First attempt at CNC milled PCB

[retailer]

I made a small start on a 'fixer upper' AVO CT160 that I purchased from a fellow forum member a while back - amongst other cosmetic and possibly electrical issues the meter movement is open circuit, I'm planning a movement transplant with an op-amp to drive the 'new' movement and decided to try my CNC mill at making a circuit board. Once the circuit was drawn up in Fusion360 (free hobby version) the circuit is transferred to the board router where one can move the components around for best placement - I chose to manually route the tracks but there is an auto router that will do this for you, I used 30mil tracks (30 thou' or around 0.75mm). Once happy with the layout the board is transferred to the CAD/CAM or manufacture screen where one needs to decide on cutters and tool paths, this where I initially stumbled - after watching a myriad of youtube videos I was still non the wiser and had produced 5 or 6 failed boards so I resorted to 3 days of experimenting trying different options, all the time telling myself that if I had used veroboard the job would have been done days ago.

I ended up using the 2D contour operation to mill the tracks, even though I did not intend to use it, I set the cutter as 4 flute with diam 0.05mm and material to mild steel - the max spindle speed of my mill is 4K RPM so I needed to slow the whole thing down, once all of the tracks had been selected I was able to simulate the toolpath to double check that all was ok before generating the GCode. I used a 30deg engraving cutter and had to add a Z axis offset to reduce the depth of cut to 0.05mm, just enough to cut through the copper - I was quite pleased - after the initial 5 or 6 failed attempts I had a board that looked presentable, total time to mill the tracks was around 4-5 min. In the past I've always etched PCB's and used a variety of methods - spray on positive photo resist, press n peel, Dalo pen, nail polish, photocopy transfer etc, etc, etc - having now gone though the milling process I doubt that I'll ever etch a board again.

[mole42uk]

Very interesting! That looks to be a really good use for a small CNC mill. The appearance reminds me of the Plessey PCBs of years ago, ISTR they printed the copper onto the substrate rather than etching.

[Malcolm T]

Impressed . What is the CNC machine you are using and how do you hold down the board to the table?. Beats chemical etching .

[Terry_VK5TM]

Nice job.

Just check that there are no shorts - I can see what appears to be two places that haven't quite milled properly.

[Julesomega]

A very convincing case that we should all have a CNC router - I certainly want one
I've been looking at what's online or in the shops - there's a UK manufacturer I found once, or of course Banggood have a wide selection of routers and laser cutters, though the latter is not relevant for PCBs. The problem for me is I have not found any cutter bits less than about 2.5mm dia. It seems that a spindle dia. of 5mm is the standard, but where do you get 0.05mm cutter?

Banggood 3018 Mini Cnc Router

[Phil__G]

A while back I had a batch of boards professionally milled as it was urgent and much quicker than waiting for fab. It seems to me that yours are just as good as the pro's, well done!
Phil

[G0HZU_JMR]

Realistically, trace cuts of 0.004" width are possible but I would recommend 0.008" as a practical minimum with a PCB router/mill unless you are prepared to spend lots on money on the best tools and the best PCB laminates that give the cleanest cuts.

I bought a secondhand T-Tech 7000S PCB router about 20 years ago and still use it. At the time this was the biggest machine available and it can handle boards up to 19" x 13" in size.

https://t-tech.com/store/systems-and...ircuit-qc7000/

I think the tool chuck size for my machine is 1/8" and this is fairly standard. I haven't bought any tools in many years so I have forgotten. I have boxes and boxes of tools here of various types and size. This includes end mills, routers, drills and V cutters. The software and hardware also allows double sided boards to be milled.

The main limitation with these machines is the operator. Either the operator has the finesse and patience and skill to get the best from these machines or they will blame the machine for their poor results and maybe also damage it at the same time.

These days, the option to buy good quality PCBs made in China for very low cost means that PCB milling is becoming less attractive. I generally use mine for milling exotic PCB laminates or for making laminated tools or for engraving. At work the company bought a PCB mill back in about 1994 and one of the directors bought one even earlier than this. Where the PCB shines is when you need to mill a board quickly and maybe have to do several iterations in a day. This might be for a printed RF filter or a coupler. It can also be very useful for rapid prototyping of a piggy bodge board to add to an existing PCB.

[G0HZU_JMR]

Here's a couple of old images showing some fine trace cuts. To give a sense of scale the finest cuts in the SMD board are 0.01" (ten thou) wide.

The first image has a metric ruler next to it. This was done with my old T-Tech machine. Modern LPKF machines will be easier to use and will give better results.

I think my machine was manufactured in about 1995 and the original owner of my T-Tech 7000S was Hewlett Packard up in Scotland. According to the LPKF rep HP lost the (ultra expensive) Isopro software dongle rendering the machine useless to them. I was able to buy it very cheaply because of this!

[G0HZU_JMR]

As Terry mentioned, there are a few places where there may be a short and this often happens even with the dedicated machines from T-Tech and LPKF.

I'd recommend trying out a green kitchen scouring pad to improve the PCB finish before milling the board. This helps give an even cut and the board will be easier to tin and solder. I use the flat green scourers and these offer a fairly even and shiny PCB finish after 'running them in' on some old scrap PCB material beforehand. A hot soldering iron can also be useful for removing any residual copper from the PCB after milling.

Some software packages offer a 'rubout' function where larger unused areas in the PCB get removed via an efficient zigzag action. This takes a while to complete but it makes the PCB easier to solder with less risk of generating solder shorts.

See the orange rubout pattern in the little RF PCB below. Another useful tip is to always leave a couple of tiny gaps in the router cutout pattern. You can see this in the pink traces in the upper left and the lower right corners of the image below where there is a tiny gap between the pink traces. This tiny gap prevents the board from lifting up after it has been cut out using the pink router traces. If your machine has a vacuum extractor then the vacuum can lift up and try and swallow the PCB if it is small enough.

[Ed_Dinning]

Hi Folks, we used these at a couple of firms I worked at in the 80's. They were designed to produce quick prototypes, either double sided or copper plus annotations.

They worked from the Gerber file outputs, ran from a PC and were dongle protected. I wonder what happened to them all?

Ed

[Julesomega]

Thanks Jeremy, they list router bits down to 0.6mm. Also what about using a laser engraver? Fine depth of cut is now possible with a 0.02mm focus area as in this ATOMSTACK A5 M40 Laser Engraving Machine
Has anyone used such a machine?

[retailer]

My mill is an old Arboga - made in Sweden sometime in the 1950's converted from manual to CNC - it is a good hobby size - I've had it for over 20yrs - I started the conversion in May last year - here's a video of the first trial I did in Dec that year - it's not very exciting and video is a bit wobbly as I was trying to keep the E-stop button within reach just in case. https://www.youtube.com/watch?v=IgIwLlxmeOA

I was contemplating a small 3d router such as the 3018 from ebay/bangood/aliexpress which is very affordable at just over $AU200 - it's capable of not only pcb milling but also engraving such things as front panel artwork - but eventually decided on the conversion as I would be able to do real world milling of not only aluminium but steel as well. If you take into account the cost of etching solutions, photoresist film, chemicals etc and the fact that a small CNC mill can also engrave panel artwork, then for me it is worth the investment

With regard to the cutter I don't have a .05mm cutter - that is quite small and I don't even know if they available - I have ordered some 0.5, 0.6 and 0.7 mm flat bottom cutters, just waiting for them to arrive - Fusion360 lets you specify your own cutter parameters so I designed a .05mm cutter as this is the approx width that a 30deg D bit engraving cutter will cut if it is set to a depth of 0.1mm, once the tool path is defined Fusion shows how far away from the tracks the tool path is - the copper between the blue line and the track would be milled away - I could see that the track would be left intact which proved to be the case - this is where I was stumbling at first - I initially was using the Fusion360 engraving option and Fusion was placing the point of the cutter directly over the edge of the track, being tapered the width of cut increases as the cutter was plunged down into the board, which resulted in the track being milled away. The engraving cutter throws up a burr, which a few have pointed out looks like imperfect milling - I'm not sure how to fix this or if it can be fixed - its not that hard to go over it with a fine file which removed 99% of it, but under a strong magnifyer a burr is still evident, I have soldered some components on and the burr is not causing any issues. I have fitted a donor meter movement into the CT160 meter case - not wanting to wander off topic I made this a separate post in the test gear section.

The pcb board was clamped to a bit of mdf board which was itself clamped to a milling vice, the pcb board should be perfectly flat - in youtube videos some use double sided tape some use hot melt glue and others use clamps

G0HZU_JMR: thanks for the tips - I still have a few things to learn when it comes to PCB milling - the next one when it comes along is going to be double sided.

[duncanlowe]

Some years ago, I came across someone offering this service on the internet and had a few one offs produced as prototypes for some projects. All of them were good quality with no shorts. Which unfortunately couldn't be said of a small batch production run of one of those baords where 1 in 4 boards had a short, and not all in the same place. The manufacturer was entirely uninterested in the lack of quality I reported.

[retailer]

A small update :- The first board made is completed and working ok, as I was looking back over the files I realised I had made an error in post 12 - I set the tool path with a 1/32" cutter not a 0.05mm cutter as I stated in the post 1/32" is very close to 0.7mm, a senior's brain fade moment from having tried so many toolpaths, milling methods and cutter sizes - I still used a 30deg D bit engraving cutter to actually mill the tracks, - I also noticed something else - there is a few spots on the board where the tool path of the 1/32" cutter overlaps - I watched the onscreen simulation all the way through and noticed that the cutter lifts off and does not cut the portion that is overlapped - this ties in with the comments made that the tracks do not look completely milled out in a few places - the blue line shows the path taken by the cutter - specifying a 0.5mm cutter results in no overlap og the tool paths. In going over the board the first time it looked like a burr had bridged the tracks as it didn't take much to clear it out with a scalpel
Thinking that if I had been on the ball I could have made a board that would fit neatly on the back of the meter eventually got to me and I decided to make another board - this one took a bit more time to mill as the board was not perfectly flat and level, to within a few tenths of a mm - the copper layer is only around 0.035mm thick so if the board is bowed, warped or not level the depth of cut will vary, this is what I experienced - I let it run until it finished, reset the Z axis origin 3 or 4 tenths of a mm lower and re-ran the program, that cleared up the few bits that had not been cut away.