PCB Air Cut Bands.

[DodgeA]

Hello, thank you all for viewing my question and any help anyone can provide in advance.

I am cutting a double sided PCB with 1.4mil copper layers and the circuit is around 20mm x 42mm total and I am using AutoLeveller probing at 4mm spacing, however I am seeing two-three small narrow bands of copper that does not cut deep enough. My circuit uses 0603 imperial SMD. I have attached a picture and the probe file, keep in mind the picture is before I clean up the board. I am using 0.1mm V-Bit 10deg, I have 4 alignment pins and a make shift vac table to hold the board in place.

The problem is when I cut at 0.04mm I get rid of almost all the copper perfectly, but two-three narrow bands that horizontally go across the PCB uniformly. I have had a few that cut well, but most of the boards come out with these bands in the same places, and doing both sides adds to this complexity. The bands look to narrow to be a bow in the board, and they seem to fall between the rows that are being probed. One is around 10mm up Y and the other is 14mm. Unfortunately I have gone deeper down to 0.075mm however in going deep enough to get past these narrow bands I start eating away at the very small traces I need to go between the 0603 components, since it requires two passes between these components to isolate and leave a trace in between the pads.

I have tried everything I can think of, in adding a filter cap to the probe line. I do need to reskim my table as it is slightly higher on one side you will see in the probe file, however this is 90deg opposite of the narrow bands. The biggest thing is under the Microscope the circuit looks perfect everywhere but where these bands are. They look to be to narrow to be a bow in the board, and I would think that when these PCBs are made copper cladding would put a very uniform layer of copper across the board. I think it may be where there is a depression on the board from rollers on a conveyor belt as they are being made, but I would think the copper would still be perfectly uniform depth. They are cheap from Amazon made by uxcel. However I am frustrated running the mill over and over the circuit moving down till I can get rid of these ridges without removing too much from the traces. I have thought about writing a software to lower all points within these ridge areas to compensate. I have also wondered about putting the boards under heat ahead of time to flatten or even them out, however I think the bands are too narrow to be solved by this.

Thanks again for any help, and I would be happy to answer any questions.
Thanks, Andrew.

[Country_Bubba]

Andrew,
Reading your post and the additional information that James forwarded to me, I feel like the problem may be with the rigidity of your setup.

First off, you mention your using a "makeshift" vacuum chuck and getting minimal vacuum. How minimal are you getting. On my setup, I achieve about 25 to 26" Hg (635 torr). Again referencing my system which has a milled flat section for the pcb, I also use some tape around the edges to make sure they are sealed. This guarantees that it will be pulled down flat. I did this because I also use cheap board from ebay and found it to be quite warped and that was the only way to pull it down over the entire board. This was discovered early on when I found that probing pressure on the board was nill (done by testing with a piece of wrinkled aluminum foil and it never moved), but on close observation, I found the board would deflect ever so slightly when I was trying to mill the traces. Just for reference, I have been successful at .002" (.05mm)depth of cut. When I was running around 10" (254torr) I could see deflection on the board when it was cutting.

The other thing I can't stress enough is the "calibration" of you bit and this must be done for each bit that you use as I have found them to very all over the place even when I buy from someone whom I feel is making a quality product. When you dealing with the small tolerances that we are looking at, every little bit counts.

Another area that you have not mentioned is the machine that your using to produce your boards. Again, rigidity is key to the success of your boards and beyond that, the actual quality of the lead screws/nuts/couplings etc that is driving everything. You might try placing your raw board in a different location on the table to see if the exact same patter repeats in the same place. If it shows up in a different section of your board, it may be related to the physical properties of the machine itself.

As for the software your using, I can't comment as I don't use it. I use Eagle and Pcb-gcode to produce my gcode files.

In any event, hope this gives some food for thought and let us know what you find. Who knows may trigger another brain fart of an idea.

[DodgeA]

Hi James,

I have a HY-3040 5 Axis mill. It is using BallScrews, and I have calibrated all of the axis's and they seem to be spot on, down to .01mm. I am using a er20 Collet not a air chuck, just a vac table. I also used a dial indicator to check runout and calibrate the spindle. The interesting part is I can remill the same parts of the board over and over with no problem, so if I was losing steps or changing position I would see it. It is also in the same exact place across all of these PCB's even though I have moved the PCB's on my table. I have also had a couple PCB's that have had no issues in the same spot on the table where I have had issues.

The problem I have with it being the mill or even the probe, is I can see this without using the probe software when traveling across the PCB, without changing the Z axis, where I will be cutting through the copper and then have these 1.5mm bands on the PCB where it doesn't completely go through the copper. If it weren't for the fact that these bands are very narrow I would think it could be many other things like a bow in the board. Not sure if because of the tolerances it is crush on the boards when they packaged them, but they were falling in between probing rows. I do a lot of plastic and metal machining with this and have really good tolerances across everything I do, except for this one issue.

My vac table is a make shift table I made from MDF and I am using a 6hp shop vac, however I have a full HVAC vac that I need to pick up a filter for then I will be moving over to it. This is a stop gap measure. Maybe not such a time savor though. LOL. However I used to bolt down the boards but went this route trying to solve this exact problem thinking it was bowing in the board and I can't really bolt down in the middle of the board, and I am using a part of the board and double sided which complicates bolting.

I use to use a pyramid bit, and switched to the 10deg only to solve this problem, as that way I could dig deeper, but I have checked runout with the bits, but I didn't check the size of the path at the depth to put that into flatcam for my diameter for the 10deg bit I just switched too. Regardless with the 10deg bit it should be a narrow path not a wider one which shouldn't cause this issue.

So since I posted I resurfaced my vac table top to make sure it was perfectly flat, before it was off by half a mm from when I anchored it on one side, but it is very rigid, and I could see this slight elevation tilt in the probe file. I still don't believe this would have any effect on it whatsoever as it is perpendicular to the problem. I also moved my Probing down to 2mm from 4mm and ran it twice, takes longer, but its better then remilling the board, or having to deepen the cut and remilling the board. The results with this one look really good, all though I have had a few before that were good too. The other possibility is all these boards I have been using all came from the same batch, so it may not happen with the next batch I use. I prefer to know the exact cause which is driving me nuts though.

Thanks for your help, let me know your thoughts.
Andrew

[TassyJim]

Can you try mounting the PCB at 90 degrees and doing a test cut.
If the problem is a fault in the PCB stock, you should see the shallow bands also rotated.
If the bands stay in the same position related to you mill, you can't blame the PCBs.

So far I have only gone as shallow as ~0.003 thou and use rustic clamps so I have a long way to go...

Jim

[DodgeA]

Hi Jim,

Really good idea turning the board 90 degrees. I can't believe I didn't think about that, all though moving the board up and down should accomplish the same thing and the spots stayed with the board and in the same place. However I will definitely try that, I don't like to blame things like the board without being 100%, but I also wasn't sure how to fix it if it was.

I have run two boards now though using 2mm spacing on the probing and had great success, no issues and clean boards, on a cut of 0.04mm deep on both sides. So I am thinking it has to be slight dips in the board from maybe a cooling rack during manufacturing. Like I said these are really cheap boards and I am still on my first package of them. However I am not sure if the problem exists on both sides of the board, which you would think would be the opposite and be raised on the other side, but once I run into it I cut down deeper or give up and adjust things on my mill and start a new board. I have a tendency of changing to many things at once. I have run into it on the other side after the first side has gone well in the past. Like I said though 2 boards down with 2mm spacing with good results. When I moved the board down and cut, I was just going straight across the board with no probing, as a test to see if the bands were still there and they were, and not cutting a circuit, but my probing appears to have been on either side of the dip from the placement of the ridge and the probe files in the past.

I will give the board a spin this weekend and give it a try. I will also upload a picture of the good one, or maybe under the Microscope as well for fun. It is so cool to be able to machine stuff like this.

I was kinda dumb in the beginning as I bought parts for the 0603 smd stuff and had designed the circuit to be small as it is for a mobile purpose, but didn't realize the trickiness of cutting a trace in between the spacing of such parts. However I have learned a lot, and I am sure I am not done learning yet. Attendee of the school of hard knocks, I just hope to graduate soon.

Thanks for all your input. I will update more soon.
Thanks, Andrew

[Country_Bubba]

Andrew,
To followup on you last post, shortly after making my spoil board flat by milling, I put a fresh piece of board on and probed the whole thing. As I remember, I used a rather short probe spacing (as I remember about .1" [2.54mm]) and had a ton of data, but it showed the board to have what I considered a significant waviness to it. Don't remember the amount right now, but think it was on the order of .001" (.02mm) or so. Been quite some time since I did it.

Point is that you may have been onto something in your original thoughts of using a smaller probe spacing. Might be interesting to probe a section of the board especially in the area where your having a problem with both a standard probe spacing and then again with a "very close" spacing to see if your missing a valley or something????

Just a thought.