|
Yeah the machines built before tangential milling strategies with high feed rates, low radial engagement and high axial engagement were all about being big and rigid as you got mmr from big wide shallow cuts vs the current approach of fast, deep and light that needs way higher accelerations of the tool relative to the workpiece. It's real interesting how understanding of cutting tool dynamics, computer power for path calculations, and machine design/control dynamics all kinda pushed and pulled on each other to really change a lot of the core principles in how production machining is run these days.
|
#
?
Mar 6, 2024 00:32
|
|
|
|
| # ? Apr 29, 2024 00:44 |
|
|
CarForumPoster posted:EDIT: drat just checked the prices of the 5th Axis ones you mentioned, the DM25 looks pretty fukken pricey at $902 and yep it and similar 5th axis dovetail chucks are $700+ on ebay. 5th Axis stuff is definitely pricey, they're quality products and I've had great experiences with them (on someone else's dime!) But I would point out that the bulk of the price in what you linked is the palletized base/riser, not the dovetail vise itself. You can buy the DM25 by itself for $339, or you can probably replicate it fairly easily. https://5thaxis.com/product/dm25/ 5th Axis also have fantastic educational deals: at an HTEC conference I went to last year they were giving out the drawings for I think the V562X as a project for students to make themselves, and they'd sell schools any parts they couldn't DIY at-cost. Just in case anyone here's an educator!
|
|
#
?
Mar 6, 2024 01:00
|
|
|
The stupid-big laser cutter works! Or at least so far it's proven itself to be a very fast way of cutting paper... Still plenty of quality-of-life improvements to do, for example a mechanism for holding the door open (probably a pair of gas struts).  So far I've only taken a wild guess at power and speed settings to check that it worked, 10% power at 100 mm/sec only marked the paper while 50% and 100mm/sec cut it. The first attempt where the paper was only marked was also a good test because it showed the lines were quite wobbly from the gantry ringing after direction changes. There are some things I can try to fix that but a lot of it I think will just come down to everything being so big, ultimately when cutting more representative materials at much slower speeds I doubt it will be an issue.
|
|
#
?
Mar 9, 2024 04:55
|
|
|
Where did you source the honeycomb mesh?
|
|
#
?
Mar 9, 2024 08:53
|
|
|
Rapulum_Dei posted:Where did you source the honeycomb mesh? A Chinese factory called Foshan City Linwell Honeycomb Hardware Co., Ltd. I found on Alibaba. I bought 6 x 0.5m x 0.5m sheets of honeycomb for $15 USD each, it's 19mm thick. Shipping was pretty significantly expensive (to New Zealand) but it was nicely packed in a small plywood crate. Built a frame for it to sit on out of 25mm square hollow aluminium extrusion. I'm using a 1 horsepower dust extraction fan to suck from a boxed off area under the honeycomb for combined fume extraction and vacuum workholding of large sheets of fabric. It doesn't pull any significant vacuum until you cover the vast majority of the bed area but then it does a very good job at holding curled fabric sheets flat.
|
|
#
?
Mar 9, 2024 09:13
|
|
|
Rapulum_Dei posted:Where did you source the honeycomb mesh? Shouldnt you just make it yourself?! [Cue 4 page derail] What was the gibbis thread had a fight about waterjet fixtures corporate? OSHA?
|
|
#
?
Mar 9, 2024 16:44
|
|
|
I mean, now you mention it, it probably IS possible even with a 3d printed former if you could get a roll of thin enough gauge metal for a reasonable cost…
|
|
#
?
Mar 9, 2024 18:17
|
|
|
Rapulum_Dei posted:I mean, now you mention it, it probably IS possible even with a 3d printed former if you could get a roll of thin enough gauge metal for a reasonable cost… If you had a laser cutter, it'd be easy to slot thin sheet and then just slide the pieces together to make waffle mesh.
|
|
#
?
Mar 9, 2024 18:30
|
|
|
Just buying it is clearly superior I worked at a laser shop for 78 years and we just buy ours because we have work to do
|
|
#
?
Mar 9, 2024 18:32
|
|
|
CarForumPoster posted:Shouldnt you just make it yourself?! [Cue 4 page derail] an entire megathread held up Atlas-style by that one dude who simply would not accept that they are not made by loading in like an inch thick plate, cutting out the holes, and throwing 90% of it away
|
|
#
?
Mar 9, 2024 20:19
|
|
|
babyeatingpsychopath posted:If you had a laser cutter, it'd be easy to slot thin sheet and then just slide the pieces together to make waffle mesh. The trick would be to find strip the same thickness as your band saw blade and then you could make a slotting jig and race through making a grid whatever density you fancied. The MK3 Lowrider I made has a work area just shy of 8x4 and I’ve put a 30w laser on it - but I don’t have a honeycomb table that size. I’ve got a drag knife for it too which has been fun to do big vinyl decals with.
|
|
#
?
Mar 9, 2024 20:28
|
|
|
Meshes get expensive fast at Sendcutsend. Besides the linear cut length, I think each pierce is pretty slow. I remember measuring my fingers with calipers to determine the maximum grid size I could get away with.
|
|
#
?
Mar 9, 2024 21:57
|
|
|
Has anybody built one of those Maslow CNC things made of lumber, bricks and drive chains?
|
|
#
?
Mar 10, 2024 04:34
|
|
|
Some Pinko Commie posted:Has anybody built one of those Maslow CNC things made of lumber, bricks and drive chains? Welp, my poor impulse control is making it very difficult to not try to buy one of these kits after seeing it for the first time just now.
|
|
#
?
Mar 10, 2024 05:24
|
|
|
drat that thing looks so janky, which makes it extremely cool. Also now that I have a functioning laser cutter probably the first thing I'll try to make with it is more parts for itself as is the tradition with 3D printers... Currently the screen for the computer that drives it is mounted on an arm off the side of the enclosure but I need to make a nice tray to hang off the screen for holding the keyboard and mouse. Sounds like the perfect task for a laser cutter and some plywood...
|
|
#
?
Mar 10, 2024 07:13
|
|
|
Some Pinko Commie posted:Has anybody built one of those Maslow CNC things made of lumber, bricks and drive chains? I've got a buddy who has one in his basement, and the only stuff I ever see from him gets made on machines he rents from a makerspace 45 minutes away, if that tells you anything A Wizard of Goatse fucked around with this message at 16:10 on Mar 10, 2024 |
|
#
?
Mar 10, 2024 15:43
|
|
|
If I was to make a 20-minute reality TV show that’s basically Roadkill for CNC machines/machine tools would anyone watch that?
|
|
#
?
Mar 11, 2024 03:19
|
|
|
Yeah, I don’t know what most of that is but I’ll watch 20 minutes of drat near anything to satisfy my craving for machining.
|
|
#
?
Mar 11, 2024 03:45
|
|
|
NewFatMike posted:Yeah, I don’t know what most of that is but I’ll watch 20 minutes of drat near anything to satisfy my craving for machining. I, possibly with a cohost, go find machine tools that I rehab/retrofit and flip for cash. Some asinine challenge makes this or that retrofit a particularly difficult thing like: its gross, it’s super heavy, it’s some niche machine, I paid too much, I have stupid objectives to begin with like adding live tooling to a 1987 okuma 2 axis lathe.
|
|
#
?
Mar 11, 2024 04:27
|
|
|
Yeah that rules please do it
|
|
#
?
Mar 11, 2024 04:38
|
|
|
Please post links when you have em cause I'd def watch that
|
|
#
?
Mar 11, 2024 05:06
|
|
|
NewFatMike posted:Yeah, I don’t know what most of that is but I’ll watch 20 minutes of drat near anything to satisfy my craving for machining. ABOM79 on YT is crack for me, along with JohnnyQ90.
|
|
#
?
Mar 11, 2024 12:30
|
|
|
I don't have the attention span for Abom anymore. Eight videos to finish a vise casting is enough edging that even Noga doesn't make a tool for it. CarForumPoster, do you have a glut of cheap machines locally, are you going to solicit for 'projects', or...? Regardless, yes, I'd watch that.
|
|
#
?
Mar 11, 2024 16:32
|
|
|
Wandering Orange posted:I don't have the attention span for Abom anymore. Eight videos to finish a vise casting is enough edging that even Noga doesn't make a tool for it. I background him unless it's something I'm really interested in, like the old shaper.
|
|
#
?
Mar 11, 2024 18:14
|
|
|
Wandering Orange posted:I don't have the attention span for Abom anymore. Eight videos to finish a vise casting is enough edging that even Noga doesn't make a tool for it. lol same for abom Id do what people wanted to watch centered around automation. My BIL has directed a few reality shows for smaller networks so I’m gonna pick his brain about what works well. Ideas for content are all over the place like: -Flip CNC machines for cash by fixing them up and showing improvements with various test equipment like spindle force gauge, ball bar, etc. sell them on eBay/FB marketplace on the video release date -Add capabilities to a cnc machine like replacing a control and adding additional axis/live tooling. -build a small self contained production line like an automated pizza food truck -make a machine tool do something it wasn’t designed to do like winding BLDC motors with a 3d printer -Make a workbench that’s a Swiss Army knife of tools (combo table saw planar router 3d printer laser cutter) -whatever else viewers want to see Through it all I’d want to build to releasing my own Linux cnc fork based control after a few dozen projects. CarForumPoster fucked around with this message at 20:03 on Mar 11, 2024 |
|
#
?
Mar 11, 2024 18:40
|
|
|
That sounds like something interesting that'd I'd probably watch, but tbh I'd expect you to be at a pretty high level of skill and experience at all of this vs bumbling through it on camera. I enjoy watching technical videos where I'm learning from someone with a lot of experience and skill at the task but a lot of YouTube is 'im figuring it out come figure it out with me!' and I don't find it very compelling. Not trying to slag on your skills as I've got no idea what they're like, just a comment on YouTube content in general. For example, super fast Matt makes great stuff but a lot of his videos recently have been 'i didn't think about this so it's a problem and I'm just gonna bodge a solution' and that gets kinda frustrating to watch after a while.
|
|
#
?
Mar 11, 2024 21:32
|
|
|
meowmeowmeowmeow posted:That sounds like something interesting that'd I'd probably watch, but tbh I'd expect you to be at a pretty high level of skill and experience at all of this vs bumbling through it on camera. I enjoy watching technical videos where I'm learning from someone with a lot of experience and skill at the task but a lot of YouTube is 'im figuring it out come figure it out with me!' and I don't find it very compelling. Ty for feedback, it’ll likely be the latter. Superfastmatt is one of the YouTubers I’m taking style notes on. There’s no way I could do a “project binky” or roadkill rn because I don’t have the expertise/budget/time to make content that way. It’d be nice to have a partner with 10 years as an automation engineer and a few people helping out shooting/editing but ces la vie. That said unlike Matt’s lifted vipers and land speed motorcycle’s cD it’s pretty trivial to measure how your machine tools are doing if you have a ball bar (just bought a Renishaw qc10) and some other basic machine measurement tools. I have a line on a closing down machine shop that’ll set me up with a lot of the metrology stuff I’m missing. Anyway I’ll start with a series on me getting my Fadal 4020 together and see how it goes. I’ve got a tripod and camera set up in my garage. When I get back from vacation and get the machine moved in we’ll see how it goes.
|
|
#
?
Mar 11, 2024 22:27
|
|
|
CarForumPoster posted:Ty for feedback, it’ll likely be the latter. Superfastmatt is one of the YouTubers I’m taking style notes on. There’s no way I could do a “project binky” or roadkill rn because I don’t have the expertise/budget/time to make content that way. It’d be nice to have a partner with 10 years as an automation engineer and a few people helping out shooting/editing but ces la vie. To that point, I watch SuperfastMatt and Aging Wheels and Hand Tool Rescue. SFM is an engineer and not fabricator, and even his bumbling ends up with a pretty decent project. Aging Wheels has gotten on my nerves for stuff that even he admits beforehand might not work, or hasn't researched, or doesn't want to worry about.... which then bites him when yup, it fails. Hand Tool Rescue is great when there's a narrated video except for he doesn't ever do cotter pins right  .So Hand Tool Rescue in the style of This Old Tony with that level of down-to-earth know-what-you're-doingness. Blondihacks for industrial automation, if you will. If you go full Clickspring, I gotta pass, because watching someone churn out absolute calm and effortless perfection just reeks of "did a thousand takes for this."
|
|
#
?
Mar 12, 2024 00:03
|
|
|
It sounds interesting but personally I like watching stuff like edge precision, where the production values are decent but its a guy talking through solving complicated problems he knows a shitload about and executes at a high level. He sometimes gets into random bits about machine tool design and servicing or peculiars of old controls when it applies to the part he's working on in that video or set of videos. Stuff like way or spindle rebuilds would be really interesting if you can talk through what you're doing, but idk how much traction you'll get with a 'this machine has a bad spindle, lets open it up and figure it out video'. But also those kind of videos seems to do well, so maybe general youtube's taste is different from mine. You might have a hard time breaking into the YT market if you don't have something to set yourself apart with though.
|
|
#
?
Mar 12, 2024 00:39
|
|
|
CarForumPoster posted:I, possibly with a cohost, go find machine tools that I rehab/retrofit and flip for cash. Some asinine challenge makes this or that retrofit a particularly difficult thing like: its gross, it’s super heavy, it’s some niche machine, I paid too much, I have stupid objectives to begin with like adding live tooling to a 1987 okuma 2 axis lathe. I would pay you cash money to somehow get a green paint green CRT Okuma Lathe to live tool mill a square boss on round stock. A before and after measurement of the machine's squareness would be interesting. Especially if you're able to do the first set when it's up on wood cribbing and 'level' in the sense that it hasn't fallen over yet. Hell, a before and after of the machine's capabilities would also be interesting. There's a ton of old iron with controls about as powerful as a graphing calculator that a control rip and replace could bring to nearly modern performance. Getting a standard set of demos like the Dogbone demo would also help contextualize and make the before and after performance differences really stand out. Trying to do any kind of adaptive/HEM/Trochoidal milling on an old control can be generally described as 'cancerous', and converting a 1992 Fadal to LinuxCNC and doing a helical threadmilling operation would be neat.
|
|
#
?
Mar 12, 2024 00:55
|
|
|
Roadkill vibes with CNC content is an interesting mix. My favorite things about roadkill were the semi-half-rear end jobs they'd do because they thought (usually correctly) they could get away with it, and the bonkers projects. I think half-assing CNC stuff would mix like oil and water, especially if you're going to resell the machines. Bonkers projects feel like a great fit though. You just have to make sure the machine has a useful niche when the project is over, or it'll just piss people off. Re: production value and content density, for my money RotarySMP has the best balance. His big project has been linuxcnc-ifying an old Shaublin lathe. Marco Reps is also great on that front. He's primarily electronics, but does a little CNC stuff.
|
|
#
?
Mar 12, 2024 01:13
|
|
|
RotarySMP is great, and his content way more resembles what I might be doing at any time, but Robin Renzetti and the guy from edge precision are next level. Very little production but just amazing info.
|
|
#
?
Mar 12, 2024 02:58
|
|
|
Methylethylaldehyde posted:Hell, a before and after of the machine's capabilities would also be interesting. There's a ton of old iron with controls about as powerful as a graphing calculator that a control rip and replace could bring to nearly modern performance. As someone that has done exactly this I just want to point out just how much time and money are involved. I took a 1988 Hitachi-Sekei VK45 mill that was mechanically good and converted it to LinuxCNC. It took about a year of pretty much full time work and I have about $15k in the machine right now. Very cheap for the capability but if I charged my hourly rate for the time involved I could have bought a new 5 axis DMG. Unless you have a pile of the same machine to run an assembly line, there is no way to do a retrofit that you could sell for a profit. There is too much reverse engineering, debugging, and tuning to make it viable unless your time is worth nothing. Some points: I had the wiring diagram and full maintenance manuals, without these don't bother. Beyond levelling the machine, there really is not much mechanically that is reasonably repairable. I replaced the spindle bearings and two axis of ball screw support bearings as these are cheap ($2k for a set ebay specials). If the ballscrews or ways are bad, scrap the machine unless you want to spend $10k+ for each part. The spindle drive was dead and Yaskawa quoted $15k to replace it with something modern. A repair shop quoted $5k with no guarantees. I went with a VFD for $1k but getting spindle orientation to work was a huge pain in the rear end. Old analog drives are easy to retrofit and don't necessarily need to be replaced. More modern machines with proprietary interfaces would need replacing the drives as well. Replacing the servo drives is quite doable but budget $1k/axis minimum. Rigging usually costs more than the purchase price of old iron. LinuxCNC is not actually a good choice for getting the best performance from these old machines. LinuxCNC does not have a jerk (or higher order) limited planner and that is one of the things that makes modern controls so much more capable. Unfortunately, I don't think there are any other viable options at the hobby level right now. If I was doing this commercially (ie planned to sell the machine) I would probably go with a Siemens Sinumerik control. Adaptive milling on old iron: https://www.youtube.com/watch?v=bV75vmiyxKw
|
|
#
?
Mar 12, 2024 04:16
|
|
|
ZincBoy posted:LinuxCNC is not actually a good choice for getting the best performance from these old machines. LinuxCNC does not have a jerk (or higher order) limited planner and that is one of the things that makes modern controls so much more capable. Unfortunately, I don't think there are any other viable options at the hobby level right now. Yea I may end up in the same place you are, but I am fortunate to have at least a few years that I can dick around as a SAHdad. That said, it looks like several attempts at a jerk limited planner for Linux cnc have been made. I don’t know dick about the trajectory planner for Linux cnc but I’d be interested in some real world testing of the current look ahead + trapezoidal path planner. Isn’t as good as Siemens may well meet the TBD requirements.
|
|
#
?
Mar 12, 2024 05:47
|
|
|
drat the extra-big laser cutter I recently finished rips through its intended use-case (cutting large sheets of thin synthetic fabric). I can cut some thinner materials at 180 mm/sec and 80% power and the edges are super clean and 'cauterised' so it doesn't fray. Need to do something about the extraction though, the fan works fine but I'm just running the house outside through a door currently and the wind was blowing all of the smell straight back inside again. Had one weird issue with Lightburn and the GRBL controller tonight though, I had already cut two full sheets of fabric and on my third project it started cutting the first shape and then immediately freaked out with "G-code motion target exceeds machine travel" despite the project being entirely within the machine boundary, and I was able to 'frame' all of the pieces in lightburn to show that the laser could physically reach all of the extents of the project. I ended up resetting the controller and disabling soft limits to get it working, but that seems like a pretty lovely 'fix'.
|
|
#
?
Mar 12, 2024 08:54
|
|
|
Blackhawk posted:drat the extra-big laser cutter I recently finished rips through its intended use-case (cutting large sheets of thin synthetic fabric). I can cut some thinner materials at 180 mm/sec and 80% power and the edges are super clean and 'cauterised' so it doesn't fray. Need to do something about the extraction though, the fan works fine but I'm just running the house outside through a door currently and the wind was blowing all of the smell straight back inside again. Any chance you accidentally entered an extra number in the gcode somewhere?
|
|
#
?
Mar 12, 2024 15:03
|
|
|
CarForumPoster posted:Any chance you accidentally entered an extra number in the gcode somewhere? This was g-code autogenerated by lightburn so there shouldn't have been any errors. Also I saved the g-code and then re-imported it into lightburn and the path looked correct with no moves outside of the working area. My current suspicion (will have to check the g-code later) is that for some reason lightburn was commanding z-axis moves, my flavour of GRBL has the z-axis enabled despite there not being one physically present and it had soft limits, so if there were z-axis moves commanded it's possible that the controller would hit a completely virtual soft limit that wouldn't happen when tracing or viewing the pattern in the X-Y plane. I'm pretty sure you can disable the Z-axis in lightburn which I'd imagine would stop it trying to command moves if that was the problem, then I could re-enable soft limits. Still not sure why it would have worked fine for the first two jobs I ran on it though if that were the problem.
|
|
#
?
Mar 12, 2024 18:56
|
|
|
Blackhawk posted:This was g-code autogenerated by lightburn so there shouldn't have been any errors. Also I saved the g-code and then re-imported it into lightburn and the path looked correct with no moves outside of the working area. Let us know what you find.
|
|
#
?
Mar 13, 2024 00:56
|
|
|
CarForumPoster posted:Let us know what you find. Well can't really see a reason for it, there's nothing in the g-code I can see that would command a Z axis move and Z axis was disabled in Lightburn anyway, all I can think of is cosmic ray strike lol. GRBL did have soft limits active with an allowable Z-axis travel of 0mm, so if somehow the Z axis was told to move even by a tiny fraction it would trigger an error. I've re-enabled soft limits but set the Z-axis travel to 9999mm and will see how it goes, I've run it a few times since and it's worked fine. Speaking of which, here's a short video of it cutting some synthetic fabric today: https://www.youtube.com/watch?v=DrK8JBCADN8 The fabric is Challenge Sailcloth EPLX200 fabric in 'snow white' colour, which was quite a challenge for the blue diode laser as the material is significantly transparent to visible wavelengths. I was still able to cut it at 80mm/sec and 80% power while I was able to cut a thicker black coloured version of the same material at 150mm/sec. Edit: The red tint is the laser-safe window I have, the actual laser is blue (450nm). Blackhawk fucked around with this message at 04:05 on Mar 14, 2024 |
|
#
?
Mar 14, 2024 03:58
|
|
|
|
| # ? Apr 29, 2024 00:44 |
|
|
Blackhawk posted:Well can't really see a reason for it, there's nothing in the g-code I can see that would command a Z axis move and Z axis was disabled in Lightburn anyway, all I can think of is cosmic ray strike lol. GRBL did have soft limits active with an allowable Z-axis travel of 0mm, so if somehow the Z axis was told to move even by a tiny fraction it would trigger an error. I've re-enabled soft limits but set the Z-axis travel to 9999mm and will see how it goes, I've run it a few times since and it's worked fine. Relatedly to the Linux CNC path planning discussion, I wonder if there was some interpolation based on look ahead that caused the machine to calculate a trajectory point outside the soft limits. Also broaching the subject of jerk, a rather talky podcast video about scribing with a VMC spindle to make watch faces, tiny grooves, etc. Neat discussion if you're into "what can't most VMCs do well with rotating tools?" https://www.youtube.com/watch?v=MR3X5MkVSzs CarForumPoster fucked around with this message at 13:25 on Mar 14, 2024 |
|
#
?
Mar 14, 2024 13:14
|
|