The sound profile of the transmission can be optimized using various methods. According to VDI 3720, noise emissions can be minimized by reducing sound radiation, structure-borne sound transmission or structure-borne sound excitation (Ref. 12). The first two methods of noise reduction (secondary measures) include all types of insulation and damping of structure-borne sound waves as well as adapting the natural oscillation behavior of the structure-borne sound conduction and radiation of affected components (Ref. 11). The most effective method of reducing transmission noise is to minimize the sound excitation (primary measure) itself. Excitations can be induced in the transmission by bearings, loose members or, most importantly, gearing. Reducing the excitation of this last component has the biggest effect (Ref. 10).
Gear wheel excitations can stem from meshing impact excitation (deviation from ideal gear meshing) and changes in tooth rigidity (changing number of meshing teeth) (Ref. 2). Due to the broad load and RPM ranges of the EV transmission, resulting in different deformation of gear teeth and therefore different gear meshing conditions, there can be no optimal condition for tooth meshing across the entire load range of the gearbox (Ref. 3).
Using the appropriate grinding methods, such as generative grinding, gear wheels can be hard finished very quickly and cost-effectively. This technique can be used very effectively to create flank modifications (crownings, reliefs, biases, etc.). Design engineers can take advantage of the flank modification capabilities of grinding machines and specifically design the gear mesh for wider load and RPM ranges, to reduce the impact excitation caused by premature meshing of deformed teeth. In a grinding project, it’s possible to set up many different modifications and simulate them directly on the grinding machine control system. The required crownings, reliefs or angle modifications can be produced easily and at low cost with conventional generative grinding. Biases can be specifically influenced by topological profile grinding. Biases can be produced separately for the left and right flank utilizing a multiaxis interpolation of the paths of motion, both for dressing and for grinding (Refs. 5, 6). In addition to specially generated macro geometries and modifications, it is also crucial to precisely match the specified geometry and reduce tiny deviations. Utilizing the right grinding tools, dressers and grinding parameters, the target geometry of the flank can be reproduced very precisely, thereby minimizing the audible noises that would be produced as the teeth mesh. Minor deviations which are still within the required tolerance specifications, but which customers often evaluate visually and subjectively (e.g., negative crowning or singular elevations) can be reduced. These improvements are reflected not only in the subjective visual assessment of the gear measurement report but also in a tactile spectral analysis, which then shows a reduction in the amplitudes of the tooth meshing frequency and its multiples.
The roughness of the tooth flank can be influenced both by the choice of tool and by the grinding technique. In structural shifting, the amount of shifting is increased until regular structures which are typically created by generative grinding are broken and replaced by irregular structures. Stringent surface quality and roughness specifications required by our customers can be met cost-effectively and reproducibly using multistep processes with fine grinding or polish grinding. Typically, this involves the use of a 2-piece dressable grinding worm with differing specifications. While rough grinding (standard generative grinding) utilizes a conventional grinding worm specification, fine grinding and polish grinding are done using special specifications. Fine grinding is likewise done with a vitrified bonded wheel but with a different type and size of grain. Surface finishes of Rz = 1–3 microns can be reliably achieved.
For even more exacting surface finish requirements, polish grinding can be used with its elastic polyurethane or resin bonded wheels. This makes it possible to achieve surface qualities of Rz
In addition, the foregoing technological factors, the grinding machine itself is fundamental and a key element which must be understood and optimized to obtain optimal grinding results regarding waviness and noise excitation. Straight ghost orders or sidebands not directly associated with the tooth meshing frequency, or its multiples most often originate from the grinding machine. Even the smallest error-based deviations on a scale of less than 0.1 microns, especially if they are integral concerning the gear circumference, can lead to noticeable noises during gearbox operation (see also process monitoring example). A potential cause can be the axis drive, including its measuring system. Deviations in the measuring system can lead to a minimally uneven rotation, resulting in a waviness in the component. The oscillating torques generated by electric motors can likewise produce such effects. Noise can also be influenced by spindle bearings or the balance condition of the axis. Moreover, every machine has its own natural resonances, which can vary by machine type and workpiece clamping set-up. At Kapp Niles, these listed points are already being implemented to minimize any potential sources of error (Ref. 13).
Nonetheless, deviations can still occur in production. This is where digitization comes in by maximizing transparency, it helps to optimize quality. There are countless networking, production, machine and support solutions available to help users with manufacturing. The closed loop is a digital tool which has proven itself to be important and effective for dealing with trend-based deviations. Through the cross-manufacturer GDE (Gear-Data-Exchange) interface, gear measurements are fed back to the grinding machine in a closed loop—using preselected tolerance bands, the grinding machine selects correction values based on the measurements and optimizes the grinding process. In case of unexpectedly high deviations, a machine operator is needed to make decisions. Condition monitoring routinely checks and logs the condition of the grinding machine itself. Regular, automatically triggered reference runs are used to obtain reference values that reflect the condition of the rotational and translational axes. For high-precision components, it is crucial the axes be in good condition since worn axes can have a significant impact on the grinding result. Part tracing is a type of digital twin, a file capable of receiving all the information from the grinding process (workpiece, technology and process data). This enables traceability and the exchange of information, especially in case of inconsistencies in production between different departments, locations and companies (e.g., between manufacturing and gearbox assembly). It also simplifies and speeds up both communication and action.
One fundamental component which can be integrated into part tracing is the process monitoring data pertaining to vibrations and irregularities in the grinding process. These data are explained in more detail in the next section (Ref. 9).
Process Monitoring for Noise-Related Components
Typically, to assess the grinding process, a small sample is inspected by tactile measurement, however, individual outliers cannot be reliably detected. This is where Kapp Niles‘ own process-monitoring system comes in, enabling 100 percent evaluation in real time. Based on internal signals from the machine control system and their multiple acceleration sensors, characteristic values are obtained which can be used to evaluate both the grinding and dressing processes.
For the dressing process, a single index is compiled which reflects the entire dressing process, thus also indicating the condition of the grinding worm. This index can then be clearly represented in a diagram and compared with the preceding dressing cycles. Threshold values for this characteristic make it possible to initiate an action on the part of the machine. It is also possible to perform a deep analysis of the raw data from the dressing cycle directly in the machine control system.
