Greetings from GG's Designer

Howdy y'all.  I'm GG's designer, and am excited to see the GGFlex project enter the niche DIY CNC firearms manufacturing space.  I'd like to dispel some information presented on your github page:

1: Everything related to Ghost Gunner is open source.  If you're not able to find a document/file/etc, it's most likely because I'm lazy and didn't put it online (another possibility: I'm embarrassed by the specific implementation).  Send me a message and I'll make sure you have access to it.  The best way to contact me is to send me an email: gg dot elt at xoxy dot net.

2: GG is designed using the metric system... only the AR15.dd file itself uses imperial (out of respect to its designers); everything else is metric.  I despise the Imperial system.

3a: DDcut only identifies Arduino 328p modules with 16U2... this is due to an intermittent data transmission bug with the CH340G that can corrupt the incoming serial stream at higher baud rates.  Modern g-code can consume data considerably faster than slower serial buses can source it... the 16U2 itself has the same issue as the baud rate is further increased, but it's the best off-the-shelf solution.  In the future we're going to switch over to our own 16U2 firmware, at which point the bug will go away.

3b: If desired, I can add a hook into DDcut's config.txt file that would allow DDcut to attempt to connect to any serial port it finds.  At present we only query 328p modules with the 16U2... because that's what every single GG has inside of it.

4a: You are correct that the GG1/GG2 PCBs (i.e. GrBLDCv1 & GrBLDCv2) are "somewhat outdated".  The latest PCB (i.e. GrBLDCv3) is substantially better, being that it has a dedicated onboard microcontroller (AtMEGA32M1) that controls the spindle.  This allows us to close the loop on the controller, and basically control anything we want.  We haven't "flipped the switch" yet to actually close the loop (i.e. firmware update), but will do so when we release the AK.dd file... for aluminum the closed loop aspect isn't nearly as important as it is for steel.  

4b:  I don't see any reason why GGflex couldn't just use GrBLDCv3 PCB... the only resource it lacks is a stepper driver for the 4th axis, but honestly I don't see a huge reason for 4th axis integration in such a small machine.  As with all other things GG-related, the entire GrBLDCv3 PCB is open source.

5: Y'all definitely want to cover those ballscrews... aluminum is going to wreck them.

FYI: I have some concerns about your statement "The FLEX frame and motion components are rigid enough to machine mild steel in light passes."  I base this on the fact that the original GG prototype used aluminum extrusions and it could just barely handle aluminum.  The harmonics were atrocious.  I think milling steel would not be fun on GGflex.  GG3 is incredibly rigid, using in-house-machined parts, and even still it required a substantial engineering effort to make it handle steel.

Best of luck with your project.  Again, if you need any files, have questions, etc, send me an email... it's unlikely I'll actively monitor this forum beyond this single post.

I'm guessing you're still at DD then? Are there plans for more machines past the GG3 now that it can do steel?

Yes, still at DD (technically I'm a consultant, although I was previously a full time employee). There's always room for improvement; we'll keep releasing new hardware just as fast as we can dream it up and make it.

GGdesigner wrote:

Howdy y'all.  I'm GG's designer, and am excited to see the GGFlex project enter the niche DIY CNC firearms manufacturing space.  I'd like to dispel some information presented on your github page:

1: Everything related to Ghost Gunner is open source.  If you're not able to find a document/file/etc, it's most likely because I'm lazy and didn't put it online (another possibility: I'm embarrassed by the specific implementation).  Send me a message and I'll make sure you have access to it.  The best way to contact me is to send me an email: gg dot elt at xoxy dot net.

2: GG is designed using the metric system... only the AR15.dd file itself uses imperial (out of respect to its designers); everything else is metric.  I despise the Imperial system.

3a: DDcut only identifies Arduino 328p modules with 16U2... this is due to an intermittent data transmission bug with the CH340G that can corrupt the incoming serial stream at higher baud rates.  Modern g-code can consume data considerably faster than slower serial buses can source it... the 16U2 itself has the same issue as the baud rate is further increased, but it's the best off-the-shelf solution.  In the future we're going to switch over to our own 16U2 firmware, at which point the bug will go away.

3b: If desired, I can add a hook into DDcut's config.txt file that would allow DDcut to attempt to connect to any serial port it finds.  At present we only query 328p modules with the 16U2... because that's what every single GG has inside of it.

4a: You are correct that the GG1/GG2 PCBs (i.e. GrBLDCv1 & GrBLDCv2) are "somewhat outdated".  The latest PCB (i.e. GrBLDCv3) is substantially better, being that it has a dedicated onboard microcontroller (AtMEGA32M1) that controls the spindle.  This allows us to close the loop on the controller, and basically control anything we want.  We haven't "flipped the switch" yet to actually close the loop (i.e. firmware update), but will do so when we release the AK.dd file... for aluminum the closed loop aspect isn't nearly as important as it is for steel.  

4b:  I don't see any reason why GGflex couldn't just use GrBLDCv3 PCB... the only resource it lacks is a stepper driver for the 4th axis, but honestly I don't see a huge reason for 4th axis integration in such a small machine.  As with all other things GG-related, the entire GrBLDCv3 PCB is open source.

5: Y'all definitely want to cover those ballscrews... aluminum is going to wreck them.

FYI: I have some concerns about your statement "The FLEX frame and motion components are rigid enough to machine mild steel in light passes."  I base this on the fact that the original GG prototype used aluminum extrusions and it could just barely handle aluminum.  The harmonics were atrocious.  I think milling steel would not be fun on GGflex.  GG3 is incredibly rigid, using in-house-machined parts, and even still it required a substantial engineering effort to make it handle steel.

Best of luck with your project.  Again, if you need any files, have questions, etc, send me an email... it's unlikely I'll actively monitor this forum beyond this single post.

Awesome, thanks for reaching out! The lead dev has been AWOL for the last two weeks so things are a bit weird in the development sphere (with only two of us still active at the moment) but I'm sure we'd all love any input you have on the project. I'm mainly the hardware guy so I can't speak much to the firmware and coding choices but;

1. We haven't been able to find source for the firmware or DDCut anywhere, which has been the main reason for some of the hardware choices we've made for our next phase of development. The precompiled DD2V7 hex works fine within the compatibility constraints we've set for this version, so this isn't a huge problem. I'll definitely get in contact by email if you're able/willing to share them though!

2. It's been a while since we converted everything over, but from memory a lot of the specialized parts like the YZ plate and ballscrew idler/shaft end mounts looked like inch dimensional parts that were converted to metric and then adjusted by metric dimensions to fit the parts. This may have just been a coincidence, but a lot of the values looked like they were derived this way; regardless the CAD files were all metric native so it wasn't complicated to adjust. We kept the distance between centers and ballscrews/shafts the same, but a lot of dimensions were rounded to multiples of 5mm or just nearest mm values to simplify sourcing and modifying for any of the parametric values.

3a. Our coder had mentioned that the CH340G's stability and clock/baudrate was likely the reason, good to know for sure. We are steering away from PC hosts entirely in favor of SD hosts for our most basic hardware setup, but wanted to maintain compatibility with DDCut if possible. This basic option will probably just be RAMPs hardware with an SDramps integrated card controller (likely one of the ubiquitous LCD/SD expansions). Mounting, sending, and running code from PC to SD would be a good project and would fix bandwidth, clock, and lookahead jams that usually plague PC hosts while still offering the accessibility of the PC client.

3b. The 328P is half the problem for the above PC hosting choice; newer hardware built on ARM chips have native USB or allocated SPI or UART (plus native ethernet for some) and some are getting to the point where they are almost as cost efficient as 328P boards. But if we were to build around a specific serial device to maximize backwards compatibility with 328P/2560 boards it would probably be the FT232, since it can be piggybacked onto most RAMPS and GRBL hardware... but this still might not be stable enough. I'll ask the more programming savvy guys what they think; thanks!

4a. Good to know; the two guys who have 'low-RPM' spindle builds going are using offboard BLDC-8015 drivers. There's been some work integrating their alarm functions, but nothing concrete yet - and certainly nothing approaching full closed loop control.

4b. I'm finalizing the 4th axis work within the next week; we're scaling the machine up to a 350x150x70mm work envelope with a 50mm swing on the rotary axis (centered along X and Z) for our 'basic' 4-axis machine build. It (combined with the low-RPM development and probing) is one of the major features we're developing to facilitate producing more complex parts. We've got some tubular, hardware-store sourced CRS parts and a single-process/setup AR lower on the immediate horizon, with more advanced stuff down the road. The 4th axis itself is a geared design made mostly from printed parts and around $30 in bearings; the current test unit has an insane 1:96 reduction and fails (by shear and slip) at around 18lbft. I've got another test unit of an older design that I've pushed through 3000 full rotation cycles that has about 10 minutes of backlash (compared to about 2 minutes in fresh condition). So overall, seems very capable and strong. I'll add a plug and support section for the GRBLDC3 for the standard FLEX model (plus low-RPM install) once we can get one of the devs a unit, but the longer term development definitely requires a 2560-based board at a minimum (and probably an ARM board for any more than one or two of the advanced features).

5. Yah, we have separate dustcovers that can be added to the X table mounts without much hassle. We decided to keep them removable to keep the open build area for weird mountings of larger parts, but we'll probably have a variant of the X table parts down the road that have the covers built in.

(6). The hardware store sourced CRS parts mentioned above are 60 and 80 mil 1000 series by the look of it. 1/8" or less diameter 3 or 4 flute carbide mills with 20% width of cut and 0.5deg stepdown max (no plunges) for very shallow contouring and cutouts. Slow and screeches like a banshee but no tool or machine deformation and not much tool wear to speak of. Hopefully once Insurgent returns I can grab some videos and quality pics off of him...

Any feedback and questions are always welcome; we try to keep as much of the technical discussion and Q+A on these forums as possible, but you can also always contact me via PM or email. Thanks again for reaching out!

EEM-LLC wrote:

4b. I'm finalizing the 4th axis work within the next week; we're scaling the machine up to a 350x150x70mm work envelope with a 50mm swing on the rotary axis (centered along X and Z) for our 'basic' 4-axis machine build. It (combined with the low-RPM development and probing) is one of the major features we're developing to facilitate producing more complex parts. We've got some tubular, hardware-store sourced CRS parts and a single-process/setup AR lower on the immediate horizon, with more advanced stuff down the road. The 4th axis itself is a geared design made mostly from printed parts and around $30 in bearings; the current test unit has an insane 1:96 reduction and fails (by shear and slip) at around 18lbft. I've got another test unit of an older design that I've pushed through 3000 full rotation cycles that has about 10 minutes of backlash (compared to about 2 minutes in fresh condition). So overall, seems very capable and strong. I'll add a plug and support section for the GRBLDC3 for the standard FLEX model (plus low-RPM install) once we can get one of the devs a unit, but the longer term development definitely requires a 2560-based board at a minimum (and probably an ARM board for any more than one or two of the advanced features).

If I ordered a kit on ebay would the scaled up change be in it?

gauntlet-eternal wrote:

EEM-LLC wrote:

4b. I'm finalizing the 4th axis work within the next week; we're scaling the machine up to a 350x150x70mm work envelope with a 50mm swing on the rotary axis (centered along X and Z) for our 'basic' 4-axis machine build. It (combined with the low-RPM development and probing) is one of the major features we're developing to facilitate producing more complex parts. We've got some tubular, hardware-store sourced CRS parts and a single-process/setup AR lower on the immediate horizon, with more advanced stuff down the road. The 4th axis itself is a geared design made mostly from printed parts and around $30 in bearings; the current test unit has an insane 1:96 reduction and fails (by shear and slip) at around 18lbft. I've got another test unit of an older design that I've pushed through 3000 full rotation cycles that has about 10 minutes of backlash (compared to about 2 minutes in fresh condition). So overall, seems very capable and strong. I'll add a plug and support section for the GRBLDC3 for the standard FLEX model (plus low-RPM install) once we can get one of the devs a unit, but the longer term development definitely requires a 2560-based board at a minimum (and probably an ARM board for any more than one or two of the advanced features).

If I ordered a kit on ebay would the scaled up change be in it?

Not for the currently available listing. The 4th axis build uses most of the same parts (requiring a couple each swapped ballscrews, extrusion lengths, and linear shafts) so I'll hopefully have both an 'Upgrade kit' and complete 4th axis capable kit listed around the time that the 4th axis is finalized. But the 4th axis build is not intended to replace the basic FLEX unit, but rather to be a modular upgrade providing additional functionality.

What's the benefit of the 4th axis for completing lowers?

gauntlet-eternal wrote:

What's the benefit of the 4th axis for completing lowers?

We plan on using it to make a single-process milling file that moves between pocketing and pin position operations without manually adjusting the part or jig (but still requiring one tool change in between), but that's about it for lower completion. If we ever try a 20%/raw forging build, the 4th axis would definitely be a big plus too. Not so much for 80%'s.

We're working to use it in the tube steel feature milling mentioned above and there's talk of a few other projects in steel and aluminum ranging from handguards to monocore suppressor bodies to other more specialized receivers and frames... all this hinges on the final capabilities though, so nothing definite yet.

In the mean time it's more intended for the wider community of CNC nerds. Widely available, accessible, and free-to-use simultaneous 4-axis CAM is a relatively new thing, there's all sorts of interesting projects people have just recently started building around that tech within the Open Source community.

I haven't found any 4th axis info on githib, is there a link available??

Swhowardno1 wrote:

I haven't found any 4th axis info on githib, is there a link available??

The 4th Axis, BLDC, and inductive/IR probing support are included in the 2.1 release anticipated within a couple weeks (along with a bunch of quality of life/alternate configuration printed parts and more electronics support/firmware preloads). Due to the fork/push settings on the Git, I'll be forking it personally instead of making another release of the /insurgentindustry main. Should have an announcement up on here when it's rolled out.

1. So i could use the GrBLDCv3 instead of the Arduino R3 with Atmega16U2? This looks like the right stuff but not sure:   github.com/Defdist/GrBLDC3/

2. Do those dust covers exist and if so, where can I find them?

1. Yes; I believe that's also what's listed for sale at https://ghostgunner.net/product/gg2-control-board/ (GG also has the Arduino R3 at https://ghostgunner.net/product/arduino-board-for-gg2/ since that's what earlier gen GG2's ran on). I haven't seen any other suppliers besides DD, so I think you'd have to get it directly from them. The GRBLDC is built around the same chip and functionality so it isn't necessarily an 'upgrade'... it's a more compact package and the onboard BLDC supply is nice though.

2. There was a set on the 'iceybreakz' fork but it was pulled down because they were sliding, sheath-style covers and they kept getting jammed up on one another. We're experimenting with a couple other designs now as well as trying to source 15mm accordion bellows to fit directly over the screws. A lot of test parts are included with the forthcoming 2.2 parts and I'll try to tack on the dustcover versions that are showing promise if anyone is interested in testing.

1. I am looking to get some pcb's made using their open source files. Boards would cost only $5-$10 ea. it seems. GG officials are like $60.

2. When you say testing, I assume cutting 80 lowers. Unfortunately, finding these things, esp under $100, is quite difficult these days with all the demand, so I can test, but not sure how many of these $100 lowers I would need to cut to make testing worthwhile.

The gg3 states will do ak lowers, I assume that’s the flats and not milled?