Considerations after first PCB milling tests

[CarlosGS]

• The Y axis (the old one) is too wavy. I think both new motor&idle stands need to be made thicker (10mm)
• The Z axis needs to be made taller and more robust. Linear bearings are not held strong enough: "elastic" holder is not reliable at all.
• Also regarding the Z axis: doing the first mill, the heat of the spindle motor expanded the PLA and did literally slide and crush down into the PCB :( This was fixed tightening more the screws, but I think it will be better to add a tab to hold the motor from the bottom.
• I couldn't finish the tests because the Sanguinololu was freezing (not sure wether it was the atmel, ftdi or both!), most likely due to electrical noise from the switching power supply from the spindle. The Cyclone Host software has a "command resend" error recovery that didn't help at all: it was not just an error in communication, but a weird freeze of the Sanguinololu :(
• Be careful when milling/drilling with thin bits: the only one that has survived with no damage has been the 1mm one (5TFLU1-1 in the BOM), for thinner bits set speeds <100mm/min

Now some pictures:

• First results (drastically ended when the spindle crushed the PCB)
• Next try, after tightening the screws of the spindle holder: In the first pass, the 0.1mm tip didn't stand for long, and it became a 0.2mm tip. I should have replaced the bit and set a slower speed, but decided to make a second pass with the damaged bit instead (and same speed) which gave nice results (but too thin tracks-enough for the first attempts)
• After that, I did a second pass with a thicker -1mm- tool (5TFLU1-1, and with new gcode, of course) I was quite happy with the results of the second pass, the 1mm tool (5TFLU1-1) seems very robust and has behaved nicely. X-Y alignment could have been better, though.
• For the drills, the software was modified (original pygerber2gcode doesn't support normal drill bits, since it makes circular XY motions to achieve different diameters). The 0.4mm drill (5DRL0412-1) was broken when the software failed in one of the first attempts, so I re-tried with a 0.5mm tool (10TFLU05-1): The tool broke after 20-30 holes due to stress, Z speed was set to 40mm/min. Maybe the software should implement a "tool rest" functionality to allow drills/bits to cool down for a few seconds between moves. Also, a "tool break" detector would be superb (any idea on this?)
• For the edges, there were more problems with the electrical noise from the power supply and the job wasn't quite finished:

But, in the end, I think there are promising results: a beta console-based software has been tested (Sender.py and Cyclone Host modules) and it has helped to address many hardware issues.

To-do list:

• Try shorter and better shielded USB cable, get a proper 48V 300W power supply, see if electrical noise issues persist.
• Reinforce Y axis
• Rework Z axis
• Is 2:1 gear ratio adequate for X-Y? Maybe try 3:2? (increased resolution)
• Continue developing software (add capability for "live" user commands -i.e. Pause, Resume,ToolChange...-)

[yOPERO]

Carlos

I would love to have the spindler to perform tests along with you but I am afraid I will not have it in the short term. I think it is very important to have a very robust frame in general and I must say that using plastic on a CNC is a very bold approach and also very challenging that's why I decide to reproduce Cyclone and see how it performs :)

My perception is that we may get good stability by using the same approach that was used in older models of 3D printers which is basically using threaded rods and plastic(Huxley, Prusa2).

Usually CNCs have at least each axis as a whole unit connected together to give stability, the current design of Cyclone has the Y axis sort of disperse As you know I have done some work on the xz frame by adding an M8 threaded rod at the back of this frame and the stability has been improved, following the same approach here I present and idea for the the motor holder and idle holder to be a single unit(Y axis).

Let me know your thoughts

[CarlosGS]

Hi yOPERO, I think you are right, threaded rods can help: just let me try first with the 10mm thick supports + four screws (Take in account I still have the original version installed) The waving is not that bad, and using the least amount of vitamins is one of my goals. For the X axis that I have been testing there doesn't seem to be any waving at all.

Thanks for your suggestion and for showing your ideas graphically :)

[yOPERO]

Carlos

As you know SAV is not a good at all, last night before going to bed I switched on my printer and left printing both modules overnight :)

Now I have both modules and the results are great, stability has increased a lot and I am in the position of saying that the extra M8 threaded rod may not be needed.

In this modules I have increased the variable "wall_thickness"(in your code) to 12mm. Basically is the same module mirrored which also gives the option to place the Y motor either way front or back.

Github is not allowing me to attached pictures so I am loading them to Google+, check it out you are tagged on my post.

[CarlosGS]

I have just mounted the thicker supports (in my case 8mm) and agree completely, way more stable!! The original parts (5mm, 2 screws) were really weak. Also, I have mounted new gears (5:7 ratio, 15 and 21 tooth): I think they will give just the aditional bit of resolution that was needed, and also gives easier access to the motor screws.

I have ordered this power supply: http://www.ebay.com/itm/140986744539 But will buy another one on monday (SAV) in a physical store, to continue the tests :P

Prior to the frames, I am going to focus in the Z axis. Afterwards they will be reinforced with your hexagonal design :)

Thanks again yOPERO, and keep up the good work!!