Yesterday, I tried cutting out a square with the new machine setup.

I wrote some GCODE per hand and let the machine run.

G28 Z
G92 X0 Y0 Z0
M80

;####################
G0 X0 Y0 Z75 F1000
G7
G0 X0 Y0 Z90 F100
G8
G28 Z
;####################
G0 X4 Y0 Z75 F1000
G7
G0 X4 Y0 Z90 F100
G8
G28 Z
;####################
G0 X8 Y0 Z75 F1000
G7
G0 X8 Y0 Z90 F100
G8
G28 Z
;####################
G0 X12 Y0 Z75 F1000
G7
G0 X12 Y0 Z90 F100
G8
G28 Z
;####################
G0 X16 Y0 Z75 F1000
G7
G0 X16 Y0 Z90 F100
G8
G28 Z
;####################
G0 X16 Y4 Z75 F1000
G7
G0 X16 Y4 Z90 F100
G8
G28 Z
;####################
G0 X16 Y8 Z75 F1000
G7
G0 X16 Y8 Z90 F100
G8
G28 Z
;####################
G0 X16 Y12 Z75 F1000
G7
G0 X16 Y12 Z90 F100
G8
G28 Z
;####################
G0 X16 Y16 Z75 F1000
G7
G0 X16 Y16 Z90 F100
G8
G28 Z
;####################
G0 X12 Y16 Z75 F1000
G7
G0 X12 Y16 Z90 F100
G8
G28 Z
;####################
G0 X8 Y16 Z75 F1000
G7
G0 X8 Y16 Z90 F100
G8
G28 Z
;####################
G0 X4 Y16 Z75 F1000
G7
G0 X4 Y16 Z90 F100
G8
G28 Z
;####################
G0 X0 Y16 Z75 F1000
G7
G0 X0 Y16 Z90 F100
G8
G28 Z
;####################
G0 X0 Y12 Z75 F1000
G7
G0 X0 Y12 Z90 F100
G8
G28 Z
;####################
G0 X0 Y8 Z75 F1000
G7
G0 X0 Y8 Z90 F100
G8
G28 Z
;####################
G0 X0 Y4 Z75 F1000
G7
G0 X0 Y4 Z90 F100
G8
G28 Z
M81
G0 X0 Y0 Z0

The behavior of the new machine is very different from the EDM machines that I build before.

- First of all, there is (almost) no condition that can lead to a failed machining job. In theory, the electrode could get stuck in the workpiece, but that has not happened, yet and I think that it is really unlikely because the gap between electrode and workpiece is large enough to prevent this. Otherwise, there is not much that can go wrong.

- During machining the electrode got more pointy, like if you would sharpen a pencil:

That led to a smaller exit hole at the backside of the workpiece and so a tapered hole.

It also led to strange behavior in which the electrode moved back and forth in an endless loop. I think this happened because the shape of the electrode prevented the machine from starting an arc.

I hope I can avoid this in the future by writing GCODE which leads to more constant electrode wear.

- I had to machine the holes where the strange behavior happened again per hand, but I'm hoping that I can fix that at some point.

- Besides that, the machining job was finished with no errors.

Here you can see the finished workpiece:

- With the new machine I was able to run a program for the first time without having to keep an eye on it all the time.

I started the machining job and watched TV, fixed the endless loop error, and just waited for the job to finish.

Besides the error (that I hope to find a permanent solution for in the future)  it's the way of stable operation that I was looking for.

In terms of the shape of the workpiece's edge - I'm thinking about writing a tool path generation tool to generate code for overlapping holes to reduce the waviness of the edge and also keep the electrode wear low and evenly.

Another thing to notice is that because of the high amount of machined material by using sinker EDM, the deionization filter can no longer keep the conductivity of the water low. Therefore, I can no longer machine the parts submerged in water because the current would just be conducted by the water and no arc would ignite. So, at the moment I'm machining the workpieces in the air and using the water just for cooling and flushing. This way the machining does work even without deionized water, which is likely not the best thing, but as long as it works it's OK. I will keep an eye on the performance and if it will cause problems in the future I will try to find a solution for it. If it turns out that it doesn't lead to problems, it would be a good thing, because that would reduce the cost of the machine by making a deionization filter, separate pump, and ppm sensor unnecessary. Time will tell.....

Besides that, the test was a success and it seems like with some smart tool path generation the machine should be able to produce usable parts from metal.

I think the next step is writing some more GCODE and doing test cuts to get more footage of the machine and workpieces.