Hey John... Mill your protos with this...

Note how careful they are to not show the other side. Or the assembled board.

snipped-for-privacy@highlandsniptechnology.com wrote in news: snipped-for-privacy@4ax.com:

You're mumbling again. It has to be orders of magnitude better than you and your handheld Dremel solutions.

Note how careful you are to bring up something that not only doesn't matter, but they were not "doing that".

mandag den 19. april 2021 kl. 17.02.33 UTC+2 skrev snipped-for-privacy@decadence.org:

have you ever actually tried milling pcbs?

Generally worse. Milled PCBs look cute, until you actually try to do one.

Doesn't matter? That board obviously has vias. How do they connect to the other side? What does the other side look like, and how did they keep the sides aligned? How do they manage power and ground and bypassing?

I don't think that's a real board.

A serious board, with several ICs and such, needs a proper multilayer stack with a ground plane, power planes, plated holes, and solder mask.

I dremel simple circuits, mostly to test parts in a couple of hours. Sometimes we do a quick-turn 4-layer proto if the parts are complex.

But the best thing to do is design your final product, release the drawings, and manufacture a batch. If you do that right, it will work and you can sell rev A.

You've got to do the real thing eventually, so why waste time on prototypes?

We've tried a couple of machined boards, on our Tormach, but it's not worth doing.

One of our customers had a PCB milling system which they didn't use much, so they essentially gave it to us. After a while, we gave it back.

I don't understand the point of going to that level of effort to design and home-manufacture a whole circuit you don't have the ability to make quantity of.

That machine isn't amenable to volume it takes too long. You're still going to have to design a regular PCB. So just do that!

Right. I'm running a business, it's a small business but it's still a business I'm not a hobbyist (at least when I'm not.) I don't build full physical circuits of anything I don't expect to make more of except if it's something for personal use and even in that case it's often faster to do a dead-bug it or get a real PCB made.

I have a bunch of lil circuits in a folder that will probably never be built. Oh well.

I've tried it on a CNC router, the auto-level software is really necessary to get it to work on a reasonable size board. I made a board for a DIY sump pump alarm.

You can do it with a Chinese CNC router, Flatcam (free) and open-source levelling software but I don't think it's worth it to make boards that look nothing like even a cheap commercial board.

Better to wait a week and get boards with 2 or 4 layers, solder mask, fine traces and holes etc.

P.S. if that guy keeps handling those bits like that he's going to stab himself. A 10-15mil bit makes a really nasty little hole.

What does "a regular PCB" mean? I suppose you need to allow a bit wider spacing than 4 mils, but there doesn't need to be anything special about the board design does there?

What is auto-level software? I saw him using a continuity tester. Is it so hard to zero the elevation of the bit? I've seen machinists do that before without electronics. Or does that have something to do with the planarity of the board?

Like with plated-thru vias and at least two layers n such. The sheet of FR4 in the vid only has copper on one side, the only thing most of those holes can be for is jumpers.

That is to say the board design in the vid doesn't look like it's a design conducive to automated assembly.

The board top surface and the X-Y motion of the gantry that moves the spindle are not co-planar to sufficient accuracy. You are using an engraving bit that creates a "V" shaped groove and without controlling the Z axis location wrt the board surface the width of the groove varies too much (so you get insufficient cut or the traces are entirely removed). You can't realistically use something like an end-mill because it would be too weak or too large diameter. That program seemed to be taking a very large number of measurements (a 5 x

5mm grid or something like that).

The open-source software uses the gantry with a probe in the spindle to take measurements (as in the video) and then adjusts the G-code file to compensate as a post-processing operation.

P.S. The only thing worse that stabbing yourself with a tiny brittle carbide bit would be breaking one off under your skin. I have not managed that yet.

I think all your footprints and the layout would be different from what you would use with a commercial PCB. So you're creating a new library and doing the layout twice for very little advantage.

If ever I had to make PCBs myself in quantity again (just shoot me) I'd have a silk screen made, print the resist in one go and etch panels with a DIY temperature-controlled spray etcher.

A cheap CNC with carbide tooling does a fine job of drilling the holes, quite fast and no leveling required.

Why would they be different? The footprint only has to do with the part and how you solder it to the board. The routing is only constrained by the separation required because of the routing bit. If you have a layout that works with the router, why would you change that for an etched board?

I'm not sure the leveling is that necessary, but maybe the tool is a finer diameter than others I've seen. 25 years ago I saw such a device with a conical bit used with the board clamped onto the bed of the device. When I spoke to the guy using it he never mentioned any problems with planarity, just breaking the bits.

Or any assembly. It looks fake to me.

Looser tolerances on trace width and no plated-through-holes. The latter means to get any strength you need large annular rings. So huge vias. If you're making quantity you probably want to make the board as small as practical, cuts PCB and assembly costs, so the 0805 parts will be 0603 or 0402 and the chips might be 0.4mm or 0.5mm pitch.

Doesn't take many thousands of an inch error before a thin trace disappears or the tool doesn't fully cut through the copper.

And how many thousands are your boards warped? Like I said, the guy using the machine seemed to have no trouble with it other than frequent broken bits.

I don't think it's fake, but it's a bit like rolling your own vacuum tubes at home. Yeah you can do this, but ????????

There are lots of hints at fakeness.