#linuxcnc-devel Logs

Jun 24 2023

11:55 AM pere: andypugh: hi. :)
12:34 PM pere: andypugh: how far did you get with the automatic tool changing code before you had to run? I would like to continue.
01:01 PM andypugh: Almost nowhere, I am afraid.
01:01 PM pere: andypugh: ok.
01:01 PM pere: did you have a nice trip home?
01:02 PM andypugh: I did get as far as putting entries in the INI file that referred to the correct tool change positions in machine coordinates, and setting the limit override signal high allows you out-of-bounds.
01:02 PM pere: note, your email supplier is throwing away my emails to you, so I can not reply to your emails. :)
01:04 PM andypugh: Yeah, that’s a Gmail thing, they only accept mails if the sending server has a valid SPF
01:04 PM andypugh: Recently, we've noticed that some of our customers have had
01:04 PM andypugh: > difficulties delivering emails to some Gmail accounts. This is because
01:04 PM andypugh: > Google has updated its security policies, and now requires anyone
01:04 PM andypugh: > sending emails to a Gmail mailbox to have a valid Sender Policy
01:04 PM andypugh: > Framework (SPF) record configured on their domain.
01:04 PM pere: looking at atc.ngc, I suspect classicladder used to take care of the 'limit override', 'open door' and 'unclamp' pins.
01:05 PM pere: andypugh: gmail is in practice a intranet, but not to worry, I will just not email you. :)
01:05 PM andypugh: My journey back was rather spoiled by a sudden passport queue. I had loads of time in the airport, so took a leisurely lunch, and then set off to the gate with half an hour to go, and missed the plane.
01:05 PM andypugh: pere: You can use my other account: andy@bodgesoc.org
01:06 PM andypugh: pere: I would suggest switching to G-code remap, as axis motion is required.
01:07 PM andypugh: The sample code from vmc_toolchange (which is already in your config) is a good starting point.
01:07 PM andypugh: I would practice with poition 1, where there is no tool carrier to hit ;;)
01:07 PM pere: yeah.
01:08 PM pere: andypugh: btw, can you write what you did to the vfd to get it working? I have no idea how you changed the config.
01:08 PM andypugh: I _think_ I set up the config to call the toolchange.ngc before I left, but not a lot more.
01:09 PM andypugh: I just lowered the V/F settings to 40V so that the current was low enough for the VFD to cope.
01:09 PM pere: I would rather do without classicladder, did your toolchange.ngc skip classicladder?
01:10 PM andypugh: Yes, there should be no need for CL, though it might be necessary to analyse what CL is doing, as removing it causes ptoblems. Possibly it is just passing-through signals, but without those pass-throught there are problems.
01:10 PM pere: yeah
01:14 PM pere: andypugh: btw, there is no toolchange.ngc. did you mean atc.ngc?
01:14 PM andypugh: Probably :-)
01:15 PM andypugh: Does the header say that it was copied from my VMC toolchange code?
01:15 PM pere: yeah
01:15 PM andypugh: That’s the one, then. Though it’s going to need a few changes, to that it opens the tool change door etc.
01:16 PM andypugh: (Instead of moving a tool arm)
01:16 PM pere: I also need to figure out how to properly select between the two probes. do you know any g-code examples using both tool height and work piece probe.
01:16 PM pere: andypugh: the way I read it, a lot was left to classicladder earlier, yes. it only set the 'tool change' pin and magic happened. :)
01:16 PM andypugh: And the various status HAL pins that tell you that the door is open (etc) need to be connected up.
01:17 PM pere: those pins are already connected, but not doing much at the moment.
01:17 PM pere: wired up, I mean.
01:17 PM andypugh: I think that they need to be connected to various motion.digital-in-NN pins so that G-code can read them.
01:18 PM pere: no standard way to do this? every machine is different?
01:19 PM andypugh: Yes, every machine is different. For example you have a tool-changer door, not everyone does.
01:24 PM pere: btw, sad to hear you lost your plane. how did you work your way around that?
01:25 PM andypugh: I bought another ticket. Which wasn’t my preferred outcome as I feel that I was actually at the gate _just_ in time.
01:26 PM pere: :(
01:27 PM pere: I hate airports, and that is one of the reasons.
01:28 PM andypugh: I used to fly a lot, (27 flights one year, when I counted) so I am vaguely used to it.
01:29 PM pere: what really get me is the security theater.
01:30 PM pere: anyway, I had an evening to play with the cnc the other day, and got to the point where I now try to learn FreeCAD Path to not have to create g-code myself.
01:32 PM pere: I also am back on looking at the tool database and feed and speed calculator API, to be able to keep track of the tool collection. :)
01:33 PM andypugh: There shouldn’t be too much to do with the tool table. It’s a simple non-random tool changer, I think?
01:33 PM pere: it can be, at least. :)
01:34 PM pere: seb_kuzminsky recommended to keep tools in specific pockets instead of tracking their whereabouts dynamically.
01:41 PM pere: <URL: https://www.haascnc.com/service/codes-settings.type=gcode.machine=mill.value=G37.html > talk about the operation I want and use M59 and M69 to enable and disable probe. Is this Haas specific or standardized?
01:43 PM pere: <URL: https://www.cnclathing.com/wp-content/uploads/Haas-Mill-M-Codes.pdf > claim it is M59 Output Relay Set (N), M69 Output Relay Clear (N). I guess it is the same as LinuxCNC M64/M65?
01:45 PM pere: Haas even dear to have tool clamp/unclamp as M code, M82 tool unclaim, M86 tool clamp.
01:46 PM pere: s/dear/dare/
02:57 PM andypugh: If you want to do that, you almost can.
02:58 PM andypugh: You could use M100-M199 (probably M182 and M186) calling a bash script that does a “setp”
02:58 PM andypugh: But, you might as well leave it as M64 P5 and M65 P5 as it is now.
02:59 PM andypugh: There are sensors for tool-unclamped and tool-clamped, and I would advise that you add checking those to the tool-change sequence.
03:00 PM andypugh: The sequence will be a lot of M64 and then M66 wait-on-inpiut to check the correct operation, and then a check that it wasn’t a timeout and an (ABORT, …..) magic somment if it is.
03:52 PM CaptHindsight[m]: pere: security theater like taking your small tube of liquid but allowing your CNC router in the cabin?
04:03 PM andypugh: Rumour has it that the liquids rule is going to be relaxed in 2024. https://www.airport-technology.com/features/uk-set-to-ease-airport-security-rules-in-2024/#:~:text=June%202024%20is%20the%20new,luggage%20during%20the%20security%20process.
04:21 PM pere: andypugh: thanks for the tip on M100->M199. Not sure it provide any real advantage over M64/M65. Using standardized M codes would make sense if it is actually standardized, which I realiase it is not.
05:06 PM andypugh: The custom m-codes are worth remembering, but are only one of many ways to do things.
05:06 PM andypugh: You _could_ probably remap the actual M-codes using Python (or G-code that actually calls M184…)
06:39 PM pere: andypugh: btw, did you want to learn how to update the package in debian?
06:39 PM andypugh: At some point, but probably not tonight,
11:35 PM -!- #linuxcnc-devel mode set to +v by ChanServ