[Carl] wanted to prototype his circuits quickly using printed circuit boards. He picked up a Bantam Tools Desktop PCB Mill and made a video about the results. His first attempt wasn’t perfect, as you could notice under the microscope. A few adjustments, though, and the result was pretty good.
Be warned, this mill is pretty expensive — anywhere from $2,500 to $3,000. The company claims it is a better choice than a conventional cheap mill because it uses a 26,000 RPM spindle and has high-resolution steppers. Because of its low backlash and high accuracy and repeatability, the company claims it can easily mill boards with 6 mil traces.
Of course, the mill can do things other than PCBs. [Carl] was impressed with the speed of the system, too. The boards he tries in the video are pretty small, but they took a few minutes each.
Of course, like most homemade PCBs, there are no plated-through holes or solder mask or plating. Of course, you could add all of those things using additional steps. We’ve seen people use wires for vias or even rivets. However, that sort of takes away from the main idea of push a button and PCB pops out.
We’ve looked at using a cheap mill to do the same thing, and Hackaday’s own [Adil] found that 0.3 mm traces (not quite 12 mils) were easily doable. If that’s sufficient for your needs, you might save quite a bit of money over the mill presented here. We’ve seen others do 10 mil traces, so that’s probably doable, too.
I’ve had one for a few years. If you need to rapid prototype a design or are making a few of something, it’s ideal. Perfect for the single/double sided smaller stuff / SMD where breadboarding doesn’t cut it. Then send the board out for making when it’s ironed out.
sponsored by JLCPCB.com :P
Milling is great for quickly whipping up a board, beats chemicals in most respects, vias are a problem each way, but $2500 to $3000 is WAY expensive for a rather limited mill like that. I didn’t top $2500 for my Taig CNC mill, which is drastically more capable.
I acknowledge that it’s a small market, but charging those prices for a machine made of sheet plastic, linear rods and other simple components is a bit crazy.
Depends on the level of customer service and support you buy into. Similar things have been said about Lulzbot for example, but they do treat their customers really well by all accounts.
If the fixtureing is repeatable enough can always have it drill the holes take it off for a quick dunk in the copper depostiting baths and then mill on it. Personally think I’ll stick with just drilling the through hole and soldering wires between sides – simple.
That said I think if I was doing more serious pcb one-offs/proto types I’d build build my own for that price tag, and to be more useful to me would have to be a seriously expensive build as most of what I do that could use my own cnc is larger- trying to get accurate and large working area in smallish space is a serious challenge…
I think they’re competing with the likes of LPKF, whose pcb mills start at about $20K. Getting the very high speed spindle is a significant chunk of the cost of this, and taig and sherline headstocks certainly have the accuracy but nowhere near the speed of these. I run my sherline at about 8K for doing PCB’s, and the edges of the copper are nowhere near as nice as when I run the same board on my LPKF.
There’s also value in having software already set up to flip the board geometry side-to-side for topside-to-bottomside and having the axis of symmetry already set up for you.
Anyone who is doing boards like this, here’s an assembly tip: take a long piece of wire and string it through all the vias, like lacing up a shoe, and solder them, then cut off all the excess wire. It is so much faster and more precise than cutting off snippets of wire and soldering one via at a time, with way less chance of bridging because the wire’s twisted or moved during soldering. No chance of little wire snips melting and falling out while soldering on the back side and melting the front side fillet, either.
Those are the low-end LPKFs, though. The advantage of those is that they can upgrade to very capable machines: 100k spindles and precision depth control.
Higher spindle speed is definitely advantageous for PCB milling, but as the entire mill is about a 1 foot cube, it can’t be a massively expensive spindle. I acknowledge and agree with Foldi regarding support. If they’ve made it a very simple process, that’s definitely worth something, although $2500 and up is still a lot.
And awesome tip on the wire vias. I’ve spent many a expletive soldering them individually and having many drop out the bottom when I’d try to flow a little solder on the top.
dave over at eevblog went over custom pcb not to long ago and with the prices dropping the way they are on custom pcb’s unless you absolutely need something right now this gadget is not worth it. below is a link and it is not about solar freaking roadways or the batteriser ;) it is about 1hour and 9 minutes but a good watch
https://youtu.be/mT2Vp3HcIP4
For me it is absolutely the right-now aspect. The stuff I do often goes through five revisions before I’m ready to send it out for fab, and being able to do all five on copper over the course of an afternoon has really helped. Similarly, both at work and home I often run into stuff like needing to add a buffer, or needing to transiently short output lines to ground, and I can have that working and have data the same day. 90% of our stuff goes to pcb fab houses, but a three day turn time for that remaining 10% would be really painful now that we’re used to having two hour pcb fab capability.
yea and dave doesnt live in a part of the world where you pay and extra 35 dollars for DHL shipping just so you can get it in your hands in under a month, that crap adds up
I did not try (yet) but I think my mpcnc could do this.
Might mention the AntPCB, seems like a nice little machine, definitely more DIY route tho.
AntPCB does supposedly get down to 8mil traces.
is this really so much better than an 3018 mill which can do the same?
Ah yeah, this is on the round tuit list, hot glue the spare dremel to the guts of a couple of CD drives.
Looks like a machine that works, but it has the same simple parts as all the other cheap ones and it does not seem to warrant it’s USD2500 price.
Maybe the axles of the linear guides are a bit thicker, which is a plus because of more rigidity, but still, it’s nothing special.
Completely supported rails such as even the thin MGN12 (clones) are much better. and I wonder why you seldom see them on routers and light milling machines.
I just bought a few of them on Ali just out of curiousity and to experiment with.
A single rail with one linear guide cots around USD15.
If you buy a bunch of them and screw them on some pieces of thick plywood you already have a better base then this bantam machine.
For this level of machine, adding stiffnes is always a plus. Think on the order of 40 mm thick solid plywood, or 22mm plywood with sheets of aluminimum or steel glued to it’s sides.
Addition:
This EUR1600 Sorotec is probably already much better quality then the bantam thing.
https://www.sorotec.de/shop/Configurator-Hobby-Line-10560.html
The challenges of this method include bed flatness, machine vibration and tip/cutter sharpness. I used a Proxxon MF70 with a German parallel port interface for years, more recently an Arduino Nano and GRBL to gain auto leveling. It was always a major job getting usable pcbs from this low cost sub $700 system. Cleaning the rough cut edges and shorts from copper whiskers was a time consuming drama every time. I recently got a cheap Chinese 5W laser engraver and it’s a game changer. Now I can do beautiful pcbs with super fine tracks and spacing in under 30 minutes, excluding the time for the paint to dry on the blank boards. Never ever (ever!) going back to milling, except to trim laser pcbs to size.
“We’ve seen people use wires for vias or even rivets. However, that sort of takes away from the main idea of push a button and PCB pops out.”
It’s just a mill. It’s not like it’s a mill, pick n place and solder oven all in one. (that would be cool though).
You’re already going to have to solder the parts on anyway. If we are talking about a small project and you keep the number of transitions between the layers to a minimum how hard is it to just line a couple of pads up on both sides and solder a wire through? Better yet the leg of a through-hole component can be soldered on both sides.
Sure, for bigger projects with lots of vias or for mass production that kind of sucks but for a lot of hobby projects I don’t see the issue.