[psychoul]

Hey guys. I am trying to create my first pcb here.

Currently I don't have a camera with me, but I will attach a picture here asap.

The problem: When the CNC begins cutting, the copper lines are clean and nice as they should be. However, as the job progresses, the quality of cutting degrades. The lines go thinner and not so CLEAN cut. It's like the copper is not cut but torn out of its place!

At the beginning I thought it was backlash. I don't have any backlash measuring equipment so I made an experiment of my own. I mark the initial position, then shift a bit to the X axis, and then go up and down the Y axis many many times and finally mark the final position. I measured no backlash at all. The same goes for the X axis.

What do you think is the problem?

Thanks!

[doorknob]

Is it possible that your spindle and tooling is not "trammed" to your table surface? Maybe the problem areas have the cutter tip going too deep into the circuit board substrate.

Does the degradation of the milling show up in a specific area (or areas) of the board? Is it possible that, rather than being related to the start and end of the job, instead the problem is related to the location on the board that is being milled? Is it possible that your board material is warped in some areas and is not sitting flat on the table?

[psychoul]

Hi and thank you for the reply. What you are saying is spot on!

I noticed that I also had this problem (where the table is not flat) but I didn't include this info here because (as you already noted in your post) the problematic areas are nearby. I still don't have my camera here but will soon, which will clear this up. So I think this is a different separate problem which I am looking around to see how to solve.

Therefore unfortunately I still don't understand why this happens.

[Jay C]

Tell us about the bit you are using, the feed, and speed of the spindle (and what spindle you have). Based on what you describe, I'm guessing you are cutting too deep into the FR4 which is extremely hard on bits especially if they are HSS.

[psychoul]

Hi and thanks for the reply.

I am attaching the photo in this post. The point indicated by 1 is when the CNC started cutting while the 2 is towards the end.

I am using a 60 degree bit from drewtronics.net . Feedrate is 5inches. Spindle speed is Kress set at the option 3 (which I guess it means 3000 RPM).

I understand the fact that I am cutting too deep. I am doing this because when I tried a less depth, at some points the bit would barely touch the surface of the board.

[Jay C]

Hi and thanks for the reply.

I am attaching the photo in this post. The point indicated by 1 is when the CNC started cutting while the 2 is towards the end.

I am using a 60 degree bit from drewtronics.net . Feedrate is 5inches. Spindle speed is Kress set at the option 3 (which I guess it means 3000 RPM).

I understand the fact that I am cutting too deep. I am doing this because when I tried a less depth, at some points the bit would barely touch the surface of the board.

60 degree is more delicate than a 90 degree. You should be using higher speeds 20K RPM or more. Feed, if your machine can handle it, can be 12-30ipm.

You need to level your spoil board before you mount your circuit. When you do mount your PCB, you might try carpet tape so it's fully supported, and pulled flat if it's warped. Thinner material like I use (1/32" thick) is more compliant.

Can you post a photo of the board mounted in your machine?

-Jay

[psychoul]

You need to level your spoil board before you mount your circuit. When you do mount your PCB, you might try carpet tape so it's fully supported, and pulled flat if it's warped. Thinner material like I use (1/32" thick) is more compliant.

Thanks for the information. What is the procedure you follow to level the board? I now use a beeper to let me know when the tip touches the board. I can see that on the four corners of my tiny board, the z difference could be as much as 0.006 inches. I believe this is a lot isn't?

Can you post a photo of the board mounted in your machine?

Sure, attached

[Jay C]

I mount a second spoilboard to my main table surface. In my case I just took a chunk of aluminum and added a piece of polystyrene on top (https://picasaweb.google.com/lh/phot...eat=directlink). Then using a square endmill I run a program to shave just the top .01" of the surface. The toolpath is either a spiral or raster your choice.

In that photo linked above, you can see that I have learned to use strips of carpet tape. This makes it easier to remove the board and less liley to drill or mill into the tape and gum up the bit.

-Jay

[psychoul]

Then using a square endmill I run a program to shave just the top .01" of the surface.

Wow! This is very clever. I haven't thought of that! So the table will be leveled according to my machine's x-y plane, no matter how many unknown parameters make it uneven...

Will try and report back Thank you very much...

[txcncman]

Surface your spoil board with a face mill, end mill, or fly cutter. Yes, 0.006 is a lot. Copper is one of the toughest metals to machine. It dulls cutters fairly quickly. Slower is usually better. The double face tape idea usually works well. A vacuum table works even better but can be fairly expensive, especially for a good vacuum pump.

[Jay C]

Wow! This is very clever. I haven't thought of that! So the table will be leveled according to my machine's x-y plane, no matter how many unknown parameters make it uneven...

Will try and report back Thank you very much...

Also, try .003" for the depth of cut. If it's too little, add .001" and try again until you get through the copper but not into the substrate.

[psychoul]

Surface your spoil board with a face mill, end mill, or fly cutter. Yes, 0.006 is a lot.

Hi again,
Today I got some break from my usual stuff and I worked a bit on the CNC. I leveled a piece of wood (shown in the attached picture) using an end-mill. I did it with 8in/min feed rate and a 50% overlap at each pass. It took about 1:20 hours.

By touching the surface you can really tell that it was uneven before!

Then I attached my PCB using a double face tape (i guess this is the carpet tape?). I took some measurements and it seems that in an area of 3x3in the touching height fluctuates +/- 0.0015. That makes it 0.003 difference between 2 points. I believe this is a considerable difference isn't?

How much fluctuation should I aim for? Do you have any other suggestions on this problem?

Thanks again!

[psychoul]

By the way, I created a small software to create the g-code for smoothing the surface. It runs in your browser so you don't have download anything. I hope it's useful to you.

[Jay C]

I typically mill at .003" for depth of cut to go through .5oz copper and .005" for 1oz. Any deeper and you are into the substrate. How did you board turn out after you leveled the area?

-Jay

[psychoul]

Hey Jay,

Well the board turned out better than before. On monday I will do the soldering and see if everything works!

However, I still have concerns about the quality of the board. First, I used a 60deg tool with milling at .006". However, using a multimeter, I could see that some lines, even though seemed OK, failed the continuity test. I didn't want to dig any deeper with this tool, since that meant it would mill even bigger tracks. So I changed the bit to 30deg and run it again (on the same board) having it to dig deeper (I think about .008"). Now the continuity test seems ok but I will be sure once the board is working.

I am attaching some pictures. Also notice the bad quality of the cut. The copper seems like its torn out of its place instead of a clean cut. Maybe I have large feed/Spindle_RPM ratio?

Thanks

[eaglezsoar]

I would like to ask what machine you are using? From the picture that shows the spindle
the CNC machine looks like a beauty. I think that when they talk about carpet tape they are not talking about the thick double sided tape but rather the wide, thin stuff that is protected on both sides with paper you peel off. You should be able to find carpet tape
at any home improvement store. It is not much thicker than a sheet of paper.

[psychoul]

Thanks,
We got the machine from here Welcome to World of CNC | Marchant Dice Ltd

[Jay C]

Actually I use the stuff from Lowes and it is "thick" compared to turner's tape which is only as thick as a piece of paper.

I think that when they talk about carpet tape they are not talking about the thick double sided tape but rather the wide, thin stuff that is protected on both sides with paper you peel off. You should be able to find carpet tape
at any home improvement store. It is not much thicker than a sheet of paper.

[Jay C]

I am attaching some pictures. Also notice the bad quality of the cut. The copper seems like its torn out of its place instead of a clean cut. Maybe I have large feed/Spindle_RPM ratio?

Thanks

I have my isolation set so that there are at least two passes per line. Experience showed me that it's easier and more reliable.

On 1/16" board 1oz copper my settings are:
Milling depth: -0.062"
Spot Drill: -0.005"

Isolation:
Default: 0.001"
Max: 0.016"
Step 0.002"

Tool size: 0.003" (this is the precise bit etching tool)

And Z depth: -0.008"

X/Y feed 35ipm
Z feed 5ipm

FWIW,
Jay

[Drools]

Just a quick comment on your Z axis. Is there any other bracing on the bottom of the axis? From the pics shown it is hard to tell but it looks like you are using the linear rails as the only support for the for the bottom piece of extrusion.
A over picture of the machine might help.