#linuxcnc Logs
Jan 04 2024
#linuxcnc Calendar
02:32 AM Deejay: moin
04:07 AM Tom_L: morning
04:07 AM Tom_L: 28°F High 43°F chance of snow late today
04:10 AM lcnc-relay: <TurBoss> Good morning
04:38 AM travis_farmer: Morning... Bad back ache this morning... :-(
05:04 AM JT-Cave1: morning
07:43 AM JT-Cave1: Tom_L, https://paste.debian.net/1303037/
07:44 AM JT-Cave1: figured out how to have multiple boards and skip outputs!
07:46 AM JT-Cave1: so now the tab labels are wrong they say joint 0 joint 1 etc and they are really the stepgen or analog outputs so what to name the tabs?
07:48 AM Tom_L: cool
07:48 AM Tom_L: stepgen or analog out ;)
07:48 AM Tom_L: 1 2 3 etc
07:49 AM JT-Cave1: thinking about Drv or Drive
07:49 AM JT-Cave1: that way I don't have to change them in the program
07:49 AM JT-Cave1: the 7i77 is labeled DRV
07:50 AM Tom_L: ok
07:50 AM JT-Cave1: the labels need to be short so the tabs fit in 800x600
07:52 AM JT-Cave1: I can push what I have if you want to look
07:52 AM Tom_L: no time this morning
07:53 AM JT-Cave1: ok
07:53 AM Tom_L: well i got ~20min...
07:54 AM Tom_L: booting linux..
07:55 AM Tom_L: did you push it yet?
08:04 AM JT-Cave1: sec
08:04 AM Tom_L: today is 1/4 btw :)
08:04 AM JT-Cave1: k
08:04 AM JT-Cave1: lol
08:05 AM Tom_L: i need to pull up a 5i25 or something?
08:08 AM Tom_L: any certain firmware?
08:08 AM Tom_L: yeah that looks ok to me
08:08 AM JT-Cave1: I just tested with 5i25 7i76 7i77 then skip some drives all you have to do is select the axis letter
08:08 AM JT-Cave1: then build and only the hal file will build
08:10 AM Tom_L: right i see that
08:10 AM Tom_L: but the * remains on the Build Config at the top
08:11 AM JT-Cave1: the only thing it does it build a bit of the main.hal file everything else is turned off until I get more coding done
08:12 AM Tom_L: ok i'm headin out
08:13 AM JT-Cave1: later
09:53 AM Unterhaus_ is now known as Unterhausen
10:05 AM Unterhausen: I screwed up my design and have to get it cut again
10:05 AM Unterhausen: parts are nice though
10:07 AM roycroft: congratulations
10:07 AM roycroft: i had a glue-up failure yesterday myself
10:08 AM roycroft: i'm not sure what happened - i was using hide glue and it never set completely
10:08 AM roycroft: i had frozen the batch i mixed up the other day, which i've never done before, but folks say it is fine to do that
10:08 AM roycroft: i also did not have the heat on in the shop, but the parts were at room temperature
10:09 AM roycroft: the frozen glue looked like tapioca balls, but it looked fine after heating
10:09 AM roycroft: anyway, i have a sticky gluey mess now, and i'm probably just going to remake the parts instead of trying to clean them up
10:09 AM Unterhausen: I have started forcing myself to throw away titebond bottles after I had a glue failure
10:10 AM Unterhausen: I don't think buying the small bottles is a good solution to buying old glue though, nobody buys those
10:12 AM roycroft: i buy titebond iii by the gallon, which usually lasts me a bit less than a year
10:12 AM Unterhausen: at least hide glue cleans up fairly easily
10:12 AM Unterhausen: and you can glue on top of the residue
10:13 AM Unterhausen: I have never had a successful hot hide glue joint though
10:13 AM Unterhausen: need more practice, I think
10:13 AM JT-Shop_ is now known as JT-Shop
10:14 AM JT-Shop: I know if you store tb3 in an unheated shop it can go bad at low temps
10:15 AM NoSpark: I'm getting so fed up with this stupid PCI Yakasawa interface card
10:17 AM NoSpark: what drive options are there for ~1kw AC servos that is as close to plug and play as possible?
10:17 AM roycroft: it usualy works fine for me
10:17 AM NoSpark: roycroft: ? wdym?
10:17 AM Unterhausen: my failure could have been the surfaces, tbh
10:17 AM roycroft: but this time it is sticky and rubbery
10:18 AM NoSpark: roycroft: soz, nm
10:18 AM roycroft: yeah, i was not talking about your issue, nospark
10:18 AM Unterhausen: Yaskawa seems particularly locked up to me
10:18 AM NoSpark: roycroft: Dammit !!!!! :P
10:18 AM roycroft: i'm gluging some edge banding and did not want to use brads to hold it in place
10:19 AM NoSpark: I have a PCI interface card that works wonderfully with linuxcnc
10:19 AM roycroft: i thought i could just glue one piece at time with hide glue and clamp it lightly for a few minutes
10:19 AM roycroft: next time i'll use titebond and masking tape
10:19 AM NoSpark: roycroft: use a clothes iron
10:19 AM Unterhausen: there is a yaskawa pci interface that works with lcnc?
10:20 AM NoSpark: Unterhausen: yep, really nice card
10:20 AM Unterhausen: surprising
10:20 AM NoSpark: might have one for sale soon :D
10:20 AM Unterhausen: I am tempted to look into the Chinese servo controls
10:20 AM Unterhausen: Some people say they work well
10:21 AM NoSpark: The card works fine, its the PCI part that is causing me massive problems
10:21 AM NoSpark: If it was available in PCIE it would be perfect
10:22 AM NoSpark: Unterhausen: http://yurtaev.com/ymtl2p.html
10:22 AM NoSpark: Unterhausen: http://yurtaev.com/yio.html
10:23 AM Unterhausen: Used dells have pci ports
10:24 AM NoSpark: Unterhausen: I have one on my desk atm
10:24 AM NoSpark: the problem is reliability, and latency
10:25 AM NoSpark: atm I'm using a PCIe to PCI adaptor and a RPI4
10:26 AM NoSpark: but I'm having reliability issues
10:26 AM NoSpark: I'm at the point where I might just go and buy a new computer
10:40 AM Unterhausen: servos don't need stellar latency
10:42 AM NoSpark: I agree, but I'm struggling to find a computer +/-50us jitter
10:42 AM NoSpark: My RPI can do it, but every other computer I've tried has weird latency spikes
10:54 AM roycroft: a mesa ethernet board would be less expensive than a pc
10:55 AM NoSpark: roycroft: I am using mesa for IO already,
10:55 AM NoSpark: the mesa stuff is less important, as it's just for pneumatics and none hard-realtime stuff (limits, homes, probe etc)
10:56 AM NoSpark: *non-hard realtime
10:56 AM NoSpark: I don't really care if the limit switch is 100us behind :/
10:58 AM NoSpark: But I would need to buy 4 new servo drives, and interface cards for the MESA. I left myself room to expand the mesa, so could add a 50pin card to handle the drives, but I'm guessing that would be about USD $1500 at least?
11:06 AM NoSpark: USD $1500 for the MESA card and drives, not the mesa card on it's own
11:45 AM rigid: NoSpark: did you experiment with a headless setup? Running X in the background causes a lot of jitter
11:45 AM rigid: you could try using remote processes in your .nml file, run the GUI on another (non-realtime) host and connect to your pi via network
11:46 AM NoSpark: I'm not exactly sure what you mean by that? I have X running on DISPLAY :0 and most of my testing is done over ssh with X11 forwarding?
11:46 AM rigid: also debugging that PCIe port might help
11:47 AM NoSpark: rigid: I remember reading about that setup years ago, is there any documents that describe how to do headless properly?
11:47 AM rigid: yes. you're still running X on the pi. configure your linuxcnc to use remote processes and stop the X server on the pi completely.
11:48 AM rigid: you can look at tests/linuxcncrsh-tcp/* for a working setup. also there's some horribly old stuff in the wiki
11:48 AM NoSpark: Sorry, you misunderstood me. The Pi works fine, even running X, Its the half dozen PC's I have tested that don't work
11:48 AM rigid: ah, well same for PCs...
11:48 AM NoSpark: the problem is, the PI doesn't support PCI, so I'm currently working my way though that
11:49 AM rigid: yeah, it probably reacts quirky to adapters
11:49 AM NoSpark: I suspect half my problem is with either intergrated graphics, or PCIe nvidia cards causeing latency, so running headless would be ideal
11:49 AM rigid: did you actually run a RT kernel on those other PCs?
11:49 AM NoSpark: ofc?
11:49 AM roycroft: i tried x forwarding so i could use a pi to run lcnc and interface with a laptop
11:49 AM rigid: ok
11:50 AM NoSpark: roycroft: how did it go? any tips?
11:50 AM rigid: X forwarding is sluggish
11:50 AM roycroft: i found that the most reliable way to do is is with something like vnc
11:50 AM roycroft: you want to be sure to run the app and the gui on the pi
11:50 AM roycroft: if your connection drops it's easy to reestablish it with vnc
11:51 AM NoSpark: my understanding of what rigid suggests is that the realtime PC communicasd over NML?
11:51 AM NoSpark: *communicates
11:51 AM rigid: NoSpark: here's a minimal .ini and .nml as a starting point: https://github.com/LinuxCNC/linuxcnc/tree/master/tests/linuxcncrsh-tcp
11:52 AM NoSpark: there is no X11 or any gui running on the host
11:54 AM NoSpark: rigid: how do you configure the frontend with this? I'm only seeing backend config? or is rsh the fronted?
11:55 AM NoSpark: (I thought rsh was the precursor to ssh?)
11:55 AM rigid: yes, rsh is the frontend. There's also configs/common/client.nml and configs/common/server.nml (where client is the host that runs the GUI)
11:57 AM NoSpark: to be clear, the host PC would run linuxcnc, and be connected to the PCI IO card and the mesanet ethrnet card, and then a different PC would just run a non-realtime copy of linuxcnc, with whatever GUI ?
11:58 AM rigid: yes
11:58 AM rigid: you can use DISPLAY=dummy on the host PC with the PCI IO card
11:58 AM NoSpark: how does that work with extended functionality of the different guis?
11:59 AM rigid: or linuxcncrsh which also doesn't need X
11:59 AM NoSpark: I really love the idea of a cmd lin CNC machine :)
11:59 AM NoSpark: *cmd line
11:59 AM NoSpark: but unfortunaly, other ppl need to use this machine too
12:00 PM rigid: what doesn't work via NML doesn't work. If you encounter bugs you could bring them up. I'm currently using axis to send gcode files and watch the machine going. Also jogging and homing not much more.
12:00 PM NoSpark: I was hoping to use something like qtDragon or touchy
12:01 PM rigid: should work aswell but tooltable probably doesn't work if you didn't compile with "--with-toolnml" (and still might be buggy). Also GUI has no access to HAL
12:02 PM rigid: if you encounter any bug and open an issue, i'll look into it eventually.
12:02 PM NoSpark: qtdragon has a lot of hal functionality
12:02 PM NoSpark: and seems to have an SQL backend for tool managment
12:03 PM NoSpark: so that probably is a no-go
12:03 PM rigid: never uses qtdragon. And experimenting with a production machine is probably not wise
12:04 PM rigid: adding dedecent network capabilities to HAL shouldn't be an issue. Tooltable code seems a bit messy aswell.
12:04 PM NoSpark: It's kinda "my" production machine, so I can handle any issues that crop up, and I have a spare CNC as well
12:05 PM NoSpark: no disagreeing with you though
12:05 PM rigid: Then experimenting with remote GUIs probably is feasable. Latency will be no problem anymore.
12:05 PM rigid: ...but other stuff maybe will until someone fixes it :)
12:06 PM NoSpark: There is some functionality in qtdragon it that I REALLY wanted, specifically an easy to use table mapping feature, camera interface, and a cpl of other things
12:07 PM NoSpark: but nothing I can't live without
12:09 PM Roguish: rigid, be very wary and carefull messing with hal. particularly in the real time stuff.
12:09 PM Roguish: non-realtime, like tool stuff. probably no big deal.
12:11 PM rigid: Roguish: The HAL layer is decently written and NML was made for realtime. It'll work fine.
12:11 PM Roguish: hey, i
12:12 PM Roguish: m just saying be very, very careful
12:12 PM rigid: without networking, it will just use shmem and (iirc) 2 additional function calls
12:12 PM rigid: hey, we have a decent test-suite, haven't we? :)
12:13 PM Roguish: test suite is pretty good, I think, and slowly getting better. the gui stuff is difficult to test suite.
12:14 PM rigid: the linuxcncrsh test is a good place to test GUI<->milltask stuff headlessly when "emulating a GUI"
12:15 PM rigid: I added some stuff there that revealed some bugs, but I didn't look into what GUIs are doing in addition to that.
12:15 PM Roguish: check on the -devel channel.
12:16 PM rigid: I won't touch it anyway unless someone needs it and I don't need it. There's other things to do.
12:18 PM rigid: i.e. why does my avg. latency jump from 1us to 4us when I use taskset/isolcpus to put every RT thread on a separate CPU core. And why does it stay at 4us after I change it back to the exact original affinity setup
12:18 PM Unterhausen: solidworks is busy running a command, doesn't look good
12:19 PM rigid: Roguish: are you aware of anything GUI<->HAL related that actually uses realtime? I didn't see anything and can't even imagine anything
12:20 PM rigid: I suppose it's emulating halrmt which isn't realtime.
12:20 PM NoSpark: Wouldn't making anything in the GUI realtime be a really, really bad idea?
12:21 PM rigid: NoSpark: well, currently we are polling keyboard/mouse events on a realtime system. That's bad enough I'd say. People are even talking about running sound/pulseaudio which isn't very realtime friendly.
12:21 PM rigid: *with the default setup
12:22 PM NoSpark: The ESTOP loop is part of the realtime system, so what is the need for polling keyboard/mouse?
12:23 PM rigid: Xorg needs that to support keyboard/mouse input
12:23 PM NoSpark: I assume by polling, you mean bypassing the X11 IO layer?
12:23 PM NoSpark: (I'm not sure what the KB/mouse handling subsystem of X11 is called)
12:23 PM rigid: xevent and xinput iirc
12:24 PM NoSpark: I am 99% sure that kb events are pushed though?
12:24 PM rigid: not 100% sure about the internals but it was very irq heavy when I checked
12:24 PM rigid: "pushed through"?
12:24 PM NoSpark: Maybe I'm showing my age here, but they used to even fire an interrupt
12:25 PM Roguish: Unterhausen, SW sucks.
12:25 PM NoSpark: polling = pulled
12:25 PM rigid: yeah, you don't want interrupts on a realtime kernel
12:25 PM rigid: or you want as little firing as possible
12:26 PM NoSpark: I'm pretty sure you can't disable KB IRQs, but that may have changed. Something to do with the MAGIC SYSREQ key?
12:26 PM Unterhausen: I have only crashed sw once or twice in the decades I have used it. This seems particularly weird.
12:26 PM rigid: iirc PS/2 uses interrupts and USB is polled at systick/x
12:27 PM Roguish: Unterhausen, Then you are very lucky....
12:27 PM rigid: NoSpark: you can disable it by not connecting a keyboard/mouse and not running Xorg
12:27 PM NoSpark: I have a USB ethernet device that consumes 50% of a core handling IRQs
12:27 PM Unterhausen: maybe I just use good computers
12:28 PM Roguish: it can be pretty picky at times.... there's a whole lot of math going on inside....
12:28 PM NoSpark: Unterhausen: Thats not it, I run SW on a decent computer, and it crashes all the time. I find crashes are related to what functions you are using, more than anything
12:28 PM rigid: NoSpark: USB ethernet is not HID
12:28 PM NoSpark: I can crash solidworks in under a min if I try
12:28 PM NoSpark: rigid: I get that, was just an example of bad software/hardware for realtime
12:29 PM rigid: iirc there are also tools to decrease the HID polling rate. But headless is the way to go imo
12:29 PM Unterhausen: big, if true. Considered submitting a bug report?
12:29 PM NoSpark: ?
12:29 PM rigid: NoSpark: well, if it uses 50% you're still good. That sounds like time you really need to spend. Keyboard/Mouse - not so much
12:30 PM NoSpark: rigid: I def agree with going headless. no, 50% is sitting idle, I suspect it is a timer triggered interrupt that just calls a polling function
12:31 PM NoSpark: it doesn't change much based on traffic flow
12:31 PM NoSpark: I just changed to another device with a different chipset.
12:33 PM Unterhausen: headless would be nice, what's the nml workaround?
12:34 PM Unterhausen: I think everything went bad when I made something tangent that was already tangent
12:34 PM NoSpark: Unterhausen: Spline of complex curve?
12:35 PM NoSpark: s/of/or
12:35 PM Unterhausen: I ain't that fancy
12:35 PM Unterhausen: just making a relief cut so something isn't as likely to be over-constrained IRL
12:35 PM NoSpark: about 90% of my crashes are related to splines, so that may explain why you are so lucky
12:35 PM Roguish: Unterhausen, yes, constraints can cause lots of problems. I always lock as much down as possible, in sketches too,
12:36 PM Roguish: the fewer things that can move around the better off the model is.
12:37 PM Unterhausen: I don't know why it didn't warn me I was overconstraining it
12:38 PM Unterhausen: I always make sure my sketches are all black
12:38 PM Unterhausen: I'm going to actually make a full assembly this time before I order the parts
12:38 PM NoSpark: something else to keep in mind is to minimise stacking constraints. 3 or 4 references deep is ok, more than that causes problems
12:39 PM NoSpark: it you can't get around it, it's usually worth finding a point and "fixing" it, and then making new constraints from there
12:40 PM Unterhausen: mirroring apparently caused all my problems, which is annoying
12:42 PM Roguish: I avoid mirroring.... i sometimes will mirror a part, but then break the link...
12:42 PM NoSpark: or mirror the part, not the sketch
12:43 PM roycroft: i mirror stuff all the time
12:44 PM Roguish: hey roycroft. bought a new place yet????
12:44 PM roycroft: no, i'm still looking
12:44 PM roycroft: there's no inventory
12:44 PM roycroft: it's like the worst time in a generation to buy a house
12:44 PM roycroft: but interest rates went down again
12:45 PM Roguish: rates will bounce around. maybe up to 15% again....
12:45 PM roycroft: there hasn't been much inventory here since the early 2000s, when the californians accelerated their moving up and buying everything
12:46 PM Roguish: my nephew and his wife just closed on a new construction house up near Bremerton.
12:46 PM Roguish: they're both Nave civilian employees. enginerds.
12:47 PM roycroft: you know the folks - they bought a small house in pacifica for $70k in the '70s, sold it for $1.5 million, and bought a house 5x as big for $500k in oregon
12:47 PM roycroft: a house that was $200k until the californians started invading us :)
12:47 PM Roguish: yup
12:48 PM Roguish: i paid 230k in '98 for my house. it's now around 1m
12:48 PM Roguish: only owe about 100k
12:49 PM Roguish: those damn Calies.
12:50 PM roycroft: there's also not been any construction since the before times
12:51 PM Roguish: down here there's lots of planned building,but the fricking bureaucracy gets in the way... and the union costs are outrageous
12:52 PM Roguish: non of the governments (fed, state, county, city, ) can get off their fat asses and get anything done.
01:14 PM roycroft: the feds get a lot done
01:14 PM roycroft: the democrats investigate republicans, point fingers at them, and blame them for everything that is wrong with the universe
01:14 PM roycroft: the republicans investigate democrats, point fingers at them, and blame them for everything that is wrong with the universe
01:15 PM roycroft: all that investigating and blaming is a lot of work
01:17 PM lcnc-relay: <simonbasil> Has anyone here done knurling but turned with a single point tool
01:17 PM lcnc-relay: <simonbasil>
01:17 PM lcnc-relay: <simonbasil> I'm thinking depth for a 1.5mm pitch thread but program with 6mm pitch and do 2 starts each direction.
01:17 PM lcnc-relay: <simonbasil>
01:17 PM lcnc-relay: <simonbasil> Feeding z- start at 0 and 3 then z+ 1.5 and 4.5.
01:17 PM lcnc-relay: <simonbasil>
01:17 PM lcnc-relay: <simonbasil> Thoughts?
01:26 PM roycroft: knurling is a forming operation, not a cutting operation
01:26 PM roycroft: you can cut a decorative hatch pattern, but that's not knurling
01:30 PM Rab: Technically correct, and the formed peaks are what actually promote grip: https://upload.wikimedia.org/wikipedia/commons/b/bf/Knurling_closeup.jpg
01:31 PM Rab: But IIRC I have seen something like a single-point knurl-alike in a youtube video, once.
01:34 PM roycroft: i'm not making any judgements about either type of operation
01:34 PM roycroft: just making a distinction between them
01:34 PM XXCoder: isnt kurling also strength metal there since it deforms
01:35 PM Rab: If you go deep enough, it would be functionally identical to a formed knurl. It would be a really intense interrupted cut if you want cross-hatching. I doubt tooling would hold up in harder materials. The "right" way to do it is probably engraving with a v-bit on a rotary axis.
01:35 PM roycroft: well it does tend to work-harden metals
01:35 PM roycroft: and it does not remove metal
01:35 PM roycroft: so yes, it would be a bit stronger than decorative hatch-pattern cutting
01:36 PM roycroft: and the raised bumps do promote grip, although i think the cut grooves would do so as well, but to a lesser degree
01:37 PM roycroft: there are times when only knurling works
01:37 PM Rab: Recently I was thinking about just that: turned aluminum knobs with CNC-engraved "knurling" in decorative patterns not possible with simple knurling wheels. E.g. a celtic knot motif or whatever.
01:37 PM roycroft: for example, if you have non-inert valve guides in a cylinder head and they are worn, you can knurl them to reduce the diameter and then ream them back to size
01:38 PM roycroft: a cutting operation alone cannot accomplish that
01:38 PM roycroft: non-insert, rather
01:38 PM roycroft: i don't know of any valve guides that are inert :)
01:39 PM XXCoder: "kurling" that follows curve of knob would be interesting. usually its pretty hard to do, and clickspring does it for for his knobs
01:40 PM roycroft: yes, i like the knobs that clickspring makes
01:41 PM Rab: I remember who made the video, more-or-less...it's someone in Japan, they do like really dense 15-minute turning videos full of brilliant, hacky techniques. No narration or other fluff. The machines are meticulously clean.
01:41 PM roycroft: it's something i'd like to start doing at some point
01:41 PM Rab: I have no time for YouTube machining in general, but I was hooked on that channel.
01:41 PM lcnc-relay: <simonbasil> Hey man I understand I used the term knurl as it easilly describes what I am trying to achieve.
01:41 PM lcnc-relay: <simonbasil>
01:41 PM lcnc-relay: <simonbasil> Pretty much make aliminium parts that get epoxied into wood and want to make a good surface for bonding that is resistant to force in twist and sliding.
01:41 PM lcnc-relay: <simonbasil>
01:41 PM lcnc-relay: <simonbasil> I don't have the space in my gang tooling to fit a knurling tool but already have single point screw cutting tool
01:42 PM roycroft: one could knurl a celtic knot motif or any repeating pattern like that
01:42 PM roycroft: the imporant thing to do when the pattern is that large is to get the diameter of the part spot on before starting the forming operation
01:42 PM lcnc-relay: <simonbasil> Reboots that channel got me good too haha
01:43 PM roycroft: even conventional knurling, i.e. diagonals and cross-hatches, works best at particular diameters
01:43 PM roycroft: but for large form patterns it's crucial to get the diameter correct
01:43 PM lcnc-relay: <simonbasil> I am going to have a go probally later this year to do solid metal gear knobs with "knurling" on a radius
01:44 PM Rab: Pretty sure it's Swap Lamp: https://www.youtube.com/@SwaplampJapan
01:44 PM roycroft: which of course is some multiple of the pattern length divided by pi
01:45 PM lcnc-relay: <simonbasil> Yep that's the guy
01:45 PM Rab: Use of knurling tools detected, no single-point so far
01:45 PM roycroft: i don't know japanese, but just the thumbnail pics on the playlist makes the channel look interesting
01:46 PM Rab: Maybe I'm remembering this video, I was totally floored: https://www.youtube.com/watch?v=rrq_hgRY5yg
01:46 PM lcnc-relay: <simonbasil> I think the face knurl one was single point but can't remember
01:46 PM roycroft: i think one of the biggest problems beginning knurlers have is not paying attention to the diameter of the part
01:46 PM roycroft: having an off diameter, combined with poor technique, makes for really awful results
01:48 PM roycroft: machinery's handbook used to have a chart of proper diameters for various knurling patters
01:48 PM roycroft: i'm not sure if it still does - i don't have a recent edition
01:48 PM lcnc-relay: <simonbasil> Agreed i remember when I started was getting all sorts of crap knurls then one of the old hands explained it to me
01:48 PM lcnc-relay: <simonbasil> But havnt done it in 15 years
01:48 PM Rab: Yeah, it must've been face knurling I saw.
01:49 PM roycroft: if the diameter is wrong then on every revolution you're doing the whole forming operation anew starting at a different point on the part
01:49 PM Rab: But I don't know why cylindrical single-point knurling would be any harder.
01:50 PM roycroft: it's like thread cutting without synchronizing the tool
01:50 PM lcnc-relay: <simonbasil> Yep and you are just mashing the metal all over the place
01:51 PM Rab: simonbasil, but I wonder if this technique will get you significantly more holding power than just threading the OD of the part.
01:51 PM Rab: Or cutting some grooves.
01:52 PM lcnc-relay: <simonbasil> I already do grooves honestly ill probally just cut left and right handed thread at 2x the programmed pitch just to give a rough surface
01:53 PM Rab: Here's (effective) cross-hatching engraved with a v-cutter: https://www.youtube.com/watch?v=cDfMI5ahbJI
01:54 PM lcnc-relay: <simonbasil> I really love fiddling with the 3d printed version that I made
01:55 PM XXCoder: wow that face cutting thing is very interesting
01:56 PM lcnc-relay: <simonbasil> Unrelated do you think I can radial drill 3.3mm in aliminium on lathe with a delta 750w servo as a spindle motor in a 1:1 gear ratio?
01:56 PM roycroft: it's slightly reminiscent of a rose cutting engine, but without the fancy machine
01:56 PM lcnc-relay: <simonbasil> I think it should work but wondering if anyone here had done something similar
01:56 PM XXCoder: yep roy
01:56 PM Rab: Reminds me of the various patterns in off-the-shelf threaded inserts: https://www.tappex.co.uk/wp-content/uploads/2014/07/tappex-self-threading-inserts-montage-21-900x711.jpg
01:57 PM lcnc-relay: <simonbasil> Very similar
01:59 PM Rab: simonbasil, 750W is like 1HP? Seems very likely to me.
02:00 PM lcnc-relay: <simonbasil> Yeah 1hp exactly
02:00 PM lcnc-relay: <simonbasil> https://www.ebay.co.uk/itm/374477880971?mkcid=16&mkevt=1&mkrid=711-127632-2357-0&ssspo=UYbRqHWVRia&sssrc=4429486&ssuid=aZLFPLJgQAy&var=643619048423&widget_ver=artemis&media=COPY
02:00 PM Rab: You might want to increase speed/feed and use coolant if the motor bogs down.
02:01 PM lcnc-relay: <simonbasil> Oh you misunderstand sorry the lathe main spindle 750w delta ac servo
02:01 PM Rab: right
02:02 PM Rab: So the workpiece would be fixed, and the drill would be in the spindle?
02:03 PM lcnc-relay: <simonbasil> https://jauriarts.org/_matrix/media/v1/download/jauriarts.org/AMzBMCwxTLyuKQdnOCCCexZd
02:03 PM lcnc-relay: <simonbasil> Making this part I want to drill the 3 holes with the pnumatic drill on the x axis
02:05 PM lcnc-relay: <simonbasil> I can tap them with a battery drill and a tap after
02:06 PM Rab: simonbasil, so the servo-driven spindle would index at 120 degrees and hold the part stationary for drilling?
02:06 PM lcnc-relay: <simonbasil> Yep
02:07 PM Rab: Sounds legit, but I base that on absolutely nothing. ;) I guess if the spindle rotates, you'll need a brake.
02:09 PM lcnc-relay: <simonbasil> I think it will be ok
02:10 PM Rab: I would worry more about starting on the curved surface. If the drill wanders, you might need stub-length or enlarged shank for more rigidity. That seems like a small part, maybe a center drill would be long enough...then you could chamfer, too.
02:11 PM lcnc-relay: <simonbasil> Yeah I will use stub length drills with around 20mm flute length
02:12 PM lcnc-relay: <simonbasil> And if it works ill get some custom drills ground to chamfer the top and geometry specifically for ali
02:13 PM lcnc-relay: <JT (@jt-shop:matrix.org)> A spotting bit is best to start a hole
02:16 PM lcnc-relay: <simonbasil> Yes it is Mr bot
02:16 PM Rab: If you're feeling ambitious and impatient, you could try a combination drill/tap...but I think it will need a more controllable driver than the pneumatic drill. https://www.mcmaster.com/tap%2fdrill-combination-tools/
02:17 PM lcnc-relay: <simonbasil> Just looked it up a center drill with a 3.15mm tip diameter has a cut length of 4.4mm so doesn't quite work
02:19 PM lcnc-relay: <simonbasil> I did consider using drill tap but can't get one with the right dimensions
02:19 PM lcnc-relay: <simonbasil> Currently the pnumatic drill is quick and dirty to see if the idea has legs
02:24 PM lcnc-relay: <simonbasil> https://jauriarts.org/_matrix/media/v1/download/jauriarts.org/tvCXyWqXalnzrbJtBWTrcJdl
02:24 PM lcnc-relay: <simonbasil> Speaking of quick and dirty this is a easy ish way to get a db25 cable into a cabinet without leaving a massive hole in it
02:26 PM Rab: Yeah, nice grommet
02:34 PM XXCoder: honestly 3d printer is amazing for random bobs and ends
02:34 PM XXCoder: holders etc
02:44 PM roycroft: that's a nice solution, although personally i prefer to avoid passing cables through a chassis like that
02:45 PM roycroft: if there's room i'll mount an idc d-shell on the back of the chassis and pass the signal through to the inside with a ribbon cable and another idc
02:46 PM roycroft: sometimes there's not enough real estate to do that, and i get that
02:46 PM roycroft: and sometimes having the extra connector would result in excessive signal loss or emf
02:46 PM roycroft: er, emi
02:47 PM roycroft: that grommet is pretty clever, though
02:55 PM lcnc-relay: <JT (@jt-shop:matrix.org)> I'm stuffed
02:56 PM lcnc-relay: <simonbasil> Agreed I couldn't fit the pc in the cabinet so had to go through and I have nightmares of chasing emi due to daisy chained dsub cables and bulkhead connectors from when I was a service engineer
03:32 PM Tom_L: JT, anything new to look at?
03:34 PM Tom_L: finally remembered to pick up my missing drill bits for the set and grabbed a few more dowel pins while i was there
04:23 PM JT-Shop: not tonight just got back from Durso Hills Winery
04:23 PM Tom_L: stuffed
04:24 PM Tom_L: snow predicted for overnight
04:25 PM roycroft: we have snow in the forecast for friday next
04:25 PM roycroft: but we often have snow in the forecast that far out, and it almost always changes to rain as the date approaches
04:25 PM JT-Shop: some kind of precip for tomorrow but they can't make up there mind
04:26 PM XXCoder: roy yep its annoying
04:26 PM Tom_L: 3-5" predicted so maybe not so bad
04:26 PM JT-Shop: i'm stuffed as well
04:26 PM XXCoder: no snow in prediction for here
04:26 PM JT-Shop: couldn't even finish 1/2 of my pizza... I must be slacking
04:27 PM Tom_L: now that guy want's to find more things to do on his cnc besides mill parts
04:27 PM XXCoder: repeat that and you'll have to turn in your pizza card
04:27 PM JT-Shop: they have plenty of to go boxes :)
04:27 PM Tom_L: not as good on day 2
04:28 PM JT-Shop: she finally remembered that a large to go box will not fit and brought two medium boxes today
04:28 PM Tom_L: :)
04:28 PM Tom_L: the problems of owning a C8
04:29 PM JT-Shop: well you just learn to deal with it
04:29 PM JT-Shop: hmm must be time to let Pea Wee and Lucy out
04:29 PM Tom_L: i should get my thread cutting macro out and adapt it for springs now
04:34 PM JT-Cave1: I did figure out a slick way to know if a dpll thingy is needed in the hal file so moving on tomorrow
04:35 PM Tom_L: how's that?
04:35 PM Tom_L: i don't think it hurts to add it anyway unless there's not room due to other functions