the wide-spread superstition that: “Digital ICE causes decreasing of the scan sharpness”.
Previously I have had Nikon LS 8000 scanner and during 3 years I never registered this problem.

Test 2: 

I discovered that NIKON SCAN program (version 3 and 4) has one unusual feature: holding the left mouse key interrupts scanning as long as the key is pressed. This feature allowed me to make the following test:

I have had prepared 4x-supefine scan (current scanning time is about 30 minutes). Every 4th minute of the scanning time I interrupted scanning temporarily by pressing and holding down the left mouse button for the following 4 minutes. Overall scanning time has been increased to 60 minutes! This experiment allowed me to achieve visibly better sharpness (approx 50% better, but not fair!) of the scan. I would called it the “Software Cooling”.

Test 3:

Opened scanner (the scanner case was dismounted) provides better results than closed, but still unusable.

Power_semiconductors_open.jpg

Gives results approaching a „90%" of final quality - the fan is too low-lying and is unable to reach the margins of the board, where are heat-sensitive components.

Final cooling.jpg

- sharpness is now perfect at the whole scan area.
I think that failure caused by overheating is not problem of the LED driver power semiconductors, but it is caused by others elements on this board, subsequently heated by these power semiconductors.

“Overheating grade” (level of decreasing of sharpness) is also dependent on Digital ICE utilization = on the amount of “slide pollution”.

Final_cooling_4x_superfine.tif

HEPA filter protect scanner from dust scattered in the surrounding air. Effective cooling needs very intense air flow and relative cold air – maximally 23 degree of Celsius scale.

I think that this problem can be partially solved by firmware modification and will require replacement of some electronic components in the scanner. Adequate solution for better cooling of the scanner interior would be also helpful.

Prague, 23rd September 2004.

This publicized exploration of facts was given with the aid of the Czech Nikon representation to the Japan Nikon Corporation. Unfortunately this work was totally ignored and all other Nikon LS 9000 scanners are up to this day affected by the above-mentioned fault. I hope that my sophisticated solution will help to improve the functionality of many other Nikon LS 9000 scanners.