A spark erosion apparatus
Ever since I broke a tap whilst making the Improved chuck for taps I have been thinking of making a simple spark erosion apparatus. When making the hexagonal hole in the mill quill lock lever I again thought that here was another application where a simple spark erosion machine could help. I can also see potential applications in drilling HSS.
My interest was awakened further when I came across an article entitled " A simple self-acting spark erosion machine" ( http://www.stockportsme.co.uk/Membersprojects/Spark_eroder.html ) by Derek Lynas. The concept of this machine is very simple. It consists of a sliding spindle, carrying the tool, attached to an armature activated by a coil. The spindle is connected to a capacitor that is charged through the coil and an external resistor. As the capacitor is charged the charging current activated the coil and lifted the spindle. When the capacitor approaches maximum charge the charging current reduces and eventually the force holding the armature in the coil is weakened so much that the spindle is released and it drops under gravity. On hitting the workpiece the capacitor is discharged, causing sparking between the tool and the workpiece. At the same time the coil re-energises and the spindle lifts from the workpiece. The whole cycle then repeats. The net result is that the sparks erodes away the metal and the tool gradually sinks into the workpiece.
During use the work and the tool are submerged in a bath of dielectric fluid. Distilled water or kerosene are commonly used.
I built the machine as described by Derek Lynas and operated it from a 21 volt power supply. I was disappointed with the erosion rate which was around 1-2 mm/hour with a 3mm tool. Derek's version operated from a 50 V power supply and probably achieved a much higher erosion rate. I did contemplate upgrading my power supply to 50 V but I decided first to see if the performance at 21V could be improved.
First I will describe the mechanical parts of the machine and after I will describe a new electrical system that greatly improves the performance.
My interest was awakened further when I came across an article entitled " A simple self-acting spark erosion machine" ( http://www.stockportsme.co.uk/Membersprojects/Spark_eroder.html ) by Derek Lynas. The concept of this machine is very simple. It consists of a sliding spindle, carrying the tool, attached to an armature activated by a coil. The spindle is connected to a capacitor that is charged through the coil and an external resistor. As the capacitor is charged the charging current activated the coil and lifted the spindle. When the capacitor approaches maximum charge the charging current reduces and eventually the force holding the armature in the coil is weakened so much that the spindle is released and it drops under gravity. On hitting the workpiece the capacitor is discharged, causing sparking between the tool and the workpiece. At the same time the coil re-energises and the spindle lifts from the workpiece. The whole cycle then repeats. The net result is that the sparks erodes away the metal and the tool gradually sinks into the workpiece.
During use the work and the tool are submerged in a bath of dielectric fluid. Distilled water or kerosene are commonly used.
I built the machine as described by Derek Lynas and operated it from a 21 volt power supply. I was disappointed with the erosion rate which was around 1-2 mm/hour with a 3mm tool. Derek's version operated from a 50 V power supply and probably achieved a much higher erosion rate. I did contemplate upgrading my power supply to 50 V but I decided first to see if the performance at 21V could be improved.
First I will describe the mechanical parts of the machine and after I will describe a new electrical system that greatly improves the performance.
The unit comprises the coil shown at the top of the photo. Beneath it is a bearing block in which the spindle slides. The spindle is terminated at the upper end by an armature which is a close fit in the bore of the coil. Both of these parts are held on a backplate which is insulated from the stand by the black Delrin block that clamps to the rod on the right hand side.
The coil former was made from two perspex cheeks, 40 mm diameter with a 12 mm centre hole, and a Delrin tube 13 mm diameter and 10 mm bore. The ends of the Delrin tube were shouldered to fit inside the cheeks. The coil was originally 1500 turns of 28 SWG enamelled copper wire but I subsequently reduced this to about 1000 turns in order to reduce the resistance of the coil. The mild steel frame surrounding the coil forms a magnetic circuit and it is all held together with M3 screws.
The top mild steel plate carries a 10 mm round x 5 mm thick steel insert that enters the top of the coil. The bottom face of this is covered with a 10 mm diameter x 1mm thick disc of EPDM rubber. This rubber disc serves three purposes. It stops the armature from sticking to the steel insert, it softens the blow of the armature when it hits the insert, and it provides some "bounce" enabling the armature to rebound from the insert.
The top mild steel plate carries a 10 mm round x 5 mm thick steel insert that enters the top of the coil. The bottom face of this is covered with a 10 mm diameter x 1mm thick disc of EPDM rubber. This rubber disc serves three purposes. It stops the armature from sticking to the steel insert, it softens the blow of the armature when it hits the insert, and it provides some "bounce" enabling the armature to rebound from the insert.
The spindle is a piece of 6mm mild steel rod with a 9 mm mild steel armature at one end. These are two separate components that were Loctited together.
The bearing block is made from a piece of steel 50 x 38 x 12 mm. This was drilled through with a 6.5 mm drill and the ends counterbored to 8 mm for a depth of 10 mm. Two brass inserts were drilled out to 5.9 mm and the external diameter reduced to be a tight press fit in the counter bores. Once pressed into position the brass inserts were reamed out to 6 mm. The final operation is to mill away the central portion of the block.
The two holes in the block are for the anti-rotation guide plate.
The two holes in the block are for the anti-rotation guide plate.
The anti-rotation collar is formed from a piece of 12 mm round brass. This was drilled out 6 mm and a 1 mm wide slit cut with a slitting saw to a depth of 1.5 mm. At right angle to the slot the collar is cross drilled and tapped M3 for a screw that locks the collar to the spindle.
The securing screw is also used to provide an electrical connection to the spindle via a flexible wire.
The securing screw is also used to provide an electrical connection to the spindle via a flexible wire.
The anti-rotation guide plate is made from a piece of 6 x 25 mm steel. A 1 mm slit was made along one edge and a piece of 1mm wide sheet steel epoxied into it. The steel was then filed down so that it protrudeed about 1 mm.
This steel strip engages in the slit in the anti-rotation collar and prevents the spindle and tool rotating.
It is important that the spindle slides freely in the bearing block and that the anti rotation collar slides freely on the steel strip. The spindle must fall freely when lifted and released.
This steel strip engages in the slit in the anti-rotation collar and prevents the spindle and tool rotating.
It is important that the spindle slides freely in the bearing block and that the anti rotation collar slides freely on the steel strip. The spindle must fall freely when lifted and released.
The tool, here a 4 mm diameter copper rod, is attached to the end of the spindle with a simple connector clamp. The tool can be almost any shape. Copper is the preferred material since it erodes very slowly but brass can also be used.
The stand was simply a piece of 6mm steel plate 125 mm x 75 mm with a piece of 10 mm steel rod screwed to it.
Various holes were drilled in the base to permit clamping of the work to the base.
The M6 brass screw at the rear of the unit is to provide an electrical connection to the stand and workpiece.
Various holes were drilled in the base to permit clamping of the work to the base.
The M6 brass screw at the rear of the unit is to provide an electrical connection to the stand and workpiece.
Electrical circuits
The basic circuit as described by Derek Lynas is shown above with one small modification. In his circuit a 6.8R 10 watt resistor is used in place of the 12V 50W lamp. When power is applied to the circuit current flows through the coil to charge the capacitor. This current lifts the armature and spindle. As the capacitor charges the current through the coil decreases until the armature is not longer attracted sufficiently to the coil thereby allowing the spindle to drop. This discharges the capacitor creating sparks at the tool/ work piece junction that erode the work. The current can once again flow through the coil and the spindle is lifted and the whole cycle repeats itself.
There were several reasons for substituting the resistor for the lamp. The first, and most important, was that my 6.8 ohm resistor originally used fried itself very rapidly (I don't understand why this should be since Derek operates his apparatus on 50 Volts and the current through his resistor must be even higher). A second reason for the change is that lamps have a very low resistance when cold. This means that the capacitor can charge up more quickly and thus potentially the frequency of operation should be higher. A third reason for changing the resistor is that the brightness of the lamp provides a visual indication of the current through the circuit.
This basic circuit provided low erosion rates, of the order of 1-2 mm/hour with a 3 mm round brass tool. In order to try to improve on this the following circuit was tried.
In this circuit a second lamp has been added shunting the coil. This provides a second charging path for the capacitor and again this should increase the current into the capacitor and hence the frequency of operation and the erosion rate. This made some slight improvement to the erosion rate but it was not dramatic. The shunt resistance cannot be too low because there will then be insufficient voltage on the coil to lift the armature.
There are a number of factors that influence the erosion rate. Firstly there is the amount of charge stored in the capacitor. This influences the intensity of the sparks. However, there is a limit to how much the capacitance can be increased because if the sparks become too intense then the tool tends to weld to the workpiece. The second factor that influences erosion rate is the frequency of operation. At low frequencies this will depend the value of the resistance and capacitance. However, at higher frequencies the frequency is determined by speed with which the spindle can be lifted and dropped i.e. the natural frequency of the oscillating spindle. This in turn is influenced by the distance the spindle has to rise and fall. Thus higher frequencies can be sustained when to oscillation distance is small. It is noticeable that when using the apparatus that the frequency of operation slows down as the tool penetrates into the workpiece because the distance travelled increases. It is clear that the operating criteria for erosion are complex and constantly changing as erosion occurs.
Other workers (e.g. Eric Rumbo in an article entitled "A simple spark erosion machine", Model Engineers's Wokshop 117, July 2006) have attempted to drive the solenoid independently of the charge discharge circuit using an astable multivibrator based on the 555 chip. This enables the charge discharge circuit to be optimised for good sparking but being a constant frequency the coil drive does not make best use of the time i.e. there are times when the drive frequency could be higher than the set frequency. Also the circuit can waste current by allowing the tool to remain in contact with the workpiece for too long.
There are a number of factors that influence the erosion rate. Firstly there is the amount of charge stored in the capacitor. This influences the intensity of the sparks. However, there is a limit to how much the capacitance can be increased because if the sparks become too intense then the tool tends to weld to the workpiece. The second factor that influences erosion rate is the frequency of operation. At low frequencies this will depend the value of the resistance and capacitance. However, at higher frequencies the frequency is determined by speed with which the spindle can be lifted and dropped i.e. the natural frequency of the oscillating spindle. This in turn is influenced by the distance the spindle has to rise and fall. Thus higher frequencies can be sustained when to oscillation distance is small. It is noticeable that when using the apparatus that the frequency of operation slows down as the tool penetrates into the workpiece because the distance travelled increases. It is clear that the operating criteria for erosion are complex and constantly changing as erosion occurs.
Other workers (e.g. Eric Rumbo in an article entitled "A simple spark erosion machine", Model Engineers's Wokshop 117, July 2006) have attempted to drive the solenoid independently of the charge discharge circuit using an astable multivibrator based on the 555 chip. This enables the charge discharge circuit to be optimised for good sparking but being a constant frequency the coil drive does not make best use of the time i.e. there are times when the drive frequency could be higher than the set frequency. Also the circuit can waste current by allowing the tool to remain in contact with the workpiece for too long.
The circuit above shows the method that I have used to control the oscillation of the spindle. The RC network is on the left. As the capacitor charges then the voltage on the spindle rises. The voltage detector on the right monitors the rise in spindle voltage and when it reaches a certain value then it turns off the current to the coil allowing the spindle to drop and discharge the capacitor through the work. If the trigger voltage is set low then the the spindle will oscillate quickly whereas if it is set high then it will operate slowly. Even after the coil is switched off the capacitor continues to charge whilst the spindle is falling so the voltage on the capacitor at the moment of impact of the tool on the work will be considerably higher than the trigger voltage. The advantage of this approach is that the capacitor is always charging except for the very brief period (about 2 milliseconds) when the tool is actually in contact with the work and that the oscillation frequency is always the highest that can be sustained for the set trigger point (i.e. no time is wasted). On the RC side of the circuit the coil has no influence on the rate of charge and the value of R and C can be optimised for maximum erosion rate, as mentioned below.
The circuit above shows the voltage detector. The two transistors, Tr1 and Tr2, form a classic long tailed pair differential amplifier whilst Tr3 and Tr4 form the output stage driving the coil. The diode across the coil suppresses the reverse voltage across the coil and the LED provides a visual indication of the average charging current of the capacitor C in the charging circuit. A cheap op amp could have been used as the voltage detector but this would have required a separate power supply since the maximum operating voltage of a 714 op amp is 18V. Using discrete components avoids this complication
This circuit, without the LED and its series resistor was constructed on a small piece of stripboard as shown below:
This circuit, without the LED and its series resistor was constructed on a small piece of stripboard as shown below:
The stripboard is mounted directly on the potentiometer. No heatsink is necessary for the power transistor Tr4 because it is either on or off and little power is dissipated. Connections to the board are via the small connector block in the lower left hand corner.
The electronics are mounted in an MDF box as shown here. At the lower right is the 21V transformer with the rectifier diodes attached. Above is the stripboard circuit attached to the potentiometer. To the left are the charging components for the capacitor. These are mounted on a piece of 3A "choccy block". This is attached to the front panel by an aluminium stand off bracket. The series resistor comprises 4 x 12V 50W tungsten filement lamps connected in series. Two of these can be shorted out by a switch to provide two selectable charging resistances. At the lower right of the choccy block are the two capacitors used in the R-C circuit. These are 470 uF 50 V components. Either one or both can be selected by a switch giving two selectable capacitances. To the far left of the choccy block is the red 2200 uF reservoir capacitor that connects across the diodes mounted on the transformer.
The front panel of the unit showing the arrangement of the controls. Connection to the coil is via a DIN speaker socket and connections to the tool are via 4mm banana plugs.
Note the front panel is square. The barrel distortion of the image is an artefact of the camera.
Note the front panel is square. The barrel distortion of the image is an artefact of the camera.
This shows the rear of the control box. The fan and vent provide cooling inside the box. The fan is a 50 mm 12 V fan salvaged from a computer. It is connected across the top two lamps in the CR circuit. Thus as the current , and consequential heat, increases so does the fan speed.
With this controller erosion rates of around 9 mm per hour using a 3 mm diameter brass tool are possible. This is not fast but definitely more usable than the 1-2 mm/hour achieved with the original circuit.
This photo shows some holes "drilled" using the spark eroder. On the left is a piece of 3 mm thick steel, in the middle a piece if 1 mm sheet and on the right a piece of hacksaw blade. In all cases the hole diameter is 4 mm. The thicker piece took circa 30 minutes to erode the hole whereas the others were completed in a few minutes.
I have used deodourised kerosene (white spirit) as the dielectric fluid. The alternative, distilled water, I have avoided because of potential rust problems.
I have used deodourised kerosene (white spirit) as the dielectric fluid. The alternative, distilled water, I have avoided because of potential rust problems.
There is scope to improve the apparatus further. A higher voltage power supply would undoubtedly increase the erosion rate. A mechanism to lower the coil/spindle assembly during erosion to maintain only a short amplitude of vibration would also speed up the process.
Disaster strikes!
In my quest for higher erosion rates I have succeeded in burning out the transformer. This was only rated at 1.2 amps. I decided to upgrade the transformer and fit an ammeter into the circuit. After checking the costs and size, both of which were substantial, of an upgraded 4 amp transformer I then thought about using a switch mode 24V laptop computer power supply. A suitable unit rated at 24V and 100W was purchased on ebay for around £8. A 0-5A ammeter was purchased on ebay at the same time for about £4.
The laptop power supply has a voltage selector enabling voltages between 12 and 24 V to be selected. It is small and self contained and can be relocated outside the control box. The creates plenty of space inside the existing box to fit the ammeter.
The laptop power supply has a voltage selector enabling voltages between 12 and 24 V to be selected. It is small and self contained and can be relocated outside the control box. The creates plenty of space inside the existing box to fit the ammeter.
The Mk 2 version.
The above photo shows the new front panel with the ammeter installed. The old front panel was reused. After disconnecting all the plugs etc. it was removed and a hole for the ammeter cut using a holesaw. The potentiometer was relocated under the ammeter and the unit reassembled. No changes were made to the circuit other than taking out the transformer and diodes. The laptop power supply plugs into the back of the box via a dc socket.
This photo shows the laptop power supply. The power supply is stabilised and has overload protection.
Making this change has increased the performance of the unit somewhat. I think the improvement results because the voltage of the new unit does not sag under high load conditions.
Further innovations.
Two further innovations have been made to the spark eroder. These are shown in the photo above. A weight has been added to the oscillating shaft. This helps to overcome any friction in the bearing resulting in a shorter fall time for the shaft and this in turn increases the frequency of oscillation.
It was also found that the weight of the wire going to collar on the shaft was having an effect on the oscilation. The connection to the shaft is now via a short coil of wire to eliminate this effect.
It was also found that the weight of the wire going to collar on the shaft was having an effect on the oscilation. The connection to the shaft is now via a short coil of wire to eliminate this effect.
The coil of green wire connects to the contact screw on the shaft collar and then goes to an additional third contact on the "choccy block" attached to the side of the bearing block where connection is made to the control box. The only forces that can affect the oscillation of the shaft now result from just the springiness of the wire coil. This is constant, whereas before the weight of the wire connecting directly to the shaft would vary according to its position etc.