New EDM Generator Used for the Machining of 1.2363 and 1.2343ESR Steels

The basic building block of any electric discharge machine is a generator, which ensures the operation of material separation itself. Not only the erosion rate and electricity consumption but also the quality of the machined surfaces depend on the design of the generator. These key factors for ecient machining have been investigated using a new energy-saving and powerful generator developed for the electric discharge machine (EDM) while machining 1.2363 and 1.2343ESR steels. In order to monitor and model the responses in the form of eroding rate and surface quality, a two-level half-factor experiment was performed with one replication at the corner points and two replications at the central points, for a total of 80 rounds. Both graphite and copper electrodes of sizes 10x10 mm and 100x100 mm were used for eroding and the inuence of parameter settings: Open-voltage, Pulse current, Pulse on-time and Pulse off-time was monitored. It was found out that the shape of the electrode and Pulse current have the most signicant effect on the erosion rate. The parameters Pulse current, Pulse on-time and their mutual interaction have the most signicant effect on the surface topography. Statistically signicant factors inuencing the occurrence of defects turned out to be Pulse current, Pulse on-time and Material of workpiece, where it can be seen that the material 1.2343ESR is signicantly less prone to the formation of surface defects. process, which takes place between

and 1.2343ESR were machined, which were chosen due to their wide range of applicability, especially in the production of plastic injection moulds. The outcomes of this study can be used to optimize the production of a range of components made of these two steels in real conditions of companies.

Experimental material
The samples for the experiment were made of two types of alloy steels, namely 1.2363 and 1.2343ESR. Mechanical properties, chemical composition and the use of both steels were compiled in Tab. 1. For the experiment, the starting semi-nished products were used always in the form of a plate with a thickness of 10 mm, and they were divided for individual samples, which is shown in Fig. 1(d), the microstructure of the semi-nished products is shown by means of a light microscope (LM) in Fig. 1(e, f). The erosion depth of each sample was always 1 mm.

EDM machine setup
Graphite and pure copper tool electrodes ( Fig. 1(b,c)) in two shapes were used for machining: 100x100 mm and 10x10 mm. All samples were made on a 433GS EDM machine supplied by the manufacturer PENTA and equipped with a new current modulating generator called P-MG2.The photographs of the erosion of the samples are shown in Fig. 1(a). The control is integrated into a eld-programmable gate array (FPGA) with the possibility of later innovation, the input voltage is converted by a very fast A/D converter with a small delay. The generator is controlled via the EtherCAT eldbus, which contains all control signals, feedback for axis control and diagnostics. Output current stages are variable (normally 0.2 A ~ 60 A), pulse timing from 10 ns to 5 ms. A very important function is the multi-stage protection against in ammation and short circuit, especially for graphite electrodes and high current densities. During the machining, all samples were immersed in a dielectric liquid, which was kerosene.

Experimental methods
The samples produced within the project were carefully cleaned of kerosene from the dielectric bath and other impurities generated during erosion. From the samples measuring 100x100 mm, small cubes measuring 5x5 mm were cut out using a wire electrical discharge machine.Thus, it made it possible to analyse them in the limited spaces of individual devices and subsequently produce metallographic preparations from them. First, the topography of all machined surfaces was analysed using a Dektak XT contact 3D pro lometer from Bruker and Vision 64 software. Measurements were performed on 5 samples at 5 random locations, followed by an arithmetic mean. The measurement was performed according to the corresponding standard for pro le ISO 4287 [15]. 3D reliefs of surfaces were also obtained on this device, which were further processed in the Gwyddion program. Furthermore, the surfaces of individual samples as well as electrodes were studied by electron microscopy, using an electron microscope (SEM) of the Lyra3 type from the company Tescan. This microscope was further equipped with an energy-dispersive X-ray detector (EDX), which allowed the analysis of the chemical composition. However, the accuracy of this device is 0.5 wt.%, So in order to more accurately study the chemical composition, lamellae for the observation and analysis of the chemical composition were made in a transmission electron microscope (TEM) of the Titan type from ThermoFisher. The individual lamellae were manufactured using SEM of the Helios type from the company ThermoFisher, which was equipped with a focused ion beam (FIB). In order to analyse the condition of the subsurface layer, metallographic preparations were created. These metallographic preparations were prepared by conventional techniques -wet grinding and diamond paste polishing using the automatic preparation system TEGRAMIN 30 from Struers. The nal mechanical-chemical nishing was performed using a Struers OP-Chem suspension. After etching with Nital etchant, the structure of the material was observed and documented by light microscopy on an inverted light microscope Axio Observer Z1m from ZEISS. The subsurface state of the layer was further examined by electron microscopy on Lyra3.

Statistical evaluation of eroding rate
EDM machining is speci c according to the erosion rate-the speedof machining -compared to conventional machining methods. The eroding ratecannot be set on the machine within the machining program, but this rateis completely generated by the machine depending on the setting of eroding parameters, such as Open-voltage, Pulse current, Pulse on-time or Pulse off-time. It results in the fact that the electrode can never come into direct contact with the workpiece.
During the machining process, the eroding rateis always shown on the machine display, from where it has been rewritten for each sample in the graph shown in Fig. 2. It is clear from the individual erosion ratesthat samples with a shape of 100x100 m were eroded at the slowest ratesclose to zero. This proves to be normal, since such a large electrode area always allows only very slow material removal. The highest values of the erosion rateof these samples with the size of 100x100 mm were achieved in Sample 5 (material steel 1.2363, eroding parameters: U=280 V, I=30 A, Ton=200 µs, Toff=30 µs) and that is 0.00902 mm/min using a graphite electrode and using a copper electrode for Sample 76, namely 0.0066 mm/min. For electrodes measuring 10x10 mm, the highest rateof 0.68211 mm/min was achieved for Sample 42 (material steel 1.2363, eroding parameters: U=280 V, I=30 A, Ton=200 µs, Toff=35 µs) using copper electrodes and using a graphite electrode, the highest eroding rate was recorded for Sample 13, namely 0.43901 mm/min. Unfortunately, similar studies such as Muthuramalingam [20] or Amorim [21] do not report eroding ratesseparately.
Linear regression was used to model the eroding ratev, which assumes homogeneity of residue scatter (deviations from the model) and cannot be achieved due to two-order differences in eroding rate. Therefore, the eroding rateitself was replaced by the order of eroding rate, i.e. the decimal logarithm of the original eroding rate. The in uence of categorical input factors on the eroding rate and the order of eroding rate is shown in Fig. 2(b), where the practically zero variability of the eroding rate for 100x100 samples can be seen.
A regression model of equation (1) was constructed for the order of erosion rate, which describes 97.46% of the variability of the monitored data. This hierarchical model was created using Stepwise Selection of Terms (p-value <0.05) from a model where all factors and second-order interactions were.
where the lower limit for categorical factors is -1 and the upper limit is 1.

Surface topography and statistical evaluation
The topography is quali ed by deviations in the direction of the normal vector of the real surface from its perfectly smooth ideal state. It is a topography that represents a key aspect in the behaviour of a real component in interaction with its environment. From the point of view of tribology, smooth surfaces have a lower tendency to wear and lower coe cients of friction than those with higher values of topography parameters. Based on the irregularities of the topography of the surface, it is possible to predict the formation of nucleation sites for cracks or corrosion. The analysis of the surface topography in connection with the setting of the machine parameters is therefore crucial, especially in cases where the part is machined only by EDM without subsequent further nishing technology, usually in the form of grinding. For this reason, 2 pro le parameters were evaluated in this experiment, which were the arithmetical mean deviation of the pro le (Ra) and the maximum height of the pro le (Rz). The individual values of the topography parameters Ra and Rz were compiled in a table and shown in Fig. 4 (a,b). It is obvious that the lowest parameters of the topography were achieved in Sample 3, with Ra being only 1.75 µm.This sample had a shape of 100x100 mm, was made of material 1.2363 and machined with a graphite electrode. Thematerial 1.2343ESR was machined with the lowest topographic parameters for Sample 50 and Ra 2 µm also using a graphite electrode. The lowest values of the topography parameters for the shape 10x10 mm were evaluated in Samples 56 from the material 1.2343ESR machined with a graphite electrode and in Sample 34 from the material 1.2363 also machined with a graphite electrode. It is thus clear that by using graphite electrodes it is possible to achieve a higher surface quality and thus lower values of the topography parameters. If we compare these values with the values using an older generator, which were analysed by the Mouralova study [5], we can see an improvement in the reduction of all values of individual parameters. 3D reliefs of the samples with the lowest surface topography values, which were Samples 3 and 50 each of a different material but both machined with a graphite electrode, are shown in Fig. 4 (c,d). From these reliefs, the difference in the size of individual craters is clearly visible, which was caused by different settings of the machine parameters for both samples.It was shown that both topography parameters are dependent (p-value <0.0005) using Spearman's correlation coe cient, which, unlike Pearson's correlation coe cient, does not assume that the data come from a two-dimensional normal distribution. Hierarchical regression models (2,3) were constructed for both characteristics of the topography Ra and Rz, where, as in the case of the order of erosion rate, the Stepwise Selection of Terms method (p-value <0.05) was used.
where the lower limit for categorical factors is -1 and the upper limit is 1. shows that due to the signi cant interaction of I*Ton, there is a greater increase in Ra and Rz with increasing I when Ton is at the upper level.

Analysis of surface and subsurface area morphology, including the defects
The morphology of electrically discharge machined surfaces forms a so-called random topographic pro le. Thus, there are no periodically recurring elements of the topography, such as during turning, when there is a constantly recurring groove on the surface from the blade of the turning knife.Large craters are formed on EDM surfaces after individual electric discharges.Also here the eroded material adheres in the form of small spheres, which has not been washed away by the dielectric liquid stream. Due to the effects of short-term electric discharges (in the order of microseconds), not only the individual particles of material are eroded but also the evaporation of thematerial due to very high temperatures of 10 000 -20000 ° C [16]. Another undesirable effect of the process is the relatively frequent occurrence of defects in the form of cracks, which can only lead within the recast layer (completely melted and re-cooled layer) or can interfere with the base material. The occurrence of such cracks can signi cantly reduce the service life or proper functionality of machined parts, so it is essential that their occurrence on machined surfaces is as small or preferably none. During the whole analysis on an electron microscope, a backscattered electron detector (BSE) was used, always rst with a magni cation of 150x and then with 500x. Fig. 6 shows the morphologies of different samples, which differ signi cantly from each other. Sample 50 made of 1.2343ESR steel machined with a 100x100 mm graphite electrode with the second lowest surface topography parameters was chosen as the rst representative (see Fig. 6 (a)). Although its surface did not have the lowest Ra and Rz values, it was not covered by any cracks. In contrast, Sample 3 made of 1.2363 steel machined with a graphite electrode with a shape of 100x100 mm with the lowest topography parameters, which is shown in Fig. 6(c), was completely covered with a network of cracks. These cracks were further analysed in a cross section to assess whether they were only within the recast layer or interfering with the base material. The cracks on EDM machined surfaces are a common phenomenon, which was presented in, for example, the study by Rajendran [7], Guu [28], Tai [22] or Mandal [23]. Further for the comparison, Fig. 6(b) shows the morphology of Sample 42 made of 1.2363 steel machined with a 10x10 mm copper electrode, which had the lowest topography parameters. A large amount of recast layer is visible on its surface, as well as numerous burnt places, which are mainly made up of carbon. This is a surface condition that is unsuitable for the components, and this setting of the machine parameters can also be assessed in this way. From the chemical composition analysis shown in Fig. 6(b), it is apparent that there was no diffusion of electrode material into the workpiece. For samples that have been machined with a graphite electrode, this cannot be assessed because the material being machined contains carbon and carbon also forms a common contamination of surfaces. Diffusion processes from the tool electrode to the base material were not monitored in the case of using a copper electrode in Mouralová study [5], but when machining the TiB 2 material in the Torres study [24] they were already studied.
Due to a detailed study of the condition of the subsurface layer, metallographic cross-sectional preparations of all samples were made. The analysis was performed using electron microscopy, using a BSE detector, always with a magni cation of 1 000x and then 2 500x and 4 000x. The condition of the subsurface layer (see Fig. 7) is crucial, especially in terms of proper functionality and predicted service life of manufactured components. It makes it necessary to carefully analyse it. For this purpose, the condition of each sample was examined, while the presence of defects in the form of cracks was recorded on individual samples and is compiled in Tab. 7. Overall, cracks were detected in 25 samples out of a total of 80. This is a signi cantresult since in comparison with the use of the old type of generator in Mouralová study [5] in the same design of experiment, 41 samples with cracks from a total of 80 samples were detected. More prone surfaces for the formation of cracks are those that were machined with a graphite electrode in a total of 15 cases out of 25 and in the same number of cases the cracks have samples made with the shape of the electrode 100x100 mm. However, the cracks, as can be seen in Fig. 7 (c-h) have changed their character and extend across the entire recast layer to the base material or even originated in the base material. Cracks extending through the recast layer to the base material were also studied in the Lee study [25] during D2 steel machining, and cracks arising in the base material also appeared during the Hastelloy X machining in the Kang study [26]. These cracks are very dangerous because they can prematurely end the service life of the part produced. Cracks also occurred during machining of titanium alloy Ti-6Al-4V in study by Das [27] or Phan [28], or also while machining Zirconia-Toughened Alumina Ceramic in the study by Bilal [29].
Tab. 7 Analysis of the occurrence of defects on the surface of individual samples. copper probably results from the secondary redeposition during the thinning of the lamella on the copper holder. In the upper primarily affected EDM layer of 2 µm, the individual elements were distributed evenly up to an increased concentration of manganese, silicon and vanadium in the area close to the protective carbon layer. Furthermore, there was an increased concentration of copper, which in this case probably comes from a copper EDM electrode. A closer examination of the particle in Fig. 10 showed that the individual rays are formed by an increased concentration of chromium, while the presence of individual particles formed by vanadium was also con rmed. The exact chemical composition for both EDX 1 and EDX 2 scans is given in Tab. 8. Comparing the chemical composition of the measured areas, it was found that in the case of a more detailed examination of the lamella on EDX 2.There was an increase in carbon, silicon and iron concentrations compared to EDX 1. Elements such as vanadium, chromium, manganese, copper and molybdenum decreased. The prepared lamella from a sample of 1.2343ESR steel machined with a carbon electrode was also analysed in TEM, the accelerating voltage, current and setting of the whole device being the same as in the case of the rst lamella. A total of 2 chemical composition analyses were performed, with the rst (EDX 1) examining the entire lamella and the second (EDX 2) examining its detail. The lamella itself and EDX 1 are shown in Fig. 11, it being clear that in the area of the upper part of the lamella (2 µm from the surface) all elements were evenly distributed up to the surface area in which the concentration of manganese, vanadium and silicon increased. In the remaining part of the lamellae, as in the case of the rst lamellae, they consisted of vanadium, chromium, manganese and molybdenum. In addition, individual vanadium particles were detected. EDX 2 led to the same conclusion as in the case of the rst lamella, that the core of the particles is formed by a network of chromium veins. The chemical composition from the EDX 1 and EDX 2 sites was summarized in Tab. 9. A comparison of individual areas of EDX measurements with respect to the total concentration of individual elements shows an increase in carbon, vanadium and manganese in the case of EDX 2 measurements. The only element in which there was no change in concentration was silicon.The lamella formed from the EDM workpiece was also investigated in Mouralova's study [5], where the same materials were machined only using the old version of the generator. It was found that essentially no changes were achieved here. The TEM lamella made of EDM machined material was also created and analysed in the Murray study [30] when machining austenitic steel or also when machining single-crystal silicon in the Murray study [31].
Tab.9 Analysis of the chemical composition in individual places according to Fig. 11

Conclusions
As part of the testing of a completely new energy-saving and powerful EDM generator, samples of 1.2363 and 1.2343ESR steels were produced according to an extensive design of experiment of 80 rounds, which were subsequently subjected to many analyses and statistical evaluations, which brought the following conclusions: at the slowest rates close to zero, samples with the shape of 100x100 m were eroded, the highest values of the erosion rate of these samples with the size of 100x100 mm were achieved for Sample 5 (material steel 1.2363, erosion parameters: U=280 V, I=30 A, Ton=200 µs, Toff=30 µs), namely 0.00902 mm/min using a graphite electrode. For electrodes with a size of 10x10 mm the highest rate of 0.68211 mm/min was achieved for Sample 42 (material steel 1.2363, erosion parameters: U=280 V ,I=30 A, Ton=200 µs, Toff=35 µs) using a copper electrode, the shape of the electrode and Pulse current have the most signi cant in uence on the erosion rate, Pulse on-time, Pulse off-time and electrode material are also signi cant, the lowest topography parameters were achieved in Sample 3 (material steel 1.2363, shape 100x100 mm, graphite electrode, erosion parameters: U=280 V, I=10 A, Ton=50 µs, Toff=35 µs) and Ra only 1.75 µm, while it was found that by using graphite electrodes it is possible to achieve lower values of topography parameters,           TEM lamella made of 1.2343ESR steel, including maps of the layout of individual elements in various details.