Calling all PCB CNC Milling Experts

[Qiaozhi]

Your math is impressive.

I changed my tool diameter to 6 mils. I had it set too small. See attached photo of 44 tqfp test.

[ATTACH]31775[/ATTACH]

Your CNC machine obviously has a lot better accuracy than the CNC3020T. I very much doubt mine will mill out something quite that small, at least not with a V-shaped bit. Perhaps I should get a 10 mil 2-flute square end-mill and give it a try. The maximum spindle speed is 8,000 rpm, but I read somewhere that 25,000 rpm is recommended for milling smd boards.

[Altra]

I think with a little trial and error your machine should be able to do the same. My lead screws are 5 turns
per inch (0.2in pitch) or about 5mm/turn

[Qiaozhi]

Here's my attempt at a TSSOP to DIP24 adapter.

and here's an image of the gerber layout.


The size, when the cutouts are removed, is 20mm x 32mm. Now I've just got to figure out how I'm going to solder the 24-pin SMD onto those tiny pins. As you can see from the traces, this is getting close to the limits of the machine. The SMD I need this for requires a metal pad under the chip. If you look carefully it's connected to one of the pins. You can buy these adapters on the internet, but the only reasonably priced ones come from Hong Kong, and it's pretty vague about the shipping charges. Although it says "FREE", it looks like there could be a customs charge and/or import duty. Also, it doesn't have the extra pad under the chip. So it's pretty much Hobson's Choice ... you have to make your own.

The SMD pads are 0.45mm x 0.65mm, with a 0.65mm pitch. This leaves only a 0.2mm clearance. I managed to mill out the PCB in the image with a 30 degree V-shaped engraving bit. Unfortunately I don't know the size of the flat tip on the bit, but I highly suspect it's 0.2mm, the same as the required clearance. This seems to be the case, as the pads and clearance appear to be the same width on the finished product. Consequently, I've just ordered 10x carbide PCB engraving CNC bits with a 10 degree angle and a 0.1mm tip.

In my original equation I didn't allow for the flatness of the tip, so I need to revisit that to see how much error it introduces in Z-depth calculation. Anyway - as you can see - the results are not bad for a tabletop CNC machine, but I think it can do better. Just need to wait for the new engraving bits to arrive.

I was also able to correctly drill the holes for the DIP connector using the Excellon Drill File, and mill out the board outline.

[Altra]

That came out nice. Surface mount soldering becomes difficult, without a solder mask.
I'm going to experiment with some different lacquers that I can brush on.

[greylourie]

Beautiful work. Will be interesting to see what the new smaller tipped bits can do when they arrive.

[Qiaozhi]

That came out nice. Surface mount soldering becomes difficult, without a solder mask.
I'm going to experiment with some different lacquers that I can brush on.

I've also ordered a 0.5mm tip for my Antex 18W soldering iron, so I'll give that a go. Hopefully the [milled-out] valley between the SMD pads will make the job easier.

[Qiaozhi]

I've revisited my original calculations for the Z-depth setting to account for the cutting edge diameter at the tip.

Here's the new equation:



where:
z = depth of cut
p = required isolation path width
t = cutting edge diameter (at tip)
= angle of V-shaped engraving bit

Examples
Let's say we want an isolation path width (p) of 0.381mm (15mil), and we plan to use a 30 degree () V-shaped engraving bit with a 0.2mm (t) cutting edge diameter. The required depth of cut (z) is then given by:



Likewise, for p = 0.254mm (10mil):



In this second example, it is clearly very important to get an accurate Z touch-off when using a 30 degree bit with a 0.2mm tip, otherwise the isolation path width could vary widely between milled boards. Evidently the 0.2mm path width is at the limit for this particular choice of engraving bit.

However, if we use a 10 degree bit with a 0.1mm tip, the depth of cut (z) is (repeating example 1):



Which is a major problem, as a typical copper clad board has a thickness of 1.5mm (60mil), and the 10 degree bit would cut completely through the board.

For a 0.254mm (10mil) isolation path width:



The conclusion is that a 10 degree bit (with a 0.1mm tip) will be good for isolation path widths at 0.254mm (10mil) and below, but above this width the depth of cut may be greater than the board thickness.

For the TSSOP to DIP24 adapter, it is not necessary to make the isolation path width the same as the clearance between the TSSOP pads (which was 0.2mm in my practical test). However, I was restricted by the tip diameter of the 30 degree engraving bit to an isolation path greater than 0.2mm. This restriction will be resolved with the 10 degree bits (when they arrive), and the Z touch-off should also be less critical.

It is important to understand the relationship between the angle of the engraving tool and the isolation path widths you intend to mill, and amazing how much the tip diameter affects the Z-depth calculation.

Can't wait to try the 10 degree bits, and compare the results from the 30 degree bit.

[Qiaozhi]

The 10-degree engraving bits eventually arrived (from China). I recalculated the Z-depth for a 5mil (0.127mm) isolation path using the updated equation, and the resultant TSSOP-to-DIP adapter board is attached below. Comparing this with the previous attempt (using a 30-degree engraving bit) you can see that the results are more precise. However, I did have to mill out the isolation paths 4 times, gradually increasing the Z-depth, in order to get the final result. I suspect the copper thickness is greater than the normal 1.37mil (35um). Also, even though the bed had been flattened, it appears that the board must be slightly uneven/warped.

[greylourie]

Hello Qiaozhi

Just found something that might be of use to you.

http://www.autoleveller.co.uk/

http://www.autoleveller.co.uk/download/

[Koala]

I only did a couple on my home made mill. I used a sacrificial piece of MDF milled flat then stuck the pcb down double sided tape. Think if I was going to do again I would make a vacuum table. Should hold much better and pull the board flat. Great for SMT boards

[Qiaozhi]

Hello Qiaozhi

Just found something that might be of use to you.

http://www.autoleveller.co.uk/

http://www.autoleveller.co.uk/download/

Thanks. I'll take a look at that.
For board flattening I've been using FlatCAM, which uses the G-Code file as a template. I think the problems I experienced with the TSSOP-to-DIP adapter were caused by the copper covered PCB being slightly warped, and the thickness of copper being greater than expected.

[Hyena]

Hi, I am new here and I want to open some locked features of this forum, so this is ideal thread for me I thing

I have 1200x1200mm engraving mill. Its very old but stable machine. Now its 2,5D, but next year I will change motherboard and it will become full 3D with Mach3.

I use AutoCAD for designing PCB, but its mess. I need to draw every circle around holes. I need to draw every path with two lines on PCD to mill it.

Do you know any software which can export from - for example - .lay to DXF with boundary of every path, which I can engrave then?

I am not sure I can explain exactly what I want with my English skills

[greylourie]

Maybe Flatcam ?

http://flatcam.org/

[Qiaozhi]

Do you know any software which can export from - for example - .lay to DXF with boundary of every path, which I can engrave then?

You will need to generate a set of Gerber files from your PCB package. The best software I've found for converting Gerber files to G-code is FlatCAM (version 7). The G-code will contain the information that your CNC machine needs to mill out the isolation paths. There is also LazyCAM, but it makes a lot of mistakes, and often gets the pad sizes wrong.

[Altra]

You will need to generate a set of Gerber files from your PCB package. The best software I've found for converting Gerber files to G-code is FlatCAM (version 7). The G-code will contain the information that your CNC machine needs to mill out the isolation paths. There is also LazyCAM, but it makes a lot of mistakes, and often gets the pad sizes wrong.

Have you tried the clearing feature, to remove excess copper?