Dr. Razvan Panaitescu, AMD PM MS, Siemens Industry, Inc.
Converting print unit systems to gearless systems may be a new concept in the printing industry, but better accuracy often results.
For years, printing systems have used the same gear designs to replicate an image. With these designs, alignment problems happen, as well as inconsistencies in ink distribution. To perfect printing systems, Mechatronic Support considered changing the entire gear design and how printing systems operate. With the collaboration of B&L, it investigated the possibility of implementing a gearless printing unit. By using a torque motor for each blanket cylinder, it eliminated the imperfections in the gear design, backlash and the long chain of compliant elements. The application for the printing units was created and installed using SIMOTION/SINAMICS, as the motion control equipment for Siemens’ 1PH7163 motors, driving the upper and lower sides of each unit.
The printing units of a shaftless web offset press, retrofitted by B&L Machine Design, were analyzed. While all four units run independently, no shaft ensured synchronism during the operation.
The four units have identical mechanical structure and are responsible for one of four colors (black, cyan, magenta and yellow). On each unit there were two power trains mirrored vertically. Each unit printed on one side of the paper. The motor was coupled to its 30-teeth pinion through a bellows coupling. The pinion was geared with the blanket cylinder gear (120-teeth), at a 1:4 ratio. The pinion at the blanket drives a third gear (60-teeth), which belonged to the plate cylinder, at a 2:1 ratio. With this latter gear, a number of 14-16 other cylinders were driven. Those cylinders formed the inking system, allowing the ink and the water to slowly circulate and be disposed on the plate surface.
During a printing test, the blue color seemed to be off registration, but the same effect could be seen with the other colors. The registration points were visibly oscillating with an amplitude of 4 to 5 thousands of an inch (or approximately 75 micrometers). The root cause of the oscillation seemed to be the settings of the speed and position controller in all four units. The controller had been hard tuned, with an extremely low integrator time of 1 ms. These settings made the controller dynamic and therefore, it responded promptly to all disturbances occurring in the system. The main disturbance in the system was due to the existing gaps at the surface of both plate and blanket cylinders. When the gap of the plate cylinder touched the blanket surface (whether it was above the blanket gap or 180° phase shifted), a disturbance torque (shock) affected the power train and speed. This torque excited the natural frequencies of the system, which were quite low due to the large inertias driven by the motor and relatively flexible couplings between the gears of each cylinder.
Frequency response measurements
Several frequency response measurements on the speed-controlled system were taken. The difference of the frequency response was between the position of the cylinders and other printing components of the press.
Two cases were considered:
a) The lower and upper units driven independently.
b) The lower and upper units were in impression (only one of the units was active, while the other one was acting as an additional driven inertia, through the friction forces occurring between the blanket cylinders).
In both cases, without a significant shift in frequencies, similar responses were observed. The reason for insignificant frequency shifts was that the friction contributing to attenuating the poles in the system was not high enough to act like a stiff connection between the systems.
The speed-controlled system showed several zeros at low frequencies: 13, 20, and 38 Hz, which seemed to be the frequencies of interest when tuning the speed and the position controllers. Although the controller had been hard tuned before, softer tuning permitted a more stable controller with closed-loop amplitudes situated below the 0 dB line.
The drive parameters may be further adjusted, should the printing present registration problems. The integral time may be increased to a maximum value of 80 ms, while the speed controller proportional gain may be decreased to a minimum of 170 Nms/rad. The changes in either of the two parameters must be correlated together and with a corresponding modification of the reference model frequency.
The only modification brought to the axis settings was the reduction in the position controller gain from 50s-1 to 40s-1. This reduction accommodated the lowest undamped natural frequency of approximately 20 Hz, which was observed in the plant frequency responses. Depending on the actual result in printing, this setting may be even lower down to approximately 25s-1. The lowest natural frequency, as well as the amount of damping in the power train played an important role in setting this frequency.
Time domain measurements
Using speed step responses measurement parameters: amplitude – 30 rpm; offset – 20 rpm; ramp-up time – 20 ms; setting time – 200 ms; and measuring time – 682.5 ms, the paper ran between the upper and lower unit blanket cylinders. The image was impressed on both sides. At every rotation of the blanket on the web, there should have been a gap of 0.180 in. This gap overlaps with the gap on the plate cylinder, of 0.187 in. A second gap in printing occurred when the plate touched the blanket at 180º phase to the blanket gap. The presence of these gaps (in both blanket and plate cylinders) introduced a repetitive disturbance force. Once every blanket rotation, when both gaps met, this force acted on every plate rotation with higher amplitudes.
The disturbance forces in the process also occurred from the gear teeth entering in contact during the operation. Because the gears were not precise and presented backlash, a hard tuned motor/controller fought to stay in position producing additional vibration effects. This effect would have been more visible if there were a direct measurement system and the controller had adjusted the position at the load, instead of at the motor encoder.
With the initial settings, the controller was hard-tuned and was seen in the behavior of the torque and speed, as they were oscillating with the lowest mechanical resonance. The frequency observed on the torque (or speed) response was about 13 Hz. It was the natural frequency of the mechanics and was probably the frequency observed on the printing result, not related to the disturbance forces and not dependent on speed.
When the controllers were tuned in a damping optimal setting (higher integrator times and lower proportional gains), the printing result was improved and the registration problems overcome. After tuning, the oscillations in speed and torque were dramatically damped-out. This tuning related to the speed controller settings, which became softer and less responsive to disturbances. Thus, the oscillations coming from the large cylinder inertia were attenuated and the command response was greatly improved.
Several other vibration tests were performed on each printing unit. The test results showed mechanical problems were also present in the gear train of the black and the blue units. Both units presented unacceptable vibration levels at the plate and blanket cylinders. Because the vibration spectrum only contained multiples of the gear mesh frequency, it was concluded that the root cause of these vibrations was likely to be one or more bad gears in the power train, at the black and blue printing units. B&L was advised to check the mechanical accuracy of the gear train and observe any possible inaccuracies at the teeth level on the gears.
During constant or run-up operation, a number of vibration measurements were taken in radial directions (X and Y) at the mounting flange of the blanket and plate cylinders on each side of a printing unit. Measurements were taken at the current operating speed. The time domain values of the acceleration are in peak instantaneous values and scaled to represent g levels (g = 9.87 m/s2).
At 600 ft/min the blanket rotated with
600/2π *14.484in/12 = 79.116min-1
The motor was spinning four times faster than the blanket, with a frequency of
(79.116/60)* 4 = 5.27 Hz
The disturbing frequencies were related to the number of gear teeth as follows:
zmotor = 30 → ƒ = 30*5.27 = 158 Hz
zplate = 60 → ƒ = 60*5.27 = 316.5 Hz
zlanket = 120 → ƒ = 120*5.27 = 633 Hz
The integer multiples of these frequencies (first, second, third order harmonics) were equally spaced with multiples of 158 Hz. Similar calculations were done for the 900 ft/min case, with the base frequency of 237 Hz (mesh frequency of the motor gear), and multiples at 474 Hz, 711 Hz (third harmonic), 948 Hz, and 1185 Hz (fifth harmonic).
When measured at the front of the printing units, both in X and in Y directions, the vibration levels were not exceeding 0.5 g, meaning that they are within manufacturer specifications.
In the Y direction, the black upper unit seemed to present vibration peaks of 2 g. In the X direction, the vibration readings reached the highest level. The black unit, both on the upper and the lower cylinder, displayed peaks at 3 g. Similarly, the blue unit displayed peaks close to 2.5 g.
Both black and blue units were checked to observe the gear execution. The level of vibration was due to an increased amplification of the 60th harmonic, which came from the plate gear mesh frequency. The vibration levels increased with the production speed. At 900 ft/min, vibration levels reached close to 4 g in the black and blue units. These values were not acceptable and immediate attention was required to correct the errors suspected to have occurred at both gears.
Conclusion to modal analysis
A compliance frequency response testing was performed by exciting the load and measured its response at the action point. Some of the measurements performed were in the opposite direction to the direction of impact (thus misleading behavior of the phase). The units were tested in pairs to show the consistency of modal readings throughout the whole printing press.
The Y-direction modal analysis reflected the torsional modes, as the directions of readings and impact were tangential to the cylinder surface. The torsional modes (more visible in the frequency response characteristics) could not be identified in the compliance frequency response. Some modes showed more amplification than others (for example 400 Hz in Y direction) as resonance was reached when the frequency became an integer multiple of the frequency of rotation. The Eigen frequencies were well damped and did not contribute too much to increased levels of vibration.
Mechanical problems observed through vibration measurements were reported in the black and blue printing units. Using Mechatronic Support investigation resources and techniques, the registration problem was corrected and showed improvement by rigorous testing procedures applied and implemented. Once the registration problem was corrected, an insight of how to further assess improvements in the controller settings was given to the customer, which now can perform a similar analysis on further systems. The registration problem can still be improved if rigorous testing procedures are applied and implemented.
B&L Machine Design
Siemens Industry Inc.