Great results milling PCBs from Eagle with the Nomad

2c6f464cb14fdcbf

Liyanage over at Carbide 3D Community site writes:

Milling PCBs was one of my main reasons for getting a Nomad. I finally found some time to learn how to use Eagle and the related tools and mill my first boards. I am very happy with the results and once again very impressed by the Nomad, especially its precision.
This post provides some results and lessons learned from my first two test projects.

DetailsĀ atĀ Carbide 3D Community site.

Join the Conversation

15 Comments

    1. Hard to tell exactly, but one thing’s sure – nowhere near as much as that guy did. A DIY-grade CNC mill for $2500 with $150-$200 toolbit sets? Bwahaha, no. Gimme a break. You can get a machine every bit as good as that one from J. Random Chinese Seller on Ebay for anywhere between $500-$1000 depending on shipping, taxes etc., anything else you might need (including toolbits) is pocket change beyond that.

      1. I’ve got some experience with machining and there always a direct correlation between how much you spend and the tolerances you can achieve. If you have any references on cheap chinese hardware that can achieve these kinds of results I’d love to hear about them!

      2. Sorry for the late reply mate, it’s just that I prefer to back up my arguments with fact whenever possible. Direct correlation? No such thing is the hobby CNC market – there’s only ghetto quality and decent quality (which includes most any Chinese unit that can usually be had somewhere between $500-$1000) and lots of impeccable gentlemen eager to fleece idiots for up to an order of magnitude over that, for the exact same quality. Which is not to say the Chinese units are the pinnacle of CNC, but if you want to improve on them you start somewhere between $10.000-$100.000. Now, back to our thing: please see CNC model 3020 (probably the most common Chinese unit) doing a batch of SO8-DIP adapters (I just happened to knock some out over the weekend) [1]. Please note that every single hole is aligned precisely [2] even though they were drilled (obviously) _after_ all 16 adapters were isolation milled, resetting the coordinate system for each one. The picture might suggest some were of lesser quality but that is not so – they were simply not de-burred yet; after cleanup all of them look pretty much like this [3]. I did not bother with any secondary rub-out passes beyond the initial pass, which is why unused copper remains between the pads, but neither that nor my less-than fortunate choices in feeds etc. that caused extra burr are pertinent to the precision of the job, which I’m quite chuffed with. Oh, and that was a 3.9 euro toolbit…

        PS – the single “stroke-out” piece is just a humble offering my machine controller decided to make to Pythagoras – entirely different & long story…

        [1] – http://postimg.org/image/of54kgiyz/
        [2] – http://postimg.org/image/l30t40q0n/
        [3] – http://postimg.org/image/ul7httda5/

      1. Looks like they are still building the units purchased during the kickstarter. Sounds like it could be a while before they even offer new units for sale.

  1. @max
    That stroke out thing is one of the reasons the 3020 is not decent enough out of the box. Control boxes are not reliable and may send the spindle to random locations. You maybe lucky and think that everything is working but eventually the crappy stepper cables will give up, stepper drivers will blow, etc. These problems can be fixed by replacing the cables with shielded ones, changing the stepper drivers and power supply.

    But there’s also mechanical problems. That dc-motor with a ER shank is not really suitable for any precision work due to lacking proper bearings. Their runout is very bad and they deflect like there’s no tomorrow with the slightest bit of force.

    Also, in most of these 30×0 series the Z axis trapezoid/ball screws are not properly fixed. You can move the spindle up and down (the amount of play will change with the machine but I have seen 3-4mm play). It needs a thrust bearing and shaft collar.

    I would agree that Nomad is on the overpriced side but it at least has a decent spindle and the electronics are usable out of the box. It is also enclosed which is quite important for PCB milling unless you rig up a very efficient and properly filtered dust collection on your machine. Fiberglass dust is pretty bad for the lungs. Would I buy it? No but that’s because I am a cheap fella :) But I would not call anyone buying them idiots.

    1. I’m afraid you misunderstand me – the “stroke out” problem is with the _control_ box (the part that is usually handled by a PC – but not in my case), not the _driver_ box (the one that comes with the mill and actually controls the steppers). The mill was simply directed to cut straight across a diagonal by buggy software, and it did just that – not the fault of the 3020 so I’m afraid that argument is entirely invalid. As for precision, I prefer to trust actual practice over academic opinions and as everyone may see that spindle is perfectly adequate for either THT or SMT PCB work as-is. Could the bearings be improved upon? Sure, but in my experience that’s not necessary for more than adequate precision for PCBs – unless of course you intend to tell me I just Photoshopped those pics. As for the dust, no such thing – I usually mill under a thin coat of oil which traps the minute amounts of material removed. Not in the pics above though – so if I fail to reply next week the dust probably did me in. Or not.

      1. I am glad that your stepper drivers that came with your machine are working. I had to replace it in mine (installed an arduino running grbl and grblshield to drive the steppers). Anyway you can actually see lots of folks having problems with the electronics in various forums if you look around.

        That dc motor spindle (which I also currently have and am trying to upgrade away from) may be fine if you are just doing SOICs but I needed finer pitch and I also want to be able to actually mill stuff with the machine (enclosures, etc). Anyway here’s one sorta review of the thing (it’s non academic) showing what’s wrong with it.
        https://www.youtube.com/watch?v=7P5THNNjafU

        Lastly, please don’t take this the wrong way but the crappy spindle is probably the main reason that you need to deburr after milling. The linked post states that they did not do any deburring. You will also need to change your toolbits more frequently due to high runout (aka wobble). It’s good that you go with cheap bits.

      2. Well, I didn’t need any finer pitch SMD yet, but based solely on a back-of-a-napkin calculation and the fact that with 1.27mm pitch I can pass twice between pads and still leave a bit of copper in-between, it turns out I should be able to mill 0.8mm pitch with this same precision (and this is a 0.2mm tip V – I could try thinner). I’d be rather curious to see ANY hobby level mill reliably do anything significantly smaller than that.

        Regarding the spindle, I did measure the runout a while ago and it’s actually under 0.02mm – hardly ideal, sure, but given the tip size mentioned definitely not something I’ll be losing any sleep over (also kinda funny, considering that with 0.001″ the Nomad itself doesn’t claim any better). My suspicion is that the tool shape (mine was a flat V-tip stylus, his looks like a spiral 2-flute V) and other dynamic variables have a lot more to do with the amount of burr than the runout – but that’s impossible to prove until I try one of those and that may take a while, so that’s neither here nor there.

        Finally, the video was quite entertaining but no offense, (with the exception of the brief wobble demo milling thick aluminium at an ungodly feed) that’s exactly what I call “academic” – at the end of the day the only thing that matters is what are you able to achieve with a tool, not what you consider the tool ought to (not) be capable of based on some elaborate analysis of its parts. And by that metric, for the DIP/SOIC level PCBs I work with, this 3020 is nothing short of perfect. I never tried to claim anything more nor anything less.

        Those who think they get something more from the sellers of $2000-$3000 machines are absolutely welcome to waste their money whichever way they like as far as I’m concerned; after all a certain well-known phone manufacturer proved conclusively that there is a staggering amount of money to be made exploiting those who think more money = better quality…

  2. @Max:
    It’s hard to argue when you are still trying to deduce some definitive conclusions just based on your experience. Case in point: I purchased a 3020, returned it because of issues with the trapezoidal screws. Then got a 3040 with ball screws. I had to fix the Z axis, change to control box, replace the cabling, etc. Now I am trying to change the spindle.

    “My suspicion is that the tool shape (mine was a flat V-tip stylus, his looks like a spiral 2-flute V) and other dynamic variables have a lot more to do with the amount of burr …”
    And that is not academic opinion, dynamic variables and all ? :)
    One of those dynamic variables is the deflection of the shaft due to skate bearings fitted on a rubber oring.

    For your claim about $2000-$3000 machines. There are actually quite a few machines in that range that are more than adequate, like Sieg or Taig mini mills. You can cut aluminum and even steel with them.

    1. Strictly speaking, just my experience is all I need for what I set out to show – that 3020s _are_ quite capable of fabricating comparable PCBs with the original article. Analogously to how all it takes to prove wrong any number of people arguing that UFOs don’t exist is one single person presenting a full-size UFO full of little green men, no amount of “I couldn’t make it work” proves anything in the face of a single “well, I could”. Even more strictly speaking, we would need a statistical study of epic proportions and unfeasible scientific rigor (repeatability of circumstances and procedure) to establish whether the majority of 3020s are or are not good enough on average but that’s a bit unlikely to ever happen, so we’ll have to make do with anecdotal evidence and there one success trumps many failures.

      Sorry to hear your experience was worse but it isn’t evidence of anything any more than my inability to get rid of burrs – the next guy might be perfectly able to use my exact equipment and succeed at that too. Or not. Impossible to tell until someone manages to do it – all it takes is one who can. Until then, it really is academic to argue whether it can be done with a 3020, with or without runout (unless you take the exact same controller / driver / machine / tool / feeds / speeds with a better spindle and the burrs go away – go ahead, I’ll wait…) and that is exactly why I saw no point in raising the subject until you did – if and when I (or someone else with a 3020) manage to get burr-free, it will cease being academic and will become worthy of discussion.

      “…$2000-$3000 machines. There are actually quite a few machines in that range that are more than adequate, like Sieg or Taig mini mills.”

      I know. I operate one. Except that 1) I wouldn’t call that the same category both the 3020 and the Nomad are in, 2) ours was upwards of 3000 _euros_ (a decade ago!) including the factory CNC kit for it (I’ve looked at ours again last year – hint: it wasn’t any cheaper) which is exactly why I quoted the specific figures that I did, and 3) there’s clearly no arguing with people who just _know_ that all Chinese stuff is by definition either garbage or a toy or both, and inherently unsuitable for anything useful so I’ll just stop right here. Y’all have a nice life, and remember, whatever happens never let anyone challenge your dearly-held beliefs, not even with evidence – everybody else is by definition clearly wrong.

  3. Author of the original article here. After reading these comments I was curious to see if I could go any smaller and I just updated the article with the results of another test, this time with an LQFP64 package that has 0.5mm pitch. It worked pretty well but I feel that’s about as small as I could go. Go check out the photo if you’re interested.

Leave a comment

Your email address will not be published. Required fields are marked *

Notify me of followup comments via e-mail. You can also subscribe without commenting.