A STUDY OF TE ACCURACY OF TE MICRO ELECTRICAL DISCARGE MACINING DRILLING PROCESS D.T. Pham, S.S. Dimov, S. Bigot, A. Ivanov, and K. Popov Manufacturing Engineering Centre, School of Engineering, Cardiff University,CF24 0YF Abstract This paper examines the influence of various factors on the final accuracy in EDM micro drilling. In particular, the main parameters affecting the size and position of the hole are discussed and techniques for minimising errors are proposed. The paper analyses the technological capabilities of different methods of setting up and dressing the drill (electrode) on the machine. The paper also evaluates the limitations of the EDM drilling process. Keywords: micro EDM, micro-machining, EDM accuracy, micro holes Notations d D g d g e g meas X meas X pos y d Y meas Y pos y unit d D D guide D init D init_min g d g e g meas L guide t meas d variation D variation g d variation g e variation g meas variation variation X meas variation X pos variation y d variation Y meas variation Y pos variation y unit variation effective dressed diameter of electrode electrode initial effective diameter of electrode diameter of the ceramic guide initial diameter of the electrode minimum initial diameter of the electrode spark gap during dressing of the electrode spark gap during erosion of the workpiece spark gap during measuring cycle achieved hole diameter length of the guide time interval between each contact signal check 1
V meas X X meas X pos y d y d_init Y Y meas Y pos y unit z guide speed of measuring cycle movement X coordinate of the hole relative to the workpiece reference X coordinate of the measured point of contact between electrode and workpiece surface X coordinate of the hole relative to the machine zero point target dressing position in Y axis an initial target dressing position Y coordinate of the hole relative to the workpiece reference Y coordinate of the measured point of contact between electrode and workpiece surface Y coordinate of the hole relative to the machine zero point Y coordinate of the dressing point relative to the machine zero point length of electrode protruding from the ceramic guide 1 Introduction Electrical Discharge Machining (EDM) is a non-contact machining process for conductive materials that has been applied for more than 40 years and has proved particularly useful in the tool making industry. Due to its high precision and the good surface quality it can produce, EDM is potentially very suitable for micro-fabrication [1, 2]. The EDM process utilises the thermo-electric energy released between a workpiece and a highly charged electrode submerged in a dielectric fluid. A pulsed electrical discharge across the small gap (known as the spark gap) between the workpiece and the electrode removes material from the workpiece through melting and evaporation. Clearly, due to the contactless nature of EDM, there are only very small process forces. This, complied with the availability in recent years of advanced computer controlled spark generators that help improve machined surface roughness, promises to make EDM the preferred method for producing micro features. This paper focuses on the use of EDM to create one type of micro feature holes. The paper examines the influence of various factors on the final accuracy in EDM micro drilling. In particular, the main parameters affecting the size and position of the hole are discussed and techniques for minimising errors are proposed. The paper analyses the technological capabilities of different methods of setting up and dressing the drill (electrode) on the machine. The paper also evaluates the limitations of the EDM drilling process. 2 Process overview The electrodes usually employed for micro EDM drilling are tungsten (W) or tungsten carbide (WC) rods or tubes of 0.1mm to 0.2mm in diameter and 150 mm to 300mm in length. One end of the electrode is clamped in a high-speed spindle and the other end goes through a fixed ceramic guide positioned a few millimetres above the workpiece. Rotating the electrode improves the achievable aspect ratio and surface finish due to the fact that a rotating electrode helps to remove debris from the work zone and makes the final roughness less dependent on the initial roughness of the electrode. During the EDM drilling process, the electrode wears and this can be compensated for with Z movement while the ceramic guide remains at the same fixed position. When an electrode diameter smaller than the smallest rod available is needed, the electrode section protruding 2
from the ceramic guide can be ground (dressed) on the machine using either a WC block [3], a spinning WC wheel or a special wire grinding device [4, 5]. 3 Accumulation of errors 3.1 Process Definition This work investigates the accuracy achievable with the micro EDM drilling process. Drilling of a small hole in a block of material by single pass machining and using a single dressed electrode will be adopted as an example. The main dimensions of interest are (Figure 1): diameter of the hole X and Y position of the hole with respect to the co-ordinate system. Y X Y X 3.2 Factors affecting hole diameter Figure 1 Dimensions of Interest The achieved diameter depends on the diameter of the effective dressed electrode d and the spark gap g e (Figure 2). d: Effective Dressed Diameter g e : Spark Gap : Produced ole Diameter Figure 2 Achieved diameter 2 g d (1) e The deviation from the nominal () will depend on variations of the spark gap g e and of the effective dressed electrode diameter d. 2 ge d (2) 3
In order to reduce the initial effective diameter D down to a micro effective diameter d, an electrical-discharge-grinding unit is employed as shown in Figure 3. For this study, movement in the dressing process is performed along the Y axis. The distance y unit gives the position of the eroding point in the work area of the machine relative to the machine reference point. The electrode is eroded until the centre of the spindle reaches a target position y d resulting in an effective dressed electrode of diameter d. Taking into account the spark gap g d between the electrode and the dressing unit, the obtained effective diameter d is defined by equation 3. Y y d d/2 g d Dressing Unit y unit Erosion point X Figure 3 Dressing process d 2 ( yd yunit gd ) (3) The variation in the effective dressed diameter d (d) will depend on the variation in the position of the grinding device y unit, the variation in the positioning of the centre of the electrode y d and the variation of the spark gap when grinding g d. d 2 ( yd yunit gd ) (4) Finally the variation in the diameter of the hole drilled by a single dressed electrode will be determined by equation 6: 2 2 g e y g e d y y d unit g y d unit g 3.3 Factors affecting the position of the hole d The position of the hole is described by the following equations directly derived from Figure 4. (5) (6) X X pos X set (7) Y Y pos Y set (8) 4
Y X set X X pos Y Y set X Y pos Figure 4 ole position description In order to set up the workpiece position in the work area, an electrode of nominal effective diameter (D) is employed. It should be noted that the use of probes or other setting devices is ruled out because it would require reattachment of the high speed spindle and readjusting of the ceramic guide and therefore introduce more errors. The set-up process is represented in Figure 5. D/2 g meas X meas workpiece X set Figure 5 Set-up process X X pos ( X meas D / 2 gmeas) (9) Y Y pos ( Ymeas D / 2 g meas ) (10) The deviations are respectively: X X X D / 2 g pos pos meas meas Y Y Y D / 2 g meas meas (11) (12) The accuracy of the position of the hole will depend on the accuracy of positioning of the machine (X pos, Y pos ), the accuracy of detecting contact signal with the surface (X meas, Y meas, g meas ) and the variation in the initial effective diameter of the electrode (D). 4 Factors affecting the accuracy 4.1 Accuracy and repeatability of positioning (yd and Xpos, Ypos) The accuracy and repeatability of positioning of the machine were measured according to BS ISO 230-2:1997 and the results are given in Table 1 for the three axes of the machine. 5
The accuracy of machine movement was measured using a laser interferometer. The parameters affected by the accuracy of machine movement (y d and X pos, Y pos ) are discussed in the following sections. 4.1.1 Dressing position (y d ) y d is the target dressing position to be reached by the centre of rotation of the electrode relative to the machine reference point during the dressing process. It is defined by the operator (or by a program) in order to obtain a specific effective dressed diameter d and therefore a specific hole diameter. Variation y d will arise due to the machine accuracy and repeatability of positioning. An obvious way of reducing y d during the dressing process is always to approach the position from the same direction (unidirectional approach). Another way of limiting the error is to identify an area on the machine and to fix the dressing unit where the repeatability of positioning is at its best. For instance, when focusing on one single spot in the machine where the dressing unit is placed, the calculated repeatability of positioning according BS ISO 230-2:1997 in this case is 2.2m, which is much better compared to values given in Table 1. The measured y d is 1.98 m. 4.1.2 ole position (X pos and Y pos ) X pos and Y pos are the coordinates of the hole in the machine coordinate system. The position of the hole depends on the workpiece and its position in the work area. Thus the only way to improve X pos and Y pos is to adopt a unidirectional approach to the hole. To machine micro holes, multiple dressed electrodes might be required and therefore the accuracy of positioning of the machine will mainly affect the position of the hole while the repeatability of positioning will impact on the size and shape of the hole. 4.2 Spark gaps and effective electrode diameter (g e, g d and D) 4.2.1 Gap between electrode and workpiece (g e ) g e is defined as the spark gap between the electrode and the workpiece. Its nominal value is fixed by the chosen pulse of the generator and the dielectric used. In conventional EDM, the selection of a pulse is directly linked with the removal rate and surface roughness required. In micro EDM, electrode wear is another important criterion which also needs to be carefully considered. In addition, in order to achieve the micro features, the spark gap should be minimised. Variations in g e (g e ) bring random errors which can appear due to flushing conditions and lack of surface/material integrity [5]. 4.2.2 Gap between electrode and dressing unit (g d ) g d is defined as the spark gap between the electrode and the dressing unit. 6
As in the case of g e, the value of g d is fixed by the chosen pulse parameters and dielectric material, and variations in g d (g d ) can arise due to flushing conditions and lack of surface/material integrity. The pulse parameters are selected depending on the surface roughness required on the electrode and on the speed of dressing. Because the electrode is rotating, its surface roughness should not significantly affect the roughness of the machined surfaces. owever, due to the small dimensions involved, a high roughness will affect the strength of the dressed electrodes, which could break during the process. Estimation of g d is difficult but it was assumed that it would not exceed g e in the worst case. This is because, during dressing, sparking conditions are better than during drilling itself as dressing involves single point sparking with better flushing conditions. 4.2.3 Effective diameter (D) The effective diameter of the electrode D is determined by the initial diameter D init of the electrode (WC rod) and the assembly conditions between the electrode and the ceramic guide. The difference between the diameter D init and the diameter of the ceramic guide D guide creates a gap that introduces potential errors as shown in Figure 6. D init D guide L guide z guide Gap Figure 6 Effect of gap between guide and electrode Thus, variations in the effective diameter (D) can occur, which reflect the tolerance of the electrode and the assembly conditions between the electrode and the ceramic guide. Based on the parameters shown in Figure 6, the maximum variation in effective electrode diameter is defined by the equation 15: D guide D L guide init _ min z guide (13) and D D guide D (14) 2 init _ min 7
Thus, D D guide D D guide init _ min guide 2 Dinit _ min Lguide z (15) In the above equations, D guide is the diameter of the guide, D init_min is the minimum diameter of the initial electrode according to the manufactured tolerance and z guide is the length of the electrode protruding from the ceramic guide. In the case experiments discussed in this paper, the diameter of the electrode was D init = 0.1460.002mm, thus D init_min = 0.144mm, and the measured diameter of the ceramic guide was D guide =0.154mm. L guide was 12mm, and z guide was within 2mm. Based on those values, the calculated maximum deviation D was 13.3m. owever, this maximum variation only occur when the position of the electrode within the guide is modified to a number of extreme positions. This is only possible when there is significant movement of the electrode along the X and Y axes relative to the guide. This is highly unlikely. To support this point, two cases are considered. First, when using a non-rotating electrode, the position between guide and electrode should not change because the only movement between guide and electrode is in the Z axis and no force acts along the X or Y axis that would change this position. Significant change should only occur when altering the position of the guide with respect to the head holding the electrode, for instance, when replacing an electrode or after a significant movement in the Z direction because this would affect the angle at which the electrode enters the guide and therefore might create a significant movement along the X and Y axes. Also, changes could arise due to the tolerance of the manufactured electrode, but the variation of diameter along the length of a single manufactured rod is considered negligible. Therefore, it can be assumed that, for a small feed and without change of electrode, the variations in effective diameter when using a non-rotating electrode are negligible, D 0. Second, when the electrode is rotating, X-Y movement can occur due to the rotation and friction between guide and electrode. owever, it can be assumed that this movement follows a cycle in phase with the rotation, resulting in a small increase in effective diameter but with negligible variations between two periods of the rotation cycle, therefore D 0. It should be noted that the length of the dressed section of the electrode should be smaller than z guide. This is to avoid the dressed part of the electrode touching the guide, as this would increase the potential error and the dressed electrode could be damaged. owever, this introduces another limitation in the depth achievable using a dressed electrode. 4.2.4 Estimation of g e and g e An experiment was conducted to estimate g e and D. A 150m WC electrode was used to drill two series of 50m deep holes. The experiment was conducted with rotating and stationary electrodes and the results are given in Figure 7. 8
172 170 168 166 164 162 160 3.75m 1 2 3 4 5 6 7 8 9 10, Rotating electrode, Still electrode Figure 7 ole produced with rotating and stationary electrodes The process of erosion of a hole is represented by the following equations: 2 g D (16) e 2 ge D (17) All holes were machined with the same electrode/guide assembly to a depth of 50m, resulting in a total Z feed of only 1mm for the total process. Thus, according to the previous section, it can be assumed that for both series of holes D 0. Therefore, 2 g e g e. 2 The difference between the mean diameter of the two series is 3.75m, which represents the mean increase of effective diameter due to the rotation. From the experiment, it can be concluded that g e = 1.1 m when the electrode is rotating and g e = 1.6m when the electrode is not rotating. As seen in Figure 7, when the electrode is rotating g e is smaller, which might be due to the better flushing conditions and a better way of removing debris from the cutting zone. 4.3 Temperature instability error (y unit ) y unit is the position of the point of erosion on the dressing unit in the machine co-ordinate system. Changes in the temperature in the room and in the machine structure create variations in the relative position between the rotating head and the table of the machine and therefore affect the position of the dressing unit with respect to the machine zero point, y unit, and to the electrode. An example of that error is shown in Figure 8. Measurements of displacement along the Y axis were taken using a laser interferometer during the machine warming up cycle. The obvious way to minimise the variation is to work in a temperature-controlled room and to ensure thermal-stability of the machine structure. Each machine should be tested to find out the time for the temperature of the machine to stabilise for certain ambient conditions and the temperature relative deviation of each axis should be measured in order to plan electrode dressing with minimum error. 9
Tim e ( h ) 0 0 5 10 15 20 26-1 25.5 Tem perature Shift along Yaxis ( m ) -2-3 -4-5 Shift 25 24.5 24 23.5 Temperature ( ºC ) -6 23-7 22.5 Figure 8 Shift of the relative position between the rotating head and the table along the Y axis due to temperature variation Under certain conditions (temperature-controlled room and minimum dielectric temperature variance, short working hours) it can be assumed that y unit 0m. 4.4 Measurements errors 4.4.1 Workpiece surface detection error(x meas, Y meas ) During the setting up of the workpiece, when an electrical contact occurs between the electrode and the workpiece, a contact signal is registered by the machine system processor. The processor has set priorities in checking each machine status signal, which means that the checking of the contact signal is not carried out continuously. There is a time interval (t meas ) between each signal check. This causes an error when detecting the position of the measured surface when measuring X meas and Y meas. If the speed of approaching the surface is V meas, the variation will be: X meas or Y meas =V meas * t meas (18) Usually the contact signal is checked every 2-5msec (depending on the controller). Obviously, to minimise the error the speed should be as low as possible but high enough to avoid stick-slip. In this case, t meas is 3msec and the measuring speed is from 20mm/min to 1mm/min. The calculated variation is 0.05 to 1m. 4.4.2 Surface position detection error (g meas, D) During the measuring cycle, voltage is applied between the table and the spindle. The machine moves until an electrical contact is reached. As the surfaces tend to oxidise, a different gap, or different pressure is needed for the spark to break through. All these factors contribute to an error of surface detection introducing variation in the measuring spark gap g meas. To minimise the effect of surface detection error, a measuring probe can be used. The variation using such a probe on a WC block approaching the surface with a very low speed of 1mm/min was measured to be g meas = 1m. As explained earlier in the paper, it will be difficult to remove the high-speed spindle and the ceramic guide to perform the measurement with the probe and then replace them on the machine, because more errors will be introduced. The electrode itself can be used to do the measurement instead of the probe. In this case D will be included in the measurement as well. Unlike for the erosion or dressing process, D is not negligible. This is because as mentioned 10
in section 4.2.4, the rotation of the electrode creates a cyclic movement in X and Y. Therefore, contact between electrode and the surface might occur at different positions in the cycle, as shown in Figure 9. Y c_min y c_max a b Figure 9 Surface position detection with D influence The accuracy of measurement is dependent on the speed of measuring. The lower the speed of the approach is in relation with the speed of rotation of the electrode, the smaller the error will be. This is comfirmed by an experiment, where the variations of surface detection on a WC block with 150 m WC electrode using different speeds were measured. For speeds of 20mm/min, 5mm/min and 1mm/min (the lowest speed on the machine), (g meas +D/2) is respectively equal to 5.7m, 3.9m and 3m. 5 Experimental set up The experiment consisted of producing two series of 10 holes, for 5 different dressing positions. The first target dressing position was y d_init and the 4 others were respectively (y d_init - 10m), (y d_init -20m), (y d_init -30m), (y d_init - -40m). The measured values are given in table 2. A reduction of 10m in target dressing position y d should result in a reduction of 20m in diameter for the produced hole. In the experiment, the mean differences between each series of diameter were 20.49, 19.55, 21.31, 19.08, which gives a variation of 2.23 m. This shows the potential for an accurate dressing process using EDM grinding. 6 CONCLUSIONS This paper has given an overview of the factors affecting the accuracy during micro EDM drilling and the results of the conducted experiments can help to plan the process within the expected tolerances. The following conclusions may be drawn: The machine used for micro EDM drilling should be placed in a temperature-controlled environment with a constant ambient and dielectric temperature. If a grinding device is used, the type of device should be justified and its position should be selected after careful investigation of the geometrical accuracy of the machine. Tests should be made in advance for the preferred sparking conditions and the spark gap deviation should be measured. Speed is the main factor contributing to errors when using measuring cycles. 11
When assigning process tolerances for micro EDM drilling all activities during the process, such as type of electrode grinding, type of positioning and duration of the operation, should be considered. All these activities will accumulate errors, which should be taken into account. This paper has discussed the level of accuracy achievable with micro EDM drilling which has been studied as the first step in investigating micro EDM milling strategies. AKNOWLEDGMENT This work was carried out as part of the ERDF (Objective 1) project Supporting Innovative Product Engineering and Responsive Manufacturing (SUPERMAN), the ERDF (Objective 2) project Micro Tooling Centre, the EC Network of Exellence 4M and the EC Network of Exellence I*PROMS. REFERENCES: [1] o K.., Newman S.T. State of the art electrical discharge machining (EDM). International Journal of Machine Tools&Manufacture, 43, 2003, pp. 1287-1300. [2] Ehrfeld W., Lehr., Michel F., Wolf A., Gruber.P., Bertholds A. Micro Electro Discharge Machining as a Technology in Micromachining. SPIE, 2879, 1996. [3] Ravi N., Chuan S.X. The effects of electro-discharge machining block electrode method for microelectrode machining. Journal of Micromechanics and Microengineering, 2002, pp. 532-540. [4] Masuzawa T, Fujino M, Kobayashi K, Suzuki T. Wire electro-discharge grinding for micro machining. Annals of CIRP 1985, 34(1). [5] Lim.S., Wong Y.S., Rehman M., Edwin Lee M.K. A study on the machining of highaspect ratio micro-structures using micro-edm. Journals of Materials Processing Technology, 140, 2003, pp. 318-325. [6] Radjurkar K.P., Wang W. M. Improvement of EDM Performance with advanced monitoring and control system. Transactions of the ASME, 119, 1997, pp.770-775. Repeatability of positioning Unidirectional A Unidirectional A Bidirectional 3.76m 2.95m 5.33m Repeatability of positioning Unidirectional B Unidirectional B Bidirectional 4.49m 4.92m 7.83m Repeatability of positioning Unidirectional C Unidirectional C Bidirectional 1.48m 1.27m 1.9m Accuracy of positioning Unidirectional A Unidirectional A Bidirectional 14.03m 11.85m 15.7m Accuracy of positioning Unidirectional B Unidirectional B Bidirectional 5.03m 5.18m 7.91m Accuracy of positioning Unidirectional C Unidirectional C Bidirectional 2.39m 1.82m 2.73m X axis Y axis Z axis Table 1 Positioning results y d_init y d_init -10m y d_init -20m y d_init -30m y d_init - -40m exp 1 19.4m 3.6m 6.0m 13.9m 11.9m exp 2 12.2m 12.2m 9.4m 13.9m 8.6m Table 2 Experimental 12