PCB making by copper removal using CNC milling not popular, why?

[Merlysys]

I came across this machine which cuts out unnecessary copper from a copper clad PCB, that eliminates several steps including the messy chemical stripping.
But I don't hear about this process despite the advantages, is that because it is not well marketed, too costly, or what?

Forward to about 3 minutes to see it in action

[wiss]

I used one some 15 years ago, it took 40 minutes to route a (small) board and that was without removing all copper.
It was also very noisy...

[CrashO]

Noise, speed (it takes a lot of time to mill away all those lines), initial costs (CNC machines don't come cheap), running costs (Isolation Milling tools wear down pretty fast and aren't all that cheap, same goes for the drill bits). You need ventilation / suction to remove all the (hazardous) pieces of fiberglass, etc.

Another thing to consider is precision. With enough practice (and good clean etching solution and stuff), 10mil or smaller precision can be achieved with photo-etching. Must pcb cnc's don't go down to that precision. And the ones that do are even more expensive.

For small time hobbyists etching is the best bang for buck way to go. And if you have the time, just to order them for $10 per 10pcs's in China..

[wraper]

But I don't hear about this process despite the advantages, is that because it is not well marketed, too costly, or what?

The only advantage would be if you want some dirty prototypes fast. Otherwise such PCBs are inferior in quality to any other methods.

[necessaryevil]

I think milling or routing is populair for large scale production, but I can be wrong. For hobbyists, milling or routing will be expensive, and it won't be easy after all.

That being said, my old university has one of these: http://www.lpkf.com/products/rapid-pcb-prototyping/circuit-board-plotter/index.htm. Specially designed for pcbs. These could be easy to use, I don't know, but for sure won't be cheap.

[CrashO]

That being said, my old university has one of these: http://www.lpkf.com/products/rapid-pcb-prototyping/circuit-board-plotter/index.htm. Specially designed for pcbs. These could be easy to use, I don't know, but for sure won't be cheap.

I use a ProtoMat regularly at my Uni but would prefer not too. It's great for (lazy) fast prototyping, just flick the switch and wait. But its disadvantages are clearly present.
The milling tools only last a dozen or so boards and have to be changed out regularly. The resolution is not that great, and more than often (albeit, probably caused by the uni running the tools/drills past their life) copper is left between pads or tracks when clearance is under 16mil..

So the way I usually work is I design some quick (and unnecessarily bulky (at least 16mil everywhere)) bulky proof of concept. Pop it in in the morning and pick it up an hour or more later, solder it, test it. And then design the real prototype and have them produced in China for a few bucks.

[rx8pilot]

It's not terribly popular because of what other have mentioned.

Slow.
Expensive.
Loud.
Dust control.
limited capability.

I had looked at the LPKF machines and got quotes for $30k-$50k USD. To do multilayer requires another machine that sounded like a real challenge. Silk printing was yes another machine on top of the others. I do my prototypes at Sunstone Circuits which are not the cheapest. In general, my prototype runs of 2 pcs are about $200 for a 4 layer odd shape PCB with silk screen. I would have to do a massive amount of board iterations to make that money back. LPKF touts you can get the PCB the same day, but it takes a person to do it along with the milling equipment. I usually have to gather other parts and prepare firmware anyway - so a 3 day wait for PCB's is not so bad.

On the hobby side of things, plenty of people have devised their own 'low cost version' of a PCB milling machine. With those, they all have the same issues as any other milling machine - they either suck beyond being usable or they are expensive and complicated and work. Cheap and milling are two terms that do not go together - I have tried for 2 decades to prove that wrong. In the end have have ended up spending 7 figures on milling machines that actually work.

Making your own PCB's requires a special circumstance like geographic isolation - ie you live in the middle of no where and the 10's of thousands of PCB fab houses around the world are still too far away and expensive. Maybe your design is triple top secret and no one else can be allowed to see it. Otherwise, I cannot image why PCB milling or chemical processing would be done DIY.

[liquibyte]

Otherwise, I cannot image why PCB milling or chemical processing would be done DIY.

Because you can and it's a challenge.

[free_electron]

because it is practically unusable

here are the problems wit this stuff

- singlesided works , anything else don't as you cant make thru-hole metallisation
- board needs to be perfectly flat or you get variations in trace width. they bank on 45 angle mill bits. the depth controls the width of the cut ... if your carrier material is not perfectly flat you get problems
- works for isolation cuts only. rub-out ( like a real board ) costs a lot of time and a lot of tool changes. i can etch a eurocard ( exposure, develop, etch ) in 20 minutes. rubout on the same board costs 2 hours... if o tool breakage occurs and i can etch finer structures than the milling machine can handle.
- tools bits are expensive
- noisy
- dusty ( and you do not want to breathe milled Fiberglass !!! )
- slow
- below 8 mil track and gap problematic.

for the cost in materials ( bare board , milling bits ) to make  a single-sided pcb you can get 50 made in china , 2 layer including soldermask and silkscreen. on a 5 day turn basis.  so why bother ?

[awallin]

The LPKF machines work quite well once you get the hang of it. I've only used 10+ year old ones, so the newer (30-50keur) machines should be quite a bit better.

The main issues IMO are:
- vias need to be inserted/punched/soldered
- no thru-hole plating
- the copper layer isn't attached very well to the PCB-material. Small pads tend to lift off if you change components once or twice while prototyping.
- no soldermask ofcourse, makes soldering small parts so easy on commercial boards...

It's really good for prototyping where you have an idea in the morning, make the schematic and layout mid-day, and solder and test in the afternoon.

[G0HZU]

These machines are best suited to RF research work and I've used T-Tech and LPKF machines since the 1990s and I think that no serious RF design lab should be without one. They are simply brilliant for designing oneoff prototypes for RF filters, oscillators or other smallish test circuits.

Nothing else competes on time, cost and versatility for RF board design because you can mill a suite of simple circuit designs on any laminate you choose in under an hour. i.e. design it on a PC, then mill it and then test it in a morning.

I've got a couple of T-tech machines here at home and we use an old LPKF machine at work.

However, I don't think they are suitable for general purpose PCBs using FR4 and they are at their worst if you want to mill a large board with lots of digital components. It takes ages to mill all the tracks and it isn't cost effective.

Therefore, the key to success with these machines is to be milling small and simple RF PCBs on exotic PCB laminates, especially if you want a PCB with a non rectangular outline. They are also useful for making laminated tools or for making simple front panels.

Like I said, nothing else can compete for the kind of work I do. But if I wanted to make a few MCU based boards that were fairly large and complex and I wanted a decent service life from the board and I wasn't in a rush for the boards then I'd just order the PCBs from a cheap PCB manufacturing house like most people do   The T-Tech or LPKF milling machines can't really compete here unless you absolutely needed the PCB the same day and were willing to spend an hour or two milling it and put up with the lack of proper PTHs and lack of solder mask etc etc.

You can also buy them cheaply if you are brave and buy an old/ used T-Tech 7000S machine. I bought my first one maybe 12 years ago for about £800 complete with tools and materials and Nilfisk extraction system and bought a second one (albeit with worn bearings) for about £600 a few years later. This also had tools and materials and a Nilfisk extraction system.

The first one paid for itself within months because I milled a series of RF filters on Rogers material for a local company and they paid me £70 per PCB for a same day service.

[G0HZU]

- singlesided works , anything else don't as you cant make thru-hole metallisation
- board needs to be perfectly flat or you get variations in trace width. they bank on 45 angle mill bits. the depth controls the width of the cut ... if your carrier material is not perfectly flat you get problems

You can buy a crude rivet based system from LPKF for making PTHs for double sided boards. It is very fiddly and time consuming to fit them but it does give reliable results for RF prototypes etc.

board needs to be perfectly flat or you get variations in trace width

For very accurate milling I use end mills rather than the angled or pointed tools. I've got various end mill sizes from 0.005" through to about 0.04". These mill quite accurately even if the cut depth changes. But they are expensive!

[IconicPCB]

The issue of plated through holes is a separate topic to mechanical ablation of copper.

I make multilayer boards in 0.2mm core material and two sheets of prepreg up to eight layers using mechanical ablation.
The process is stable repeatable albeit slow. It is slow because only one tool is doing the work and copper is removed in sequential operation unlike chemical etching which works on the entire panel at once.

Any discussion of riveting is equally applicable to mechanical ablation as it is to chemical etching and needs to be viewed as a separate issue.

The topical finish ( soldermask, ident, metalization finish ) are equally divorced from mechanical ablation as they are from chemical etching.

The question of flatness of bank laminate is a problem which has been addressed in two ways to date.

The first method is a software  approach and to my way of thinking only a band aid solution.

The blank panel is attached to the bed of the machine. This is followed by a grid based probing of the laminate using a contact probe.
The probe is so arranged as to generate  a signal upon contacting laminate. At that point Z axis is recorded to compensate the laminate deviation from flat. The compensation is applied to the G code file  hopefully minimizing the effects of flatness imperfections.

The second way, a better solution in my opinion is a floating head on the milling machine.  The head is allowed to move freely in  Z axis over a limited stroke length while resting on a pressure foot. The net effect is the depth of penetration of the cutter is fixed and maintained by the floating action of the head.

[necessaryevil]

...The milling tools only last a dozen or so boards and have to be changed out regularly. The resolution is not that great, and more than often (albeit, probably caused by the uni running the tools/drills past their life) copper is left between pads or tracks when clearance is under 16mil..

I see the problems there. Could be a bummer in certain scenarios. But it still sounds like it's very usefull for universities, I guess using components which are a little bit bigger doesn't make much of a difference for pcbs which aren't going in to mass production in a later stage. So I'm still jealous jou have acces to one ;-).

Do better machines exist the way? 

[IconicPCB]

I use tools which will clear a gap of 0.004"

[G0HZU]

In terms of setting the cutting depth accurately I've milled a 2 sided flexi PCB that feels like copper coloured printer paper but actually it's a two layer PCB with an inner dielectric that is incredibly thin and flexible. It feels just  like regular paper! The T-Tech machine can do this because it uses the 'foot' method to set the cut depth and the trick is to stick the PCB material to a very hard and flat backing material before milling it and to use a (brand new) precision pointed milling tool. The cut depth is set by a multi turn vernier control.

[nctnico]

It's really good for prototyping where you have an idea in the morning, make the schematic and layout mid-day, and solder and test in the afternoon.

I can do the same with my 20 euro (including chemicals) etching machine which sits in the shed. Having to deal with fiber glass and copper dust is much more of a health risk than using ferric chloride.

The only area where milling has an advantage is for RF boards where the trace width need tighter tolerances that you can achieve using a home etching machine. OTOH you could work around that by choosing thicker material so the traces may be wider and the variance in trace width affects the impedance less.

[SundayProgrammer]

I have found engraving very handy way to get prototype PCBs. With one machine I can get drilling, cutting and engraving done. Using my DIY machine I can go up to 0,5mm pitch and the smallest trace I have done is 8mils.
Someone said that fine pitch work is very difficult because the material is not completely flat. It is true and without calibration the result is not good. I run surface calibration first so it does not matter how twisted the board is.

Axel.

[G0HZU]

I can do the same with my 20 euro (including chemicals) etching machine which sits in the shed. Having to deal with fiber glass and copper dust is much more of a health risk than using ferric chloride.

The only area where milling has an advantage is for RF boards where the trace width need tighter tolerances that you can achieve using a home etching machine.

The milling machine can also drill holes quickly and accurately and it can cut out complex shapes very accurately including cutouts within the PCB and also it's great if the PCB itself has a complex outline.

[nctnico]

That still leaves the glass fiber and epoxy dust problem. Besides that I use SMT components so the number of holes I need to drill afterwards is 4 or 5.

[G0HZU]

The milling machine can also register two layer PCBs extremely accurately, it can also cut very precise large circular cutout holes for connectors and is especially useful if you want to use the PCB itself as a PCB based front or rear panel because it will mill the SMD artwork and will also cut the outline accurately and also register the screw holes and connector points really well. You can also deeply engrave text into a PCB or panel or aluminium sheet and then infill it with paint even with exotic fonts.

[liquibyte]

Man, you guys worrying about milling dust from PCB's should never, ever go work for a boat manufacturer.  I've worked two and if you think milling PCB's creates hazardous dust, let me tell you, mounting a bridge on one yacht creates more fiberglass dust than you'll ever encounter in a lifetime of milling PCB's.

[SundayProgrammer]

I have desktop milling machine in my flat and I use vacuum cleaner. Actually it is the cleanest way for me to get PCB done (engraving, cutting and drilling).

Axel.

[IconicPCB]

Fiberglass dust is much less of a problem than solder mask dust.
In any case my machine has integrated suction system which consist of a cyclonic chamber where most of the garbage ends up and a hepa filtered vacuum cleaner.

The cyclonic dust collector is in fact an in-line attachment which can readily be used on any vacuum cleaner.

Yes dust can be a problem but very little dust gets kicked up during operation without suction.