TABLE OF CONTENTS Dec 2003 - 0 comments

Rotogravure print quality evaluation analysis, pitfalls and statistics

TEXT SIZE bigger text smaller text
By: H.U. Heintze
To download the full version of this article including table images, please click here.

Abstract: Print quality problems were evaluated in commercial rotogravure prints. The substrates ranged from roto-news to high quality coated papers. The evaluation focused on the frequency of missing dots. The variation of skipped dot counts in the prints confirms the need to count a large number of skips in a commercial print to obtain statistically valid rankings of skipped dot counts. This sufficiently large paper sample area must also be taken into account in making laboratory prints or carrying out other paper smoothness tests.

Rotogravure involves printing from a smooth metal cylinder that carries the image to be printed in the form of small cells etched or engraved onto the cylinder surface. The cylinder is rotated in a bath of very fluid ink, the excess is doctored off and the printing substrate pressed against the cylinder. Ink is transferred from the cells to the substrate by wetting and film spreading forces.

When the printing substrate is too rough it does not make sufficient contact with some cells on the cylinder surface. This results in skipped cells in the print. These skipped cells, also sometimes referred to as missing dots or speckle, have long been one of the main print quality problems attributed to paper in the rotogravure process.

The paper industry has used various laboratory gravure proof presses to assess paper roughness for gravure printing [1,2,3]. With the advent of image-analysis systems it became possible to quantify skipped cells in laboratory prints, and to relate image-analysis skipped cell counts to visual quality [4,5]. Development of robust image analysis methods continues, using more modern equipment and software [6]. Speckle frequency in laboratory prints has been shown to exhibit high variability [5]. It also has been shown to depend strongly on press operating conditions on a GRI laboratory proof press [7].

The quantification of rotogravure paper quality and of cell skipping by either visual or image-analysis methods must take into account the statistical nature of skipped cell counts. The GRI proof press, which has been the subject of several studies [1,4,5,7], uses a relatively large light tint area of 88 mm ¥ 38 mm to assess cell skipping. Even with this large area, the high/low skip count ratio for average lots of 10 prints on coated board ranged from 1.4 to 10.7, with an average ratio of 2.5 [5]. Samples with lower average skipped cell counts showed greater variability of skip count. All laboratory presses evaluated in one round robin study exhibited high variability of cell skipping in multiple sheet tests [2].

The quality improvement problem for the papermaker is compounded by paper machine roughness profiles that have been shown to relate to cell skipping in laboratory gravure prints [8]. Such profile issues can be studied with the Tapio Paper Machine Analyzer and the Tapio Lab Printing Machine, which can analyse continuous narrow strips of paper up to 2.5 km long [9].

Electro-Static Assist (ESA) was introduced to reduce the severity of skipping by using an electrostatic field across the printing nip to deform the ink meniscus in the individual cells, and thereby improve ink transfer. The reduction in speckle due to ESA has been shown to depend strongly on the voltages/currents across the printing nip [10]. This effect of ESA voltage can significantly alter skipped cell counts and mask the effects of paper roughness to varying degrees.

The objective of this study is to define typical skipped cell levels in current commercial samples and relate them to paper evaluation issues both in the laboratory and in commercial printing trials.


Commercially printed samples of major rotogravure publications and of sales flyers were collected from Los Angeles, New York, central Canada and Germany. The paper substrates covered the spectrum from roto-news to high-quality coated paper. Samples were subjected to a visual screening to identify and select sufficiently large uniform single-colour areas where cell skipping might be expected. This approach depends on the presence of suitable printed areas in the selected publications.

Uniform light to medium tints on the selected prints were scanned in 10.0 mm by 10.0 mm squares at an optical resolution of 2,400 dpi using a Canon D2400UF scanner with an optical resolution of 2,400 dpi ¥ 4,800 dpi. The pages were backed by a black sheet to minimize the effect of print-through from the reverse of the sheet. The scanned areas were then printed to dimensions of 20.6 cm by 20.6 cm on an HP LaserJet 2100 printer. This magnification made it simple to visually count skips.

Higher magnification images showing details of printed dot shape were obtained using an Olympus C-3020Z 3.2 mega-pixel digital camera fitted to a Wild-Heerbrugg microscope.

Apparent screen rulings for the rotogravure prints were determined by counting dots in the images scanned at 2,400 dpi. This count may differ from the screen count on the engraved cylinders because of paper expansion or contraction effects during and after printing.

Magnification of the scanned images and of the microphotographs was estimated by imaging a glass scale marked off in increments of 0.13 mm (0.005-in.).

Basic image analysis measurements of printed cells were made using ImageJ (version 1.28), obtained from the NIH (National Institutes of Health) Web-site.


The average skipped cell counts per square centimetre on the selected paper samples are summarized in Table I. The number of areas measured for cell skipping was determined by the images available in each publication.

Cell skipping was found to be almost non-existent in the best quality sample (A). Some speckle was found in the lightest tones, but the contrast of these printed cells was so low that the skipped cells were not visible to the naked eye.

Publications B and C, printed on coated or filled SC papers, exhibited low levels of cell skipping in medium and light tones. They had average skip levels of five and three skipped cells per cm2 respectively. There was no obvious sign of two-sidedness of cell skipping in these papers.

Publication D had a large uniform tint printed on both sides of the same sheet. This uniform tint was large enough to permit measurement of 11 areas of 1 cm2 on each side of the sheet. Side 1 had an average of five skips per cm2. However, the sample was very two-sided, with an average skip count of 109 skips per cm2 on side 2.

The roto-news publications had high levels of cell skipping and other noise in the printed cell pattern.

The minimum and maximum skipped cell counts in Table I confirm for commercial gravure prints the high variability of cell skipping found in laboratory prints. The impact of this variability on evaluations of cell skipping is significant.

The ranking of a set of papers for cell skip tendency will require a large area of light tones in the print. This area must contain enough skipped cells to produce useful statistical confidence limits for the average skipped cell counts. If the paper quality differences are large this will not be a problem. However, the statistics of skipped cell counting will be a problem in mill-improvement trials that give small to moderate quality changes or in competitive benchmarking trials.

While examining the various prints for skipped cells, it became evident that the papers were not all prepared at the same screen ruling.

The apparent screen rulings obtained by counting printed cells extended from a low of about 60 cells per centimetre to a high value of about 100 cells per centimetre. This produces apparent cell centre spacings in the range of 100 to 172 microns. A lighter tone cell, whose diameter may be only 10% of the cell centre spacing, is thus in the size range of softwood fibre widths.

An interesting observation on the commercial prints is that almost none of the lighter tone printed cells are circular in shape and that the majority of printed cells have a skewed doughnut shape as shown in Fig. 1. Figure 1 also contains open-ended cells and a small number of solid dots. This indicates variability in the engraving process or in the ink transfer.

A magnified printed single cell in Fig. 2 illustrates the appearance of the typical cell on coated or SC paper. Figure 3 illustrates the variation in relative reflectance across this area. The higher quality prints in this collection showed printed cells that tended to be darker at the trailing edge of the cell in the direction of web travel through the press. This means that more ink was transferred from the back edge of the cell.

In view of the very fluid nature of gravure ink, this makes sense if the initial contact of the cell with the paper surface takes place at the trailing edge of the cell. This will be expected if the inertial forces in a cylinder rotating at high speed cause the ink to be pushed to the trailing edge of the cell. This results in relative depletion of ink at the lead edge of a cell and an ink bulge at the trailing edge. The bulge will then be first to contact the paper surface and ink will wick to the paper around the circumference of the cell from this initial contact point. This hypothesis could explain the open doughnuts visible in Fig. 1. There was not enough ink in the cell to wick all the way around from the initial contact point. The ink displacement in a cell is expected to become more important at higher rotational press speeds, resulting in variations in printed dot shape with increasing speed. Cells engraved on the cylinder with a high-speed stylus will have a shape dictated by the stylus and will generally not be circular.

Many of the prints on uncoated paper exhibited not only very high skip frequencies, but also a number of other print defects related to coarse fibres at the sheet surface. Figure 4 illustrates typical ink lay patterns that occur on a surface containing large, coarse fibres. Skipped cells are evident. There are also clear indications of ink flow along surface fibres. Where initial contact of the ink has been made on a large surface fibre the ink tends to flow along that fibre rather than spread laterally and form a dot. The improvement of roto-news print quality thus requires an emphasis on the coarse long fibre fraction of the fibre furnish. Paper machine modifications such as pressing and calendering are not necessarily the best means of achieving the goal of reduced cell skipping since they only mask the effect of coarse softwood fibres that may be present.


1. The rotogravure printing process continues to exhibit troublesome skipped cells, even after years of paper quality development, the implementation of Electro-Static Assist and the evolution of presses and inks.

2. The cell skipping found in these commercial prints follows the general patterns found in laboratory-proof prints. Skips are more frequent in lighter tones and skip counts per unit area exhibit large variability both within a sheet and between sheets.

3. The low numbers of skipped cells in these commercial prints lead to high variability in skipped cell counts. Using the Poisson approximation as a rough guide, it can be estimated that 900 skipped cells must be counted to obtain a coefficient of variation of 3% on the count.

4. The high variability of cell skipping in an area of one square centimetre within a single printed page indicates that laboratory predictors of gravure print smoothness based on measuring only a few such small areas will not be useful in mill practice. This is true for laboratory-proof presses and for all other devices.

5. The ink lay of individual printed gravure cells is non-uniform and points to a complex ink transfer process.

6. The fibre-direction ink flows on roto-news complicate the measurement of cell skipping by image analysis because of the general noise they introduce in the ink lay.

7. Large, coarse softwood fibres at the paper surface are detrimental to rotogravure print quality. A sheet surface needs sufficient fine material to help fill the holes and level the surface.

8. Image-analysis methods aimed at measuring the circularity of printed cells are not likely to be helpful because the typical rotogravure dot is not necessarily round to start with.


1. HEINTZE, H.U., GORDON, R.W. Tuning of the GRI Proof Press as a Predictor of Rotonews Print Quality in the Pressroom. Tappi J. 62(11): 97-101 (1979).

2. HEINTZE, H.U., GORDON, R.W. A Round Robin Evaluation of Gravure Printability Testers. Proc., 1982 International Printing and Graphic Arts Conference, Quebec, 39-45 (1982).

3. CHEN, J., OLIVER, J., SMITH, G., HAUN, J., KEEFE, S., LAUBER, D. A Print Quality Monitoring Program for Roto Grade Newsprint. Part I: Preliminary Design of Print Quality Evaluation. Preprints, 87th PAPTAC Annual Meeting, B169-B175 (2001).

4. HEINTZE, H.U. GORDON, R.W. Reliability of Videoscanner Measurement of Gravure Speckle on Coated Board. Tappi J. 63(9): 125-128 (1980).

5 HEINTZE, H.U. GORDON, R.W. The Measurement and Significance of Small Differences in Gravure Speckle Levels. Tappi J. 63(10): 91-93 (1980).

6. HEESCHEN, W.A., SMITH, D.A. Robust Digital Image Analysis Method for Counting Missing Dots in Gravure Printing. Proc., 2000 TAPPI International Printing and Graphics Arts Conference, Savannah, 29-35 (2000).

7. HEINTZE, H.U. Press Operation and Gravure Print Quality. Tappi J. 65(6): 109-112 (1982).

8. ALGUARD, M.J., LUCAS, J.M., HEINTZE, H.U. On-line roughness and printability. Proc., 1981 TAPPI Coating Conference, Houston, 135-141 (1981).

9. PERENTO, J. Paper Trials Help to Improve Quality and Printability. Paper Asia 15(7): 15-17 (1999).

10. SILER, S.J. Rediscover the Color with ESA and Water-Based Inks. Gravure 13(4): 52-56 (1999).

Résumé: Nous avons évalué les problèmes inhérents à la qualité d'impression des imprimés commerciaux en rotogravure. Les substrats allaient du papier journal roto aux papiers couchés de haute qualité. L'évaluation a porté sur la fréquence des points manquants. La variation du nombre de points manquants confirme qu'il est nécessaire de compter une grande quantité de manques dans les imprimés commerciaux pour obtenir un classement statistique valide des comptes de points manquants. On doit aussi s'assurer que l'échantillon de papier est suffisamment large lorsqu'on procède à des impressions en laboratoire ou à des essais du lissé du papier


FIG. 3. Relative reflectance variation over the printed cell shown in Fig. 2.
Larger photo & full caption

File size: 15.2 KB (285px X 243px)
Caption: FIG. 3. Relative reflectance variation over the printed...
FIG. 1. Printed cells and a skipped celll in publication C.
Larger photo & full caption

File size: 15.9 KB (288px X 215px)
Caption: FIG. 1. Printed cells and a skipped celll in publicatio...
FIG. 4. Typical image of cells printed in a mass-market roto-news publication.
Larger photo & full caption

File size: 18.1 KB (296px X 224px)
Caption: FIG. 4. Typical image of cells printed in a mass-market...
FIG. 2. A single printed cell in publication C. The web in the press ran from right to left.
Larger photo & full caption

File size: 8.8 KB (289px X 200px)
Caption: FIG. 2. A single printed cell in publication C. The web...
H.U. HEINTZEConsultantMontreal, QCheintze@sympatico.ca
Larger photo & full caption

File size: 4.7 KB (148px X 185px)
Caption: H.U. HEINTZEConsultantMontreal, QCheintze@sympatico.ca
Monitor These Topics

Horizontal ruler
Horizontal Ruler

Post A Comment

Note: By submitting your comments you acknowledge that Pulp and Paper Canada has the right to reproduce, broadcast and publicize those comments or any part thereof in any manner whatsoever. Please note that due to the volume of e-mails we receive, not all comments will be published and those that are published will not be edited. However, all will be carefully read, considered and appreciated.

Your Name (this will appear with your post) *

Email Address (will not be published) *

Comments *

* mandatory fields