Industrial Technology - Linked-in Industrial Technology - Twitter Industrial Technology - News Feed
Latest Issue

Increasing machine throughput by decreasing the time to make a move

Increasing machine throughput by decreasing the time to make a move

Today’s world sees demand for more and more features, functionality and novel manufacturing methods to deliver as example new electronic devices, mobiles without increasing weight or size but maintaining battery life. As an example adding more and more electronic features in less and less space ultimately challenges the manufacturing processes. Examples include indexing parts or process heads over small distances as quickly as possible and every millisecond costs money.

But this can cause problems because the increased acceleration required for these faster move times causes the machine base to rock, so disturbing the motion and impacting both on the achievable speed and attainable accuracy. The law of equal and opposite forces means that as a control effort is applied to the motion stage to perform its indexing motion, so there is an opposing disturbance force applied to the machine base, exciting the natural frequency of the base and resulting in position errors.

Some of this vibration can be down to the machine design. For example, screw type levelling feet can be a cause of vibration, as can the use of light weight machine bases that can’t absorb disturbance forces adequately. But even with large machine base masses such as granite it is still possible that there will be resonance within the machine structure.

To compensate for or, better yet, eliminate the problem, first we have to be able to diagnose the base resonance. Many advanced motion controllers today allow you to analyse the open loop frequency response of the motion system as a method of tuning the servo loop. This is a good way of understanding if there is a frequency that will affect settling time as well as understanding how the system performs dynamically.

Typically if a machine base is heavy, these resonances are at low frequencies, with 12Hz being a common problem area.

How, though, do we solve the issue? Let’s consider a real world example. Figure 1 shows a typical Cartesian gantry that was used to position a laser head to perform a process at several points along the part which was to be mounted on the granite base. On the granite base are three black acceleration sensors which measure any movement of the base. In this example, the feedback from these devices whilst the machine was performing its process steps showed a strong and lightly damped resonance at 12Hz. In this case when the move was completed, the difference between ‘MoveDone’ for the X and XX axis (dual bottom axis) and ‘In Position’ which is when the position error is below and stays below the specified 1µm error for the process to continue was 239ms (Figure 2). In other words, there was a 293ms base settling time. This might seem small, but in fact it was critical, adding to the process time for the system.

If we consider that there were around 100 moves per completed part, this settling time had significant implications for the overall process time and for the productivity of the machine. Can we be sure that the base vibration was the real cause of the increase in system settling time? In this case, the signals from the feedback sensors proved that it was the effective cause, since the analogue signal from the base sensor was the opposite of the disturbance in the measured position error.

So how would we look to resolve the issue? One option is to look at a larger, heavier machine base. However increased costs and size as well as shipping and handling issues would rarely justify the smaller increase in performance, especially when it is not a guaranteed fix to the problem.

An alternative method is to use the feedback from the sensors to improve the settling time of the system. In the case of this real world example, feedback from the base sensor was applied to the current loop of the servo. Some gain was added to the measured input from the base sensor so that the amount of correction added into the servo loop was enough to offset the measured base motion. A low pass filter was then applied to the signal so that only the low frequency base motion was corrected and high frequency noise from the sensor was rejected. 

And the results? In the plot in Figure 3, we can see that the difference between ‘MoveDone’ for the X and XX axis (dual bottom axis) and ‘In Position’ – which is when the position error is below and stays below the specified 1µm error for the process to continue – is 103ms. So, for a single move, we have achieved a saving of time from 293ms to 103ms. In other words we have saved 136ms per index or 13.6 seconds per part. The process only takes 20ms so the machine cycle is decreased by more than 50% driving a huge increase in overall productivity of the machine, reducing the cost of the part being manufactured.

Some controllers, such as those from Aerotech, have standard tools to allow this tuning and setup of the functionality. Further, Aerotech offers a complete range of sensors ready for attachment to the machine to deliver these important savings.

Download pdf

Other News from Aerotech

Greater efficiency in the manufacture of cardiovascular products

Latest news about Machine Building

Additional Information
Text styles