Things I want or need to know (feel free to tell me the answers):\n\n* what are the typical machining pulse times and currents for thin wire EDM?\n* what wire can I buy and on what spool sizes?\n//I can get Berkenhoff wire in 5000M lengths, at about 200GBP err, too much wire and too much money :(//\n* Does carbide rust? I want to make guides from it perhaps.\n//no//\n* Why are most datasheets so badly written and why can't I read them even when they are not?\n
TheStart\n\n
\n''Control:''\n\nMultiMediaCard\nPcEndElectronics\nMachineElectronics\nThePic\n\n''Power''\n\nSparkGenerator\nGapControl
Gap control is a black art, mainly because you cannot measure the gap width directly (except for specific experiments).\n\nWhat are gap conditions?\n\nThe generator applies high voltage pulses between the electrode and the workpiece. If the gap is large then the voltage seen across the gap will be the same amplitude as the supply's and the current through the gap will be zero. If the gap is small enough then ionization will begin to occur and current will flow. This is the spark and this is what causes machining. However, if the gap is too small ''and/or'' the dielectric between the electrode and workpiece is not clean enough then the current can take a path of least resistance through the gap and a high current arc is seen. At even smaller gaps a short can occur. \n\nToo large a gap and unused pulses mean wasted time and slow machining. Too small a gap or inadequate flushing means arcing which damages the workpiece and electrode (causing spots of carbon and eventual break down of the machining process).\n\nThere are many ways to monitor the gap but most look at the voltage or current pulses seen. Some others send extra pulses to test the gap between the machining pulses but these can themselves affect the gap conditions and require longer off times for full de-ionization of the gap.\n\nMechanisms of monitoring the gap include:\n\n* Average voltage - If you low pass filter the voltage pulses these can be applied to comparators to give signals indicating too large a gap (high average v), arcs/shorts (too low a voltage) and correct machining (between the two)\n\n* Ionization time - There is normally a delay between applying the voltage and the breakdown of the dielectric in the gap. Small times indicated too small a gap, large times indicate too large a gap. This method is now very common but has been shown to be quite variable even with an optically measured constant gap. This also does not work well with current source supplies which raise the starting voltage until break down occurs if it can, these lead to constantly short times to break down.\n\n* Break down voltage- after the ionization there should be a nominal machining current. This should be within certain limits. Too low means too large a gap and too high means too small a gap and arcing. Of course the voltage across the gap is proportional to the current can be used for these measurements.\n\nThe indirect measurement of the gap conditions via the electrical signals is one cause of the "black art" nature of gap control but the other is the fact that the flushing cannot be relied upon to be constant and there are also other things such as the production of bubbles and material variations in both electrode and workpiece. Wire vibration and speed variation will also play a part.\n\nFor this machine gap control is a major issue as it will be quite important to have steady machining for good results on a small scale. Because of this although I may begin with simple averaging of the voltage I may move on to something that looks more specifically at the break down voltage. It may also be beneficial to be able to turn off pulses that become arcs half way through as has been shown to be possible.\n\nOnce you have these voltages or times the question becomes what to do with them. The time constant of the machine will be much higher than that of the generator and it may well be possible that discharges of several of these types have occurred between the points when the uP decids if it should move the drives or not. Some systems try to respond to the relative occurrence of each discharge type, keeping of course to a regime where there are more good pulses than open circuit or arcing ones.
This may seem a simple thing but it has been one of the hardest parts. Requirements to consider:\n\n* Low vibration and good stability for precise cutting.\n* Access to workpiece despite small tank.\n* threading through small holes in the workpiece.\n* Potential for 4-axis work (each end of the wire is moved independently for taper cutting)\n* No wet floors with submerged cutting\n\n''Move the tank:''\n\nBoth of these designs suffer from the fact that the lower guide is beneath the work, this makes hand threading very difficult.\n\n//Through the side//\nThere is a very common configuration where the tank sits on an x-y stage. The lower guide it mounted in an arm and this protrudes through the side of the tank. The lower arm is securely fixed to the machine base. The upper arm is often on an X,Y,X stage to provide control over the upper nozzle height and to allow for taper cutting.\n\nThe problem with this method is the seals that are required to stop the tank emptying on to the floor where the arm enters. This is normally done with a sliding plate seal and a bellow. On my tiny machine the increased rigidity would be offset by the vibration caused and the difficulty in building the thing. \n\n//Over the top//\nThis means two arms enter the tank over the top, one for the lower guide which does not move and one for the upper with the same x,y,z controls. This system of course works and can be seen on several old Agie machines however to me it is clumsy and leads to a large tank and large over hangs of the arms, a potential cause of vibration.\n\n''Move the work:''\n\nMoving the work in the x,y plane would be much like the previous method however if you hang the work vertically then things are different. To start with the tanks dimensions can be smaller, it does not matter if the work leaves the surface of the water unlike leaving via the side of the tank!. So you basically have your wire in the tank and the work held vertically from the top by a, err, y-z stage. Then there is the fact that the two arms that hold the wire guides can have zero overhang, in fact they can be bolted to or integral to the tank. 4-axis is possible by having one moving arm and one fixed. Wire threading is also a lot easier as the guides are at either side of the work piece.\n\nTo cap it all off if the two arms holding the guides are not attached to the tank then it is a simple matter to make a drop tank. You put the tank on a simple jack system and let if drop away from the work and wire. This allows complete access to the workpiece and threading, areal bonus for micro work.\n\n''Move the wire''\n\nJust like the above but you have both wire guide arms mounted to the y-z stage. The work can then be mounted to either a third arm or to some work holding structure in the tank. This method is very good for 2D as it allows easy threading, a drop tank option and efficient tank use but will not lend itself to 4-axis work.\n\n''CONCLUSION''\n\nI think moving the work with a horizontal wire and vertical workpiece is the best option along with a simple drop tank. Oh... I have just reinvented the [[Makino UPJ-2|http://www.makino.com/edm/wire_edm/technical_articles/articles.asp?id=968&title=Wire+EDM+Goes+Horizontal+]] typical!!! Only after deciding for myself this was the best option did I make the leap that in the same way as the horizontal wire makes auto threading easy for them it will make manual threading easy for me.\n\nI have been sketching in Rhino, some results:\n\n[img[one|tank2.jpg]]\nProbably not too clear. There are two pairs of two stages, each arranged as a pair but mounted to give y-z (up/down, back/forth). On one pair of stages an arm with the workpiece (orange) is attached. To the other pair of stages one of the wire guide arms is attached, this will allow 4-axis (tapering) cutting by moving one end of the wire by small amounts. The other end of the wire goes to a fixed arm attached to the frame of the machine, this can be seriously stiffened, these are just sketches really.\n[img[one|tank1.jpg]]\nSame thing again really, you can see the workpiece is lifted as high as the stage can move it. I have drawn this with a 3"X3" workpiece, the machine can move 2"X2" but I guessed the ability to hold a larger object might be handy. Obviously the tank is floating in the middle of no where, it just needs a simple mechanism to lift it. It all looks a bit tall and wobbly but the sides of the frame are only 290mm long, less than a foot.\n[img[one|back.jpg]]\nYou get a clear view of the stepper motors on the other set of stages in this picture. The back plate is designed to make it stiff although I think it may need a slight redesign as the tank may need to come back a little more than drawn.
\nThis part is "simple" a PIC reads out the pulses contained on the MMC at a rate dictated by the feedrate oscillator and tthe gap condition feed back hard ware to the motor drives. \n\nThe IO for the PIC consists of:\n\n''Start/continue/pause'' - allow the user to start after a pause or instigate a pause.\n''Advance'' - If this input is high the gap is too big.\n''Retreat'' - If this input is high the gap is too small.\nstepping clock - This tells the pic when to load the next set of steps and directions to the drives. \n''Xs'' - Step output for x axis\n''Xd'' - Step output for x axis\n''Ys'' - Step output for x axis\n''Yd'' - Step output for x axis\n''wire feed'' - wire runs when high\n''Machining enable'' - Output to enable the generator.\n\nOther:\n\nUser interface:\n\nTo begin with while getting the machine up and running a user interface will not be required, a single button to start and pause the program should be enough. However once the machine is being used for real machining more control functionality will lead to greater ease in use and practicality, the possible interface components include:\n\nJogging in all axes\nstart/pause\ntoolpath visualization/verification.\n
\nSome general info:\n\nI have some small stages with 2" travel and high quality ball slides and ball screws. The lead of the screw is 1mm which allows very high resolution movement. The drives are driven by small stepper motors and some micro stepping step/direction drives.\n\n*MachineConfiguration\n\n*TankAndDielectric\n\n*WireGuidingAndTensioning\n\n*WireThreading\n
TheStart\n\nProjectProgress\nCurrentQuestions\n\nTheConcept\nSoftWare\nElectronicStuff\nMachineMechanics\n\nReSources\n\n\nWikiEditing\n\n\n\n\n<<newTiddler>>\n<<newJournal "DD MMM YYYY">>
An MMC is the type of thing you have in your camera for storing pictures, these days they are very cheap and to me were a great source of cheap memory to dump lots and lots of 1's and o's into. One of the key assumptions made for this machine was low machining speed. I intended at first to use Mach3 to produce the step and direction and save this to the PICs memory until it was full, then pause Mach3 via an OEM inpt (yes mach3 is clever) and just drip feed the pulses. The problem is I want to use my good PC and I want to keep it where it is and not up in the attic.\n\nThe concept is to have a little box with a card slot in and a PIC, the PIC just shoves the gcode onto the card as the PC, PIC and gcode allow. I then walk the card upstairs to the EDM machine and plug the card in. The pic in the EDM then takes on the card as extended memry and starts reading out the step and direction pulses to the motor drives at a rate determined by the gap condition electronics.\n\nIt turns out that linking a PIC to an MMC is not so difficult:\n\nhttp://www.captain.at/electronics/pic-mmc/\nhttp://www.compsys1.com/workbench/On_top_of_the_Bench/MMC_Project/mmc_project.html\n\nIt turns out that MCC cards can operate in two modes, one is called SPI or Serial Peripheral Interface mode and it is designed for microprocessor to mae use of the cards. Some PICS already have serial comms built in so marrying the two is quite simple as the above link shows. Actually it is easier than the link suggests as some of the components used are for other comms with a PC.\n\nHere are some links to info on MMCs:\n\nhttp://en.wikipedia.org/wiki/Multi_Media_Card - A page on wikipedia\n\nAnd from that page:\n\nhttp://www.sandisk.com/oem/mmc.asp - Info from Sandisk (a big mfr of cards)\nhttp://www.kingmaxdigi.com/product/MMC.pdf - A slightly more digestable document on their cards.\n\nthanks to a friend I think I have a way of seeing what data has been put on the card by using some software called winhex however beware as it displays the data in a strange way, paying atting attention to boot sectors and the like.\n\nhttp://www.x-ways.net/winhex/index-m.html\n\n
Put patents here, remember for hobbiests you can use the ideas at will.
At the PC end a pic must be connected to the PC via the parrallel port and write the step and direction signals and other control signals to the MMC card.\n\nThe required IO for this is:\n\n''Xs'' - Step for x axis\n''Xd'' - Direction for y axis\n''Ys'' - Step for y axis\n''Ys'' - Direction for y axis\n''Pause'' - Allow the gcode to pause the machine (for wire threading etc)\n''"spindle"''- Allow the gcode to turn off machining, for rapids between start holes etc.\n''wire feed'' - sets wire running and stops it\n''Program start/stop'' - to indicate the start and end of gcode recording.\n\nAll of these inputs will be recorded to the MMC card (the last one tells the machine when it has got to the end of the code).
Here is where I will say what I have achieved and suggest my plan of action.\n\n''STATUS''\n----\nI have the software to record the pulses from the parallel port of a PC running a step and direction CNC control program to a camera memory card (MMC). I can read this data to the rs232 port for verification and have developed the algorithm for moving back and forth through the blocks of data stored on the card according to the gap conditions. Lack of ram (to use as a buffer) means I need to change pic to an 18FXXXX series and compiler and change to code a bit. I am making progress setting up a Nintendo DS games handheld games machine as a rs232 console for the user interface.\n\nI have not done much on the mechanics but have decided to orientate the wire horizontally and move the workpiece. Keeping the tank still, allows me to configure it as a drop tank easily. This gives fantastic access to for wire threading and removes the need for drainage taps/pumps etc.\n----\n\n''Stuff done:''\n //pathetic small steps but steps all the same//\n\n1. Purchased MCC reader and have stipped out slot and mounted to vero board.\n\n2. Purchased PIC, oscillator and 3.3v regulator\n\n//So now have required components for PC-end hardware and software developement, possibly for basic machine-end tests also with a change of software//\n\n3. I now have a 500VA torroid with two secondary windings parralleled it gives about 100v@5A after rectification.\n//This will form the basis of an RC supply.//\n\n4. Have located a decent box and most of the components for the supply. Have set up work area for pic programming and testing.\n\n5. Found some watch jewels in amongst a bag of watch gears I bought\n//I now feel that auto threading or even manual threading could be a real problem. I need to set up a water system to test.//\n\n6. I can now compile the C code and write it to the PIC. I now just need to write the correct code to the PIC!\n\n7. Have soldered up most of the test circuit, I await a small MMC to double check the pinout of the MMC socket.\n\n8. Have aquired some 0.058mm tungsten wire for testing.\n\n9. Things have gone slowly to say the least, I finished the prototype board with PIC and MMC slot and minimized the test program to put some data on the card, I can then read it with a card reader and winhex. At least that was the plan, I can write the C code, compile it and put it on a PIC but it is not working. Simple turn on pin 1 tests are also not working. I smell a rat, will buy a new PIC tommorow.\n\n10. It turns out I needed to turn brown out detection off as I am running the pic on 3.3v. Argh, Now I can at least run basic programs.\n\n11. More weirdness, bit 0 of port A goes high if I put my hand near or on Port B, turns out that low voltage programming was also enabled and that uses port B, argh!\n\n12. The pic is talking to the card and the card seems to say it has got the data A ok. But the data does not appear. Progress is hampered because I just can't work out the command format.\n\n13. Argh Argh!!!! Sandisk's new datasheet is different to one of there older versions, for some reason they ellected to miss out the command format part. Now I get it and can debug proper.\n\n14. Have drawn up my stages in ~Rhino3D and started sketching the machine (see mechanical section)\n\n15. Well it turns out it was working it just turns out that win hex shows the data at a large offset, I found it by searching the whole card.\n\n16. Various stupid errors later and I have a program that will write pulses from the parallel port to the MMC card. It just needs some cleaning up and I want to find out if the offset seen in winhex is real or something wierd.\n\n17. A long time since my last update although I have been busy. It turns out the offset seen in winhex is caused by winhex, there are issues concerning bootsectors and the like that make it display in a strange way. To avoid these problems I have gone to using RS232, the PIC program can be set to read the card and output what it reads to an ancient laptop running a program called hyperterm. The code to do this is in the MMC programs linked to in the MMC section.\n\n18. Spent about 2 hours trying to get the rs232 to work, went to bed, next day it worked, then it stopped. It turns out that if you dont click on the screen of hyperterm it does not display anything, argh!!!!!!!!!\n\n19. But now I know the basic recording part works just as planned, I can record a set of pulses to the MMC and then read them from the MMC to the rs232 and see them on the laptop confirming that is works.\n\n20. Have written the basic program to read the card and move forward or backward through the recorded pulses depending on the state of the gap control inputs. However there is a big snag, if I want to be able to move back and forth completely freely I need to be able to buffer at least two blocks of data at a time, this is 1024 bytes, my PIC only has about 300.\n\n21. Have selected a new pic, this time in the 18 series, the problem is that it requires a different C compiler, luckily microchip do a student version of their C18 compiler and I think it will do all I need. Still it means setting everything up and probably quite a bit of twiddlying in the code to get back where I was already.\n\n21. As a side line I have been considering user interfaces to the machine, this could just be a key pad, perhaps with an LCD screen or even just a couple of LEDS. The more work done on this side of things the more usable the machine will be. However it could get very complex. I think I have decided to use a second pic to keep track of the main pic, leaving it to get on with its business of reading the card. The second pic will then communicate via rs232 with a terminal. The terminal can be whatever you want, it is simple to write visual C++ or visual basic programs that will work with rs232 and it even allows the possibility of using palm PCs and modified games consoles. I am working on using a Nintendo DS which has two screens, one is a touch screen. There is an active developer network and a library that makes basic functionality very simple. Rs232 has already been implemented by some.\n\n\n''Tasks''\n\nSoftware:\n\n Sort out new compiler\n Get basic MCC card functions working with new PIC\n Get parallel port commands written to the MCC again (easy bit)\n Get basic program flow working, allowing pulses to be played back and forth.\n //at this point there is enough to make a machine that will cut something if a generator and some gap feedback is added//\n Work on a user interface of some form.\n\n\nGenerator:\n\n1. Get or build a 100v DC PSU. '\n2. Using a mosfet, mosfet driver and pulsegenerator or 555 circuit along with a rheostat for current limiting access the viability of machining and obtain the current requirements of the supply. This will require some suitable wire.\n3. Build the final version of the generator.\n4. Test in single axis mode to determine some basic "technology"\n\nMachine:\n\n1. Assemble stages and drives.\n2. Build a tank - //Need a way to very quickly fill/empty the tank//.\n3. Build arm.\n4. Build wire tensioning and speed control mechanism\n5. Build wire guiding mechanism\n\n\n''NEED:''\n\n==Small amount of suitable wire==. //I now have some 0.058mm tunsten to play with, might be OK but only got 10m//\n==100v (or so) power supply== //Well I have the bits\nto talk to someone about typical currents
*WebLinks\n*ScientificPapers\n*PatentsOfInterest
[1]\n[2]\n[3]\n[4]\n[5]\n[6]
A home CNC approach to micro wire-EDM
Micro wire-EDM project
http://www.tiddlywiki.com/
SoftwareOverview\nPcSoftware\nMicroProcessorSoftware
The flow of infomation will be as so:\n\n* This bit is the standard CNC stuff:\nCAD software - Produce a drawing\nCAM Software - Convert Drawing to gcode\nCNC control software - Convert gcode to step/direction pulses\n\n* This is my bit:\n\nCard writer: Microchip PIC which is a self contained microcontroller is used to write the step and direction pulses seen on the parallel port of the PC running the CNC software to a camera memory card known as an MMC (stands for ~MultiMedia Card).\n\nMachine: Once the program has been walked to the EDM on the MMC card, a second pic will echo the recorded pulses to the stepper drivers according to the state of the gap conditions. If the gap it too big it will play out the pulses to the drives making the wire move forward, if the gap it just right it will stay still and of too small it will play the pulses in reverse causing the machine to back up along its toolpath. Or something along those lines anyway.\n
I had thought that the spark generator would consist of a simple DC source with a basic linear current regulator in series with it. However there are a few problems with this idea. I had assumed that my max machining current would be about 2A, I did this to make things easy. Hoever in a short 100v@2amps means 200W of heat needs to be dissapated. 200W resisters are pretty huge so there was no way a poor little linear regulator chip was going to do the task even ''if'' I could get one rated for the voltage.\n\nThere seem to be three possible ways things can go:\n\n1. I realise I only need tiny currents for the tiny wire and hence go for a homebuilt linear regulator that will handle the voltage and the lower current.\n2. I build a machine with a set of power resisters that can be connected in parallel as required to set the current limit.\n3. I build a clever switching power supply that does not waste power but rather controls it.\n\nIdea 1 will be fine if it turns out to be the case that only tiny currents will be used, it should be possible to rig something with a large high voltage power transister and a feed back circuit.\nIdea 2 is how it always used to be done. It may be a bit expensive and will turn the PSU into a cooker.\nIdea 3 would be fantastic, a tiny box with some power mosfets in it and an inductor or two. But how easy it would be I am not suure. This option also allows the supply to be set up as a current source, just as the big boys do it (I think).\n\nThere is an excellent paper on a switching converter for EDM. It uses a ~LCsCp resonant converter. The circuit is this:\n\n1. Take mains and rectify it\n2. Chop this at a high frequency.\n3. Feed this into the ~LCsCp resonant circuit, the load of which was the Cp in parallel with it. The load is a small transformer.\n4. The recified output of the transformer feeds to the workpiece and electrode.\n\nI have not read the paper enough times to fully understand the curcuit but the main point is that the high frequency switching allows a tiny transformer and inductor but the resonant behaviour ensures good efficiency. And most importantly it behaves as a constant current supply.\n\n// Futher developments//\n\nTalking to some chaps at work it seems\n\n\n''Pulses''\n\nThe current limit partially governs the amount of energy delivered per pulse. The second factor is that of the pulse length t~~on~~ . The time between pulses t~~off~~ is normally reduced to a minimum but so that there is enough time for machined particles to be flushed before the next pulse.\n\nThe PulseGeneraTor is quite simple for such a small machine and consists of just two cascaded 555 timers and a mosfet, t~~on~~ and t~~off~~ will be adjusted by hand according to the material to be machined.
For the tank it just needs to be rust proof and electrically insulated from the work holding stuff.\n\nI want to be able to fill and empty the tank quickly. A tap and a second container will suffice however care should be taken to keep al of the mechanics dry.\n\nFor work holding it is usual to have an L or U shaped platform to which the workpiece can be bolted.
The idea of building a micro wire EDM from common parts to run from hobby CNC software may seem far fetched. However I am making several assumptions that are valid for my usage and make life easier.\n\n* The machine will use very thin wire.\n''Why'': This is because I want to make very small parts including injection mould inserts for gears.\n-> This means that machining currents must be low.\n-> and that means that the design of the power supply is more simple than larger machines and less expensive.\n\n* The thickness capacity will be small <3mm\n''Why'': because that is all I need, more may be possible.\n=> this means that flushing requirements are reduced, see point 3.\n- I now think this can probably be increased, it will just be slow.\n\n* The workpiece will be submerged\n''Why'': Nozzle based flushing is OK but it is quite a critical process and used as far as I can tell to make the building of some machines easier. Plus when you finish machining you don't have a tank full of water. However with my small tank drainage will be no problem and I think any mechanical problems can be overcome. Nozzle based flushing is rather a tricky process, pressures must be correct and tuned to the process. With such thin wires it is inevitable that vibration or bowing of the wire could occur and with ultra fine kerfs flushing will be very difficult full stop. I think it is a no-no for anything but cutting shim. \n- Furthermore submerged cutting is better for cutting of non planar materials like cylinders etc and I think it could make wire threading easier.\n\n* The tool paths will be short, or the parts will be small.\n''Why'': partly because that is the kind of part I want to make and partly to simplify the system. \n=> because of the small amount of material removed it should be possible to use standard automotive distilled water as the dielectric with perhaps some gentle filtering. After use the water will be discarded.\n=> Distilled water also means that there will be no horrible smells or messy waste.\n- actually the small size of the machine makes a lot of sense in terms of the dielectric. Of course in production cutting you need plenty of it as there will be a lot of contamination but the water ionizes itself if you leave it and that means de-ionization is required. Given the small cuts an small tank replacement will probably work fine and is a pretty good "filter" system.\n\n* The machine will be controlled by a microprocessor.\n''Why'': edm feed rate control is quite complicated, it is necessary to watch gap conditions and advance/retreat the wire along the tool path as necessary. No hobby CNC software have reverse facilities built in, at least not within a "move".\nHow: A standard hobby CNC program such as ~TurboCNC or Mach3 will be used to interpret G-code and produce step and direction pulse for the motors. These pulses will be intercepted by the uP and stored digitally in memory (a camera memory card as it happens). When machining is to proceed the uP will turn on the generator and begin to play the stored pulse trains out to the motor drives. While this is happening the gap conditions are monitored and the wire is slowed/halted/reversed as needed in order to maintain good machining. Another factor is that it will allow the machine to be free running without the need to use a PC, as the cutting speed may be quite slow this could be handy.\n\n* The tank will have a clear perspex front panel and submerged blue ~LEDs. \n''Why'': Well why not
At first I started worki using the PIC show in the example shown in the MultiMediaCard section. This was a ~PIC16F876 and the programming was sone using a free version of ~CC5X, a C complier for pics. The lack of ram has caused me to move to a ~PIC18F2620, this beast has the same pin out but much more ram, rom and even includes a pair of comparitors. The down side is that the ~CC5X compiler does not cover the 18 series of PIC so I need to use the student version of microchips C18 compiler.\n\n\n
This is a tiddlywiki that plots my progress in trying to build a homebrew micro wire-EDM machine.\n\nwhat is a tiddlywiki, well in this case it is a single java based html document that I can edit easily and can link information together easily. Tiddlywiki is a form of wiki that is designed for personal use and can be stored on a USB drive or uploaded to the web. If you want to edit this wiki you can but only after first down loading it [[right click "save target as"|edmwiki.html]].\n\nSome basics in use: \n\n- Use the menu on the left to get to the subsections.\n- Click close at the top right of a tiddler if you are done with it.\n- Click close all in the right hand box to start from scratch.\n- Click permaview to generate a url in your adress bar to provide a link to all the tiddlers you have open\n- Click timeline to see what I have edited recently to go straight to the changes etc.\n\nif you want to contact me then email me at eexgsidontlikespam@nottingham.ac.uk and just remove the I don't like spam bit. \n\n''WARNING'' this document should be thought of as a journal, it shows my current state of thought on the project. Some of what I write may well be incorrect and I will be altering it frequently.\n\nIn case you wonder what is so good about micro wire-EDM have a look at this:\n[img[tiny micro wire-EDM cut gear pump|cool.jpg]]\n
For research:\n\nhttp://ep.espacenet.com - patents\nhttp://scholar.google.com/ - Scientific literature\nhttp://print.google.com - Search inside books\n\n\nMachine manufacturers:\n\nhttp://www.agie.com\nhttp://www.fanuc.co.jp/en/product/robocut/\nhttp://www.makino.com/\nhttp://www.sodick.com/\nhttp://www.charmilles.com/charmillescom/en/menu_0/main_id.html\nhttp://www.mitsubishi-world.com/Products/Wire/home.html\nhttp://www.ona-electroerosion.com/eng/homepage.htm\nhttp://www.accutex.com.tw/e_au-300ia.htm\n\nFixtures:\n\nhttp://www.edm-products.com/4000/index.htm\n\n\nGood technical info:\n\nhttp://makino.com/die_mold/forging/technical_articles/articles.asp?id=122&title=EDM+Wire+Tutorial\nhttp://www.reliableedm.com/wire1.html
''Images''\n\nHere's how the code works:\n{{{\n[img[alternate text|image URL]]\n}}}\n\n''Formatting Options:''\n\n*You can create ''Bold'' text by enclosing it in pairs of single quotes:\n{{{\n''bold text''\n}}}\n\n*You can create ==Strikethrough== text by enclosing it in pairs of equal signs:\n{{{\n==strikethrough text==\n}}}\n\n*You can __Underline__ text by enclosing it in pairs of underscores:\n{{{\n__underlined text__\n}}}\n\n*You can create //Italic// text by enclosing it in pairs of forward slashes:\n{{{\n//italic text//\n}}}\n\n*You can create ^^superscript^^ text by enclosing it in pairs of carets:\n{{{\n^^superscript text^^\n}}}\n\n*You can create ~~subscript~~ text by enclosing it in pairs of tildes:\n{{{\n~~subscript text~~\n}}}\n\n*You can change the text's @@color(green):color@@ by enclosing it in pairs of at-signs (@@@@) and specifying a text color:\n{{{\n@@color(yourcolorhere):colored text@@\n}}}\n\n*You can change the text's @@bgcolor(red):background color@@ by enclosing it in pairs of at-signs (@@@@) and specifying a background text color:\n{{{\n@@bgcolor(yourcolorhere):your text here@@\n}}}
There are lots of ways to guide a wire:\n\n* Jewels: these are just hardened orifices that the wire is passed through. Very often the upper guide is a split type to allow for automatic wire treading. My instinct is that this will not be a good way for a hobby machine to work with fine wires as threading will be very difficult indeed.\n\n* Pulleys: You can just pull the wire between two pulleys, this is what they do for simple machines used for cutting up samples for experiments etc. Pulleys are very simple but do have disadvantages as it is necessary to have excellent bearings to avoid float. To me a pulley only design will not be viable for fine wires.\n\n* Grooves: Usually a v shape so that the wire can sit nicely. Precise and easy to thread, not so good for 4-axis work. \n\nI found an excellent paper where a table top micro wire EDM is discussed. This machine uses 0.02mm thick wire (less than half a human hair's thickness). They say that it is too expensive to drill holes of that size in a diamond so they opt instead for "precision v-grooves". From what I can tell these consist of cap head bolts with a groove turned on the outside face. With this machine they achieve incredible surface finishes and 1um accuracy. There machine is only good for VERY small workpieces and parts and probably does not lend itself to threading through start holes but it is certainly food for thought.
\nIn wire edm the wire is the electrode and it is well know that the electrode also suffers damage during discharge. For this reason it is normal to feed the wire continuously between two guides from the relatively large spool the wire is supplied on to a storage spool or box.\n\nThe wire guides are very important as they dictate the mechanical position of the wire, if they are imprecise then cutting will be also. Wire speed and tension is also important. Too slow a wire creates too much wear on the electrode, this can lead to brakage in roughing applications and poor accuracy in finishing procedures. The wire tension must also be maintained, a slack wire will cause non straight sides to cuts and flushing and dischargemay cause a slack wire to produce poor finish or even shorts.\n\n*WireGuiding\n\n*WireTensioning\n
\nProper tension and wire speed are important. Proper speed because if the wire travels too slowly it may be eroded too much causing breakage or lower accuracy. Proper tension is important as it effects the cut quality, perhaps more important than absolute tension is actually the maintenance of constant tension. It is easy to imagine a slack wire or a wire periodically tightening oscillating in the spark gap. This can lead to instability in cutting and even shorting. It certainly affects the finish.\n\nThere are several ways to run wire and tension it.\n\n''Feed:''\n\n*Spool to spool:\n-In spool to spool one spool pulls the wire from the other.\n-Sounds easy but can be a pain after several re-threads.\n-Speed control must be performed on an intermediate pulley as spool diameter affects wire speed.\n\n*Spool to feed rollers\n-rollers arranged like a mangle or printer paper feed pull the wire.\n-used wire dumped in a box (sometimes after chopping)\n-speed control can be done on the feed rollers as circumference is constant.\n-good for frequent re-threads and for recycling of used wire.\n\n''Tension''\n\n- Wire tends to be supplied on large spools, the inertia means it may not be possible to pull thin wires straight from the spool. As such the supply spool may be motorized.\n\n*Clutch systems:\n-A common way to regulate the tension is to use some form of arm with a transducer and a electrical clutch system to apply the required tension to an intermediate roller. This will form a closed loop control system.\n-Not overly easy to produce\n-Probably harder to get right at smaller scales.\n-Expensiveish\n\n*Gravity:\n-Several commercial fine wire EDMs seem to use gravity, I have used this myself in a coil winder and it is a neat solution.\n-Gravity is pretty constant.\n-Easy to control, hang a weight on the wire (via a pulley) and monitor the height of the weight. Adjust the feed rate on the wire supply spool such that the height is maintained. As such the wire supply spool matches the wire feed spool or roller system and maintains the tension supplied by the weight. I used bang bang control with reed switches, this is not ideal as acelleration of the mass will cause some tension variation however the wire feed rates are so slow I suspect it will not be too much of a problem. i.e the acceleration can be very small. Otherwise a simple opamp based control circuit might be built.\n\n*Other:\n-In one paper a micro spool2spool system is used, the supply spool is so small that the tensioning force was applied to the shaft of the spool using an openloop system comprising of a rare earth magnet and steel plate. Tension was calibrated with weights and the magnet-plate distance set appropriately.\n\n\n''Factors specific to micro EDM''\n\n*Wire vibration is very important, vibrations that you can hardly see may affect the cutting and such thin wire will vibrate rather easily. I have no way to prove it but perhaps submerged cutting may also be of benifit in vibration reduction? Some machines seem to put the tensioning systems right at the cutting point to help in this. I think Agie do this. The tabletop machine designed in the literature actually used little foam blocks to try and supress vibration. For very thin wire it may be a good idea to put the machine on a solid base with some vibration damping (such as some moped typres).\n\n*The wire speed at least in machining classed as micro machining is typically very low. This will probably make things easier if anything.\n\n*Spool size: thicker wires always some on big spools however due to low wire speed it may not be the case for fine wire stuff. Typical speeds for 0.02mm wire are less than 1mm per minute! Spool size will dictate the tensioning mechanism although I feel gravity may be my prefered choice.\n\nBerkenhoff in Germany make micro wire for EDM. They supply their small wires on "bedra 4" and "BK100" spools. these are the following sizes:\n\n''Bedra4 spool''\nCore diameter = 80mm\nCore length = 105mm\nOuter diameter = 130mm\nOuter length = 125mm\n\n''K100 spool''\n\nCore diameter = 63mm\nCore length = 80mm\nOuter diameter = 100mm\nOuter length = 100mm\n\nNeither are what I would call cotton reel sized.
With the decision on the machine configuration made wire threading is much easier than previously expected. It may well be possible to thread by hand through all but the smallest start holes. A vaccum pump and tube may help suck the wire through super small start holes is required. After that part of the process, the rest will be hand threaded.