[jlawton]

Craig,

If you used "released FlatCAM" that was probably version 8.5 and yes that was NOT impressive. If you used the beta that was FlatCAM 8.994 and I've found a few small bugs but overall it's pretty solid. The primary problem to me is the documentation is ancient and all but useless.

To me there seems to be some inconsistency between what FlatCAM is programming and what is getting machined. When I look at the motor it appears to have ball bearings and I don't really see or hear anything that suggests excessive runout but I can't be certain. This is why I got into the discussion about PreciseBits but those people seem to think their tools are so precise even RUNNING them on my spindle would destroy them!! Would it be a reasonable "hack" to just keep opening up what I'm TELLING FlatCAM the tool width is until I get some sort
of correlation with the trace width I was asking for? I understand Fusion 360 isn't terribly expensive but I've already studied this system, to just drop everything and start chasing down a
totally new rabbit hole makes no sense. I could consider Autoleveller maybe a little after I get past this initial stage, what does it require just Java 8? I just want to start getting some USABLE boards out of this thing, "perfection" can come later, but I'm listening for any reasonable ideas.

Jeff

[joeavaerage]

Hi,

If you used "released FlatCAM" that was probably version 8.5 and yes that was NOT impressive.

I last tried FlatCAM about 2.5 years ago, what version it was I don't know.

I had for some years prior to that experiment I had been using EAGLE. When Eagle was bought by Autodesk and incorporated into Fusion I sought an alternative. None
of the free or open source tools, including FlatCam were of use to me.

I used Fusion, or rather the electronics design modules within Fusion, which is familiar to any whom have used EAGLE.

In the event I subscribed to Fusion ($675NZD/year) and I use it all day every day. I thought initially, as you do, that Fusion is expensive, but when you consider all the features,
the quality of those features then you realise that Fusion is very VERY competitive amongst the Pro level CAD/CAM packages out there. I am now of the opinion that Fusion represents
good value for money. Like any system there are things which I would like done differently but over-all I'm very satisfied.

Despite Fusion being able to generate Gcode toolpaths for isolation routing, its is not as good as the very familiar (to me at least) as the ULP PCB-to-Gcode in EAGLE.
Thus I have a copy of EAGLE into which I pass the Fusion design, and then use PCB-to-Gcode to generate the Gcode isolation routing files. Best of both worlds.

I have attached a pic of one of several boards I have designed this week. Not only designed, but made the board, populated, tested and programmed the uP and sold
it to a customer, money is already in the bank. I also attached a few lines of the Gcode generated by the ULP for this board.

Craig

[jlawton]

This whole exercise has been SUCH a waste, SUCH a lack of understanding, it couldn't POSSIBLY have gotten further off-track...please read on...

I came into this with a SMATTERING of mechanical knowledge, but not bad for an EE, my first serious jobs in electronics were in the field of computer peripherals,
so I was not at all unfamiliar literally dealing with microinches...but that was way back in the early 80's...but I had already been laying out PCBs and such...
I was an eager early adopter of CircuitMaker/TraxMaker when it first came out WAY before they were bought by Altium, it had a "Manhattan autorouter"
which was pathetic compared to the rip-and-reroute I use today, but FOR ITS TIME it was advanced, it even had a SPICE simulator...I since "moved on" to
KiCAD and am very impressed...

This was my first mill, my first CNC, I was very eager to get started, I bought what I thought were these super-fancy blue-coat engraver bits, I'm not sure
but I think they were labeled "SainSmart" which is a known brand. It appeared that I had THREE 10-count boxes of bits but only one box of "blue-coat", but
I didn't want to make any mistakes, it LOOKED TO ME like "tip width" is something you can't really even accurately measure, although I use my dial
caliper to look at dimensions when I can...I "didn't want to take chances" so when it came time to do engraving I loaded one of the "blue-coats" into
the mill, and loaded the file...and the result was horrifying...

I just took that bit out of its case. If you shine the light just the right way and you look VERY VERY CLOSELY at the tip, it isn't a point at all! In fact there's
TWO points with a split between them!! The whole damn BOX is like that! On the other hand the two boxes of "non-blue" bits ARE true points! Now the weird
thing is I have NO IDEA what that bit design is for, or how it got like that! I suppose this COULD just be the result of a mistake by a shipping clerk!

I may owe some of you an apology...but somewhere SOMEBODY owes ME an apology...if I even had any friends I suppose this could fall into the realm of a
"practical joke" but it's MUCH more elaborate than most people would ever dream to pull off...what the heck is a weird tip like that for anyway??

[CitizenOfDreams]

Any way you can show us a macro photo of the "split tip" in question?

[jlawton]

Not exactly here and now...I never took up photography so no fancy cameras or lenses (there's a reason for that having to do with massive early trauma but hey that's TMI)...I did "kind of try" but through my ancient iPhone anything that close is all hopelessly out of focus...maybe I can find someone with some appropriate gear to help out over the weekend...I'd sure like to find a way to do that and post pix and get some closure...right now I'm going to do the "systems engineering" to finish the schematics and laying out of the "larger" boards, in my mind it seems for now I likely have identified a "root cause" but yes right now that's just a tentative finding...but at least for now I do want to thank everyone who posted, sorry for the way I "came across"...I do like the "idea" of Autoleveller but it would be nice if that "function" were inside the machine firmware so not all this additional error-prone file handling, but I guess to do that would have to violate RS-274X (gcode) so maybe can't be done?

[deadlykitten]

hy guys, is there something i can help you with ? i can create custom 2d cad cam

[joeavaerage]

Hi,

I do like the "idea" of Autoleveller but it would be nice if that "function" were inside the machine firmware so not all this additional error-prone file handling,

There is a means to do that. Mach4 has a module called mc.Surface. It probes a nominally flat bed, or other flat object like a PCB blank, and then the Z axis varies its height
based on its X,Y location. It does this without any file manipulation whatever. If you like the Z axis correction is 'baked in'.

I believe this feature was more designed to 'flatten' the bed of a machine, that is to say that the bed would be probed and for many months or even years thereafter that
correction is applied to every job. For this reason it it not a particularly convenient tool for probing PCB blanks, as each and every one of them must be probed,
and consequently I don't use it for PCB making, but rather use Autoleveller. I've used Autoleveller for seven years and have good procedures that produce good
and consistent results.

It rather sounds like those faulty tools have cost you some grief. When I first started making PCBs I tried various cheap Chinese made tools. I've never had such a significant
fault as you report but none-the-less lost confidence in cheap Chinese tools. That's when I started buying Kyocera-Tycom tools such as I pictured. They specialise in micro-tools
for the PCB industry. Worth every cent. Many have reported good results with PreciseBits products, and can only but presume they too are quality made to have garnered such a good
reputation.

Measuring the tool tip is a near impossibility, all that can really be said is that a tool is sharp OR the point is broken. I tried 15⁰, then 30⁰ tools but the tip
is just too fragile that in effect you do not know with any confidence the tool tip diameter or the tool length at any given time because you have no way of knowing whether
the tip has broken or not yet. Note also that these ultra fine tips place an extreme premium on run-out, any run-out and they fracture within milliseconds of use.

I now use 60⁰ engraving tools and they last superbly. I will often throw them out after a month or six weeks use, and perhaps a hundred boards as being blunt,
not broken, just blunt as the cut quality has degraded. Note that these larger included angle tools place a premium on cut depth, naturally any small increase in cut depth
can materially reduce trace width. For this reason Autoleveller is an essential part of my work flow.

I made a series of boards that had an ultra thick copper layer 12oz or 420um (0.42mm), for a high current project I was working on. Vee shaped engraving bits are of no use
for tis board. I ended up using 0.4mm and 0.5mm two flute endmills (Kyocera-Tycom naturally), and even the it took many frustrating attempts before I discovered all the
required techniques to do it. I must have broken 50 or more tools, but eventually got to the situation where I would remove a tool after four hours use as blunt, not broken.
The two innovations that were the genuine break-through moments were Autoleveller and flood cooling. Autoleveller meant precise depth control and therfore very limited
abrasive fibreglass dust in the chip stream which markedly improves tool life. Flood cooling was less about cooling but rather flushing the chips out of the cutzone, and it has
a night-and-day effect on tool life.

The point of this story is that I stuck with it, and learned what was required to achieve the result I wanted. The boards worked out perfectly and I had to learn about chip formation,
chip evacuation, heat removal, spindle run-out and depth control, Gcode generation an Autleveller software Gcode manipulation. Very satisfying. What ever you do
'DO NOT GIVE UP', the results will come.

Craig

[jlawton]

OK so I got in contact with SainSmart (didn't send the product back) and asked them about the split tips, and they said from my description it COULD have been defective product, as far as they were concerned it wasn't the WRONG product because nothing they make has tips like I described. Well I've been extremely busy and I've changed some of the "nonessential" software tools radically, went from Windows 10 to Linux MATE Mint, and now my control SW is Universal G-code Sender since it seems extremely solid, but just getting THAT working was a major PITA. So anyway since I'm not using the SainSmart tips I had to look through my collection of tips and I had some other 30 degree ones that came with the kit but that I still consider to be of of "questionable" pedigree. Now to refresh this whole issue a little I'm generating Gerbers in KiCAD, making Gcode from those in FlatCAM, and a lot of what I'm doing (certainly the drilling, and apparently the milling) seem to be OK, but I CAN'T believe the trouble I'm having with engraving! I just "fired up" my machine, I put the "questionable" (but still BRAND NEW) engraving bit in the machine, and now I have the opposite issue, even with the spindle cranking at 15KRPM and all, the tip not only isn't engraving, when I shut off the machine it's barely making a visible SCRATCH, and this is on the SAME Gcode file the other tip was way too wide on!! I guess the tip is too soft or too blunt, dunno...

I mean maybe I don't need the fancy "PreciseBits" type of product everyone wanted to talk me out of, but I definitely need to find a source of engraving bits WITH CONSISTENT PERFORMANCE (so I guess forget Amazon and AliExpress). With the bits I've found so far, the results are so pitifully inconsistent that if I couldn't see the excellent photographs here, I'd probably conclude it couldn't be done! Is this type of product available for sale inexpensively anywhere I could drive to (Southern California)? I'm only asking because if I order another "pig in a poke" then I wait at least a week for it to ship out, then when it doesn't work I have to fight to get an RMA to return it because the vendor is too proud to admit he sells crap, then I've lost 10 days and have to start over...not an appetizing scenario...or at least post a link to something that works for you...or does anyone think I ought to go back to the SainSmarts and try and figure out what the "effective width" of those weird bits works out to (would prefer not to but I need to find a fix among a bunch of unattractive options)...

[jlawton]

Does anyone even have some favorite "moderate priced" brands of carbide bits for PCB use that worked for them? BQLZR, Bantam, Onsrud, Amana, wegstr, others? It occurs that the bits that shipped with the kit were really for wood and plastic, not stiff at all.

[joeavaerage]

Hi,

Does anyone even have some favorite "moderate priced" brands of carbide bits for PCB use that worked for them?

Yes, I've bought dozens of these:

https://www.ebay.com/itm/381533819973

Kyocera Tycom is a good brand, $4.00 each is a fair price. The company is in Texas. My last order from them amounted to $245USD and it cost $12USD for shipping to Oregon, my shipping agent,
and it got there in four days. Another $30USD plus New Zealand GST (15% tax), and I had them delivered to my door in another 7 days.


30 degree engraving bits are too tender, you are making it hard for yourself, use 45 degree or 60 degree, you'll get better results.

The real issue you are facing is depth control, or rather the lack of it. There is ENEVITABLY a certain variation across a PCB blank and if you don't account for it
you will find that certain areas of the board are scarcely touched and other ares where its too deep.

Use Autoleveller or an equivalent. I have made if not hundreds, then thousands of boards over seven years, none of which would have been successful WITHOUT Autoleveller.

Craig

[joeavaerage]

Hi,
just finished this board, 110mm x110mm. Has 0.31mm Z axis variation over the extent of the board, but Autoleveller took care of that.

Craig

[jlawton]

Well as far as Autoleveller goes my only shop machine is now running Universal G-code Sender on Linux MATE Mint on my old netbook, with 2G RAM (maxed), it turns out my plan to run Win10 on a machine with that little RAM was ABSURD (it was barely stable under Win7). UGS is also written in Java but the Java it needs is bundled with the distribution. Now I guess that there's SOME way to invoke Autoleveller running on JRE and that there's a JRE set up for that (hell for all I know it COULD be part of the Mint distro) but from what I can tell they don't publish installation instructions (they don't even seem to say WHICH version of Java is the minimum it will run on), and I'm no expert on JRE anyway. Now I'm starting to "warm to the concept" of that tool as I start to try and engrave PCBs that need it, since it REALLY only needs to be applied against the two files per design that reflect the isolation routes not all five, but until someone "solves the install issue" for me (I spent the last TWO DAYS just getting UGS running, and THAT install was an order of magnitude easier) the premise has to be dead in the water for now, but in theory it's a nice idea.

[blinkenlight]

@jlawton have you considered using Candle instead of UGS? It doesn't need Java, it does run under Linux, and it has a built-in "autoleveler" (heightmap) function.

[jlawton]

Are you kidding me? I'm using 64 bit Linux. Candle is written in a language called Qt, specifically qt5. Until I saw the Candle website I had never HEARD of Qt! There is a .tar.gz distribution of Candle on github but it only supports 32 bit Linux. In order to install this on a 64 bit Linux (according to Denvi) you need to install the 64 bit version of the compiler, get the source code from github and compile it, download the .tar.gz distribution and expand all the files, copy the 64 bit image over the previous 32 bit version from the distribution, THEN install the program! And yet the pinheads at Denvi seem to think that entitles them to declare that they "support Linux"! I'd rather swim naked in a pool of piranhas than have ANYTHING to do with that program anywhere NEAR modern Linux! Do you even KNOW anyone with 64 bit Linux who got this working? Of course you don't because it's next to impossible! Do you have any other "great ideas" that nobody can use??

And before anyone gets any other "wild hare" ideas, both Chilipeppr and bCNC "claim to" have some kind of auto-height function. Well Chilipeppr runs in a browser but somehow it "demands" a special graphics controller standard that my Intel N Atom-based motherboard just doesn't have, PLUS installing the serial JSON server has proven to be another PITA. And while there's a .tar.gz distribution for bCNC it's in Python so you have to install the Python setup tools and a few other "special modules" that also happen to not be in the standard Mint distro, in fact for bCNC there's a .exe file for Windows and people are saying even installing THAT has been problematic! Here's a warning, don't "take it at face value" if somebody says a particular program is "supported in Linux" on some hobbyist website because that DOESN'T mean you can get it to work on YOUR machine with YOUR hardware and YOUR Linux distro, this is WAY more complex than double-clicking some install.exe under Windows!!

The "gold standard" for Linux install would be to provide a ".deb" (Debian) install module, but of course hardly anyone does...

[jlawton]

I'm sorry, I don't think talking about software is going to solve this, there's something more fundamental going on. This is what's confusing me, I'm getting into this trying to do double-sided right off the bat (I never "jump into the shallow end"). Now when I looked at FlatCAM they said if you do double-sided you want to place two or more "alignment pins" outside the layout, so when you flip the PCB over the front-side engraving aligns exactly over the back-side (and yes you have to "mirror" one of them). Now my "board stock" for these boards is only 10x7 cm so I decided on two alignment pins 85mm apart, and they make for a snug fit. And I'm using two of the "stock" clamps, one just outside of each pin, for holddowns, AND THAT'S IT. Now it occurs to me that the board stock one way is a concave "potato chip" and the other way it's convex (or I suppose it's perfectly reasonable that each way is BOTH depending on dimension or something). Now remember I started this post with tips that were giving me a "too wide cut" but at least I was GETTING a cut, and let's just say I probably "just got lucky to have it cutting at all". I think what MAY be happening now that I swapped tips is with the board stock I have in there now it's "majority convex" and when the machine moves the tip DOWN the board just flexes beneath it and there's no engraving happening. And if that's what's happening then autolevelling won't fix a damn thing!!

So do I loosen the fit of the alignment pins? Am I REQUIRED to have double-sided tape underneath the work area? (I did "in case" buy some but it's kind of thick and flexy in its own right so I was loath to introduce another set of variables, looking at it I THINK I could find something better but I didn't see any "warnings" or advice how to select/buy or whether it was necessary in the first place, just bought whatever I found on Amazon that "looked reasonable" at the time.) I'm starting to feel there's more to this "double-sided holddown requirements" than was explained to me, I'm not ashamed to admit HOW much I "don't know" but I think maybe it could have been explained better because now that I'm actually trying to "get some work output from the machine" there appears to be ZERO room for either misunderstanding, or lack of experience in dealing with the problem!! (I will say I'm looking for results from people who are getting them MUCH more than I am people "taking potshots" though!)

[blinkenlight]

Your enthusiasm concerning Candle is truly a sight to behold; one cannot help but marvel in awe at its shining flame.

Do you even KNOW anyone with 64 bit Linux who got this working?

Yes.

Have you considered attempting to find out if there are any guides for your particular situation...? Maybe something like this...?

Second hit has a step-by step guide:

    sudo apt-get install git g++ qt5-default qttools5-dev-tools libqt5serialport5-dev libqt5opengl5-dev
    git clone https://github.com/Denvi/Candle.git
    cd Candle
    git checkout grbl_1_1
    cd src
    qmake
    make
    ./Candle

That's eight commands, including downloading the source and starting the binary. It took me all of, oh, five minutes tops to follow that on Mint 20 and the result is apparently running just fine. I have no GRBL 1.1 hardware at hand to actually test it with, but to be honest my motivation is somewhat lacking too, can't possibly fathom why that might be.

Toodles & best of luck.

[jlawton]

I'm sorry but you have ZERO knowledge about this issue. Some and potentially ALL of those commands may need to be modified for Linux Mint. For example to install Universal Gcode Sender I experimented FOR TWO DAMN DAYS modifying somebody's "perfect posted instructions" UNTIL I WOUND UP WITH SOMETHING THAT ACTUALLY BROUGHT UGS UP ON MY SCREEN. And THAT was for a completely functional build, NOT something that needed to be compiled from scratch! ISSUE CLOSED!!

ALL of which is completely irrelevant, I'm NOT looking for some stupid autoleveller because such a function is COMPLETELY IRRELEVANT TO THE PROBLEM, but you were too interested
in playing this stupid "gotcha" game THAN READING THE LATEST POST AND ACTUALLY HELPING SOLVE A PROBLEM. Begone troll!!!!!!!

[jlawton]

So current "best SWAG" is I just have had MISERABLE luck purchasing usable tips, and need to go back to that wonderful craps table called Amazon and try some alternate brands. (Look I'm not blaming the viewers here for getting confused, I'll be the first to admit the symptoms don't particularly make sense even to me, and that's frankly awful!) I'm thinking probably what ships for engraving tips with your average "3018 kit" is at best usable for wood and soft plastic, and there is NO "standard marking" to warn the unwary. Nonetheless if anyone has decent recommendations for double-sided tape THAT WORKS FOR THIS "USE CASE", or anything else or other issues having to do with technique that I would find handy to bail me out of this, please weigh in!

[joeavaerage]

Hi,
I do not use clamps of any type, they always pull some warp or bow into the PCB blank, and I mean always. It might only be 0.05mm, but that's 50um,
and standard 1 oz. board has a copper layer of 35um. So the distortion pulled into the blank is something like 150% of the target cut depth.

I use double sided tape. doublesided 'Sellotape' is as good as any, but these days I get 30mm wide doulbesided tape, un branded, from our office supply company.
I did use specialized tape on the ultra thick copper boards because I was using flood cooling, and I needed a supreme tape that did not lose adhesion when subjected
to coolant. I presume you are not using coolant so any reasonable doublesided tape will work. Ditch the clamps.

I use acrylic sheet, either 8mm or 10mm thick as a spoil board. Its fairly cheap, and because it is a spoil board it will need to be replaced occasionally, but more importantly
its pretty flat, reasonably rigid and doubleseided tape adheres to it nicely.

What you may not have noticed is that the EAGLE artwork I posted (#22) had some dimensions listed. the common point of the dimensions is the (0,0) coordinate of the board.
Thus the position of the two pins I use for top-to-bottom alignment are at (0,-5) and (72,-5), where 72mm is the furthest X axis extent of the board. I displace the two pins -5mm
in the Y axis direction so that the pins themselves are outside the PCB boundary and will not therefore get 'engraved'. The advantage is that there is no having to think and plan
where the pins are going or any calculation required to flip the board.

The top side etch Gcode, that I do first, is anchored by the (0,0) coordinate at the lower left. Then when the board is flipped he Gcode is anchored by (0,0), but now at the
lower right. All thats required between top and bottom side etching is to call a g0 x72y0, ie drive to the new work zero and then 'Zero' the X axis only, you can do it with your
eyes closed!

You are getting closer to getting some decent results:
1) get some good quality 45 or 60 degree engraving tools
2) Use either doublesided tape OR vacuum hold own , if you want to get real fancy
3) Use Autoleveller or some equivalent feature for precise depth control.
4) Use the board zero coordinate and the boards furthest X dimension as locations for the alignment pins

You need to have your own solutions or procedures that address each of these points. Some people mill the spoil board flat prior to mounting the PCB blank. That will ensure that
the spoil board is perfectly 'flat' relative to the tool tip but does not accommodate any bow or warp in the board. Thats where vacuum holdown could be a real game changer
as it can suck the board down to a nominally perfect spoil board. This is and would be a perfectly viable solution to point 3), precise depth control.

Craig

[jlawton]

Observation #1: When FlatCAM tells you to "include alignment pins" in the layout, from what you're saying they're encouraging a line of thinking that by itself leads to suboptimal results (like binding and warpage)? Of course I believe you're right, that sort of "misdocumentation" winds up actually being kind of evil...

Observation #2: I never actually "made a technical selection" on the tape, I bought based on price and wound up with this, but I need advice, is this even CLOSE?
https://www.amazon.com/dp/B09K5B9S1W...t_details&th=1
It seems it's kind of thick (1/12") and likely WAY too flexible...
And yet I'm concerned if I found tape that I thought I could TRUST to use without clamps, I'd almost need a jackhammer to help me "flip" it (or I'd pry with a screwdriver and risk breaking it).
Any actual vendors and #s you would recommend that I could get from "the usual suspects" (Michael's, Hobby Lobby, Jo-Ann, Home Depot)? Or at least material and thickness?

Observation #3: The "vacuum holddown" has a certain appeal, can you recommend anything (link or picture, don't yet have the actual idea) for maybe $50 using my shopvac, or is that all wrong too? (Wish I could salvage the electrostatic holddown from an old plotter, if that would be stout enough...)

Observation #4: I have enough time while waiting for the new tips (Wednesday), I printed out the instructions, I can ATTEMPT to get bCNC or Chilipeppr or Candle running (all have
internal levelling).

And thanks for giving actual help!! (It seems to get harder to find every day!)