What’s your biggest obstacle using Laser/CNC engraver for rapid PCB prototyping

[otto_weird]

Recently, many desktop laser and CNC machines have hit the market. Videos like Stephen Hawes’  show that they can produce clean, single-sided PCB traces quickly—but despite this, they still haven’t become as popular as 3D printers.

I’m curious why. Is it because the machines are too expensive? Because they can’t drill vias? Or simply because commercial PCB fabrication services are already quick and affordable enough?

Let me know your thoughts!

[nctnico]

Lack of ability to create vias is what kills taking the encraver route. The only niche where an encraver can make sense is prototyping RF circuits but then again these can be simulated quite accurately nowadays. I've had various etching machines over the course of 30+ years but in the end I never used because ordering a board makes so much more sense. I think I have etched the last board over 10 years ago. I got rid of my etching setup as I could make better use of the space.

[Skashkash]

Gonna hop on here and agree that vias are the big hold back.
 In the past I've made many hundreds of pcbs inhouse using various methods including etching and high end engravers (LPKF).  Never found a workable solution for vias.
   Admittedly, it was great to be able to make simple boards in a couple of hours, but most of the more modern and interesting SMD devices require much finer detail.

   These days I just send out to oshpark or china for prototypes. Four layers, and a soldermask in a reasonable (IMHO) amount of time and money.

   As a bonus I no longer have to worry about breathing in fiberglass or spilling ferric chloride over everything. Those stains never come out.       

[PlainName]

That's a great idea to have a via grid. Vias are, indeed, the big problem with DIY PCBs.

[Geoff-AU]

Via grid is a nice solution for extremely simple boards.  It would be a real limitation otherwise, I'd rather wait a few days for a proper PCB with vias where I want them and better signal integrity.

[CatalinaWOW]

Another vote for vias.  Also, even if you use rivets or other methods to obtain side to side connectivity, lining up the two sides is at least as much trouble as it is when etching.   

The final thing is that the range of applications seems smaller for this CNC etcher/routers.  They are too limited to do much machining which leaves art and PWBs as their main output.  I have a full size (300x600 mm) laser cutter, and like using it to make boxes, puzzles and similar things, but find myself using the 3D printer to solve many more problems.  Replacement parts around the house.  Jigs and assembly aids.  Knobs and bezels.  Enclosures.  Molds.  Paint masks.

[Smokey]

The reason more people don't make their own PCBs is that electronics is hard. 

It's not software. 

You need a bench of test equipment and soldering stuff.  You need experience working with little parts (lots of the people I know are too blind to deal with little fiddley assembly, even under a microscope).  You need experience reading schematics and understanding how basic circuit stuff works.  You need experience troubleshooting why your homemade boards don't work and if the problem is actually in the board, or the circuit, or the firmware, or the test equipment measurements, or.....

Most people with all the required experience to get a thing working are just getting JLCPCB/PCBWay/etc boards and calling it a day. 

[PlainName]

Most people with all the required experience to get a thing working are just getting JLCPCB/PCBWay/etc boards and calling it a day.

Clearly, I am not most people.

Often I want a simple thing and I want it right now, not in 2 weeks when I've forgotten what project I was meant to be doing. They tend to not be Herculean products but a small part of something. For instance, I'm slapping 12V van lights onto the electric gates and need a breakout for some gate manufacturers connector to my simple 24->12V breakout. Not going to wait 2 weeks to find out if it works, and really don't want to use perfboard. Knock it off on the mill in half an hour and I'm already moved on to something else.

[bsdphk]

Disclosure: I'm the guy who invented the "height probing" technique everybody uses now, and which is what made CNC'ing PCB's with contemporary trace-widths possible in the first place (https://phk.freebsd.dk/CncPcb/).

For most contemporary electronics, the missing vias, and to a lesser degree the missing solder-mask are major downsides.

But for me personally the real turn-off is that I do not make PCBs often enough.

By the time I wanted to make the next one, things had been shoved aside, because desk-space is always a scarce resource, and software had been updated, producing slightly different output, and by the time I had the entire setup humming again, real PCBs, with vias and soldermask would have been in my mailbox days ago, and I could have spent the time much more pleasantly.

But if you make PCBs frequently, every two or three weeks, and can live with, or work around, the missing vias and soldermask, there is something very satisfying in seeing the traces emerge from behind the bit, and being able to boast: "I made it all by myself."

(I had hoped to be able to paint or draw soldermask with my CNC, but I cound not find a usable substance or applicator)

[buxtronix]

I use a Proxxon MF70 to occasionally mill small PCBs - I'll have a burst of activity for a few weeks then go idle for months.

It's great when I want a simple board right now - for example I am prototyping something that doesn't have a lot of complexity or vias. I can reliably mill down to at least 0.8mm pin pitch (eg QFP). I find it's faster to design and mill a board than futz around with working out a layout and laying a few dozen wires on a proto board.

The trick is to keep the setup consistent - I have an old 2013-era dedicated laptop for it, running Flatcam and UGS, and I never update those to new versions except once every few years, so I dont get nasty surprises. I've also written up a short reference doc to remind myself of the process if it's been some months.

Generally I can go from a Kicad design to a finished board in about an hour give or take. For anything more serious I will use JLC, but a simple one-off it's hard to beat a mill.