Circuit Mill Development Log: Spindle Dropping, Auto Probing, and Z Axis Failure Mode

Shane Oberloier's picture

Spindle Dropping

In order to effectively debug and recognize odd behaviors with the mill, I have started streaming runs of the mill. The first run I recorded had some very valuable insight. Near the very end of the run (well not really the end - but I stopped it due to the error), the spindle dived into the material. Reviewing the video showed that the spindle had actually fell out of the 3D printed spindle holder.

So the first remedy was to tighten the grip on the spindle. The results were looking more promising, but once again - the spindle started drooping out of the holder.

The next solution was to wrap standard office rubber bands around the spindle. The thought process being that both the spindle and 3D printed parts are very slick, and the rubber should increase the coefficient of friction. Just by placing the spindle in holder and tightening it - I could tell the issue was fixed. I could tug on the end of the spindle and it wouldn't budge. And sure enough - I managed to make it through an entire run without  the spindle drooping out of its holder at all. Also something to point out in the following video - I show the entire leveling process, and Copper Carve is screen captured, so you can get a good idea of how the software works.

Spindle with rubber bands placed around it
6 rubber bands placed around the bottom of the spindle.

The results are still not optimal - the backlash compensation in OCI Copper Carve is not working properly. This is due to the change in how G-Code is streamed. I have to reprogram the method in order to work with the new streaming method, and then I suspect that the mill should be fully operational and dependable. Below is the circuit made in the above video. Note that some of the traces have been completely cut away, and this is a sure sign of backlash.

Board from 5 March. Note the broken traces due to backlash
Board from 5 March. Note the broken traces due to backlash

Auto Probing

It is essential for a mill of this construction to probe the surface of the copper board in order to compensate for warpage. I have properly implemented the method in Copper Carve, and the process can be seen below:

There are a few outstanding bugs with the automation, but when the process works correctly (it often does), the mill is able to properly modulate the Z height, compensating for the warpage.

Z Failure Mode

If this mill is to be commercialized and sold as kits - something that I think is important is paying attention to the spindle failure mode. Being a belt driven system - when the Z axis motors lose power, the spindle will plummet downward (due to it's 1kg weight being greater than the friction of the belts). While brainstorming about the problem I considered that a spring could be implemented in order to counter-act the force of the spindle's weight. However I didn't want to redesign any of the standard D3D parts, and also realized that the springs force would need to be equal on all 4 corners. Then I realized that the full travel of the mill is not utilized (only about a quarter is necessary, or 5cm). The space on the linear rods would provide a great spot for compression springs. So I went to my local hardware store and picked up the lightest springs that were approximately the correct height. And low and behold they work perfectly. The springs are compressed about 1 cm during the steady state if dropped from no compression. And if the springs are compressed further, the friction of the belt is able to keep the spindle in place (note all of these tests were done with the movement motors turned off).

Z axis springs used to cancel the force of the spindle's weight
Z axis springs used to cancel the force of the spindle's weight