Most of the higher end CAD-CAM programs have a post processer where you can state what version/Type of machine you have, as many systems differ as to the options included and how certain features are handled when converted to machine code. (G-code).
For example, most follow the original standard used in all Fanuc CNC machines, so the post processor creates the G-code accordingly.

Hello there,

I may just end up writing my own program for the CAD part, because it will be mostly 2d, maybe a little 3d. I started a 3d cad program long ago but, hilariously, I never needed it for anything yet (ha ha). It was quite a job figuring out the math for the panel shading. As the object is rotated for viewing at different angles, the shading has to change.
But this will be mostly 2d which would be cake for me once I figure out how the G codes work.
I seriously doubt I would buy a high-end Cad program unless it was not super expensive.

The G code so far reminds me of when I first learned assembler for the Z80 microprocessor. In fact, even before that when I had to program memory chips using Octal binary. That was nuts. That was very long ago the place I worked at the time had limited hardware for programming the ROM's needed for the control part of the machines.

 

IMO, CNC G-code programing is far easier than manual machining, where for example, you have to be aware of tool offsets all the time, figure out radius programing etc, etc.
Start slowly with simple shapes and slowly advance to the more complicated shapes.

 

Attached is a rather poor photo of the board I made this morning on a 3020 CNC. 0.3mm track, 0.3mm Spacing.

 

IMO, CNC G-code programing is far easier than manual machining, where for example, you have to be aware of tool offsets all the time, figure out radius programing etc, etc.
Start slowly with simple shapes and slowly advance to the more complicated shapes.

Hi,

Sounds like a good idea. In fact, I probably could not do a PC board for example using manual movement of the head. Would be harder than drawing the pattern with a Sharpie for later etching.

I am wondering too now how the machine increments from say 0.1 inches to say 0.9 inches, in 0.1 inch steps, such as for making a DIP pattern. I imagine it gets to 0.2 inches ok, then 0.3, then 0.4, but as it gets closer and closer to 0.9 I wonder if it starts to get less and less accurate. Then there are the 0.05 inch pitch SMD packages. Each pin pad has to be fairly accurate. I wonder if I should invest a little more in getting a better machine.

 

Attached is a rather poor photo of the board I made this morning on a 3020 CNC. 0.3mm track, 0.3mm Spacing.

Hi,

Oh that looks pretty good. Are they 0.1 inch spaced holes for a DIP package (10 pin) or just 0.1 inch spaced pins for connectors?

What kind of bit did you use?

 

Hello again,

Another question came up.

Is it true that you can convert KiCad data into Gerber files and then convert the Gerber files into G Code?

 

Hello again,

Another question came up.

Is it true that you can convert KiCad data into Gerber files and then convert the Gerber files into G Code?

The first part is true ... not sure about the second part, though

 

Yes. You can convert Gerber files to Gcode.

 

Hi,

Oh that looks pretty good. Are they 0.1 inch spaced holes for a DIP package (10 pin) or just 0.1 inch spaced pins for connectors?

What kind of bit did you use?

Yes they are all 0.1" spacing, three 1206 SMD, and two solderable option links.
This was made using eagle 7.6.0, and PCBgcode to generate the gcode directly from the eagle generated files.
I use SRBF boards as it is much kinder on the tools. For the track isolation I use a 30°, 0.1mm engraving bit. I started using a 0.3mm end mill but they are very fragile.
I can also make double sided boards but of course they don't have plated through holes.

 

Yes they are all 0.1" spacing, three 1206 SMD, and two solderable option links.
This was made using eagle 7.6.0, and PCBgcode to generate the gcode directly from the eagle generated files.
I use SRBF boards as it is much kinder on the tools. For the track isolation I use a 30°, 0.1mm engraving bit. I started using a 0.3mm end mill but they are very fragile.
I can also make double sided boards but of course they don't have plated through holes.

Hi,

The idea of using SRBF boards is interesting.

Does that 30 degree, 0.1mm engraving bit end shape look like a very sharp, flat, pointed triangle?

Also, did you have any trouble getting the PC board to lay really flat so that you can get a reasonably deep but not too deep engraving cut depth?
Or, did you use some sort of sensor that adjusts the cut depth over the entire board plane? And, if so, what happens if you do not use that sensor?

Thanks much for the info.

 

I don't think it is possible to successfully "etch" PCB on a CNC machine without doing Z-mapping, except if the board is so tiny (less than 1" x 1"). The copper laminate variation in thickness and board warp is going to result in uneven mill depths. I have done PCBs successfully after doing the Z-mapping. Z-mapping is done automatically by the CNC Candle software. It takes no effort to do.

On CNC resolution, my CNC3018 does 200 steps per revolution on a 4.00mm pitch screw. This works out to a resolution of 0.02mm per step.

 

Hi,

The idea of using SRBF boards is interesting.

Does that 30 degree, 0.1mm engraving bit end shape look like a very sharp, flat, pointed triangle?

Also, did you have any trouble getting the PC board to lay really flat so that you can get a reasonably deep but not too deep engraving cut depth?
Or, did you use some sort of sensor that adjusts the cut depth over the entire board plane? And, if so, what happens if you do not use that sensor?

Thanks much for the info.

Yes, that's the shape of the bit - like this: https://routercutter.co.uk/solid-carbide-30-engraving-flat-point
Z-mapping is essential. The sensor is the tool contacting the copper layer.

 

Yes, that's the shape of the bit - like this: https://routercutter.co.uk/solid-carbide-30-engraving-flat-point
Z-mapping is essential. The sensor is the tool contacting the copper layer.

Hi,

Oh thanks, and by chance those are the bits I happened to have. I have others I used with the Dremel I had them for years so I might give those a try also for other things like 3d shapes. I wonder if there is a maximum diameter you can use for the cutter because I have some as large as 3/8 inch diameter with an 1/8 inch shaft (not for PC board work of course unless maybe you had to remove a lot of copper).

Oh so you suggest Z mapping. I do not have a sensor yet, so can you just somehow map the four corners using the tool itself?
For example, set the tool down to just touch the board at the upper right corner then somehow record that, then move to the lower right corner and record that, then the other two corners. The spindle would not be turning yet. Think that would work for now?
Of course it may be better with the spindle turning.

 

I don't think it is possible to successfully "etch" PCB on a CNC machine without doing Z-mapping, except if the board is so tiny (less than 1" x 1"). The copper laminate variation in thickness and board warp is going to result in uneven mill depths. I have done PCBs successfully after doing the Z-mapping. Z-mapping is done automatically by the CNC Candle software. It takes no effort to do.

On CNC resolution, my CNC3018 does 200 steps per revolution on a 4.00mm pitch screw. This works out to a resolution of 0.02mm per step.

Thanks again for the info.

Oh ok, I guess I have to get a sensor then. I am not *too* worried about a variation in mill depth as long as it removes the copper, but of course I don't want it to go too deep either.

Is there a special command you use with Candle to do the Z mapping, assuming you have a sensor?

Also, do you think the limit switches are mandatory?

Once upon a time I used a milling machine that must have weighed two tons and had a big platform for large objects. One use was to mill the bottoms of hydraulic pumps. It was not automatic though this was probably 50 years ago. It cut through cast iron like butter even with a 3/4 inch diameter cutting bit. I used it a few times but until now I never used an automatic CNC machine.

 

You don’t need a sensor. You are going to use the V-bit as your sensor.
The CNC controller board would have a sense input. When the sense input is grounded, that is detected as Z = zero.

You are going to connect the spindle motor bracket to GND such that the V-bit is now at GND. Then you connect the copper laminate to the Z-probe input on the controller board. The candle software will automatically step down the V-bit until it makes contact with the copper laminate.

The one setup you will be required to do is to setup the actual area to be scanned, X and Y limits. You also need to setup how many sample points to measure, i.e. you give the software the x-y grid of points to be sampled.

 

Installing limit switches is highly recommended. Sooner or later an error will occur and you will hit the physical limits of the machine. At the very least, you should install an E-STOP button.

 

Another thing to try to remember, when a limit switch is activated, all motors will be stopped.
To recover from such situation, the controller will be in manual mode.
Before you attempt to do anything else, stop, pause, and think carefully which direction you want to step the machine.
You want to make certain that you are driving the machine away from the fault condition, not further into it.

 

You don’t need a sensor. You are going to use the V-bit as your sensor.
The CNC controller board would have a sense input. When the sense input is grounded, that is detected as Z = zero.

You are going to connect the spindle motor bracket to GND such that the V-bit is now at GND. Then you connect the copper laminate to the Z-probe input on the controller board. The candle software will automatically step down the V-bit until it makes contact with the copper laminate.

The one setup you will be required to do is to setup the actual area to be scanned, X and Y limits. You also need to setup how many sample points to measure, i.e. you give the software the x-y grid of points to be sampled.

Hi,

Oh ok that sounds good, thanks again.

I can't wait to try it but I am still putting this dang thing together.
One of the problems with the assembly is the 'nuts' they use are the type that slip into a groove and the groove keeps them from turning. However, if they don't catch the groove, the nut tightens but it does not grip the side bar frame. It's difficult to get the nuts to catch just right because I have no access to the inside of the side frame. Kind of a dumb way to design it, and there are like 100 little bolts with nuts like that.

 

One thing to avoid is fitting a drill when a mill was called - bad things happen. Don't ask me how I know!

 

Hi,

Oh ok that sounds good, thanks again.

I can't wait to try it but I am still putting this dang thing together.
One of the problems with the assembly is the 'nuts' they use are the type that slip into a groove and the groove keeps them from turning. However, if they don't catch the groove, the nut tightens but it does not grip the side bar frame. It's difficult to get the nuts to catch just right because I have no access to the inside of the side frame. Kind of a dumb way to design it, and there are like 100 little bolts with nuts like that.

I must have missed this somewhere. Did you acquire a CNC machine? Which one did you get?
I think they are called T-nuts. There are tricks you can do to install them.