#linuxcnc Logs
Jul 18 2025
#linuxcnc Calendar
05:42 PM Tom_L: from the sounds of it, i may lose power again
07:24 PM roycroft: xxcoder: are you about?
07:24 PM xxcoder: yup
07:25 PM roycroft: has that hall sensor with the threaded body ever worked for you?
07:25 PM roycroft: i can't get anything out of it
07:25 PM roycroft: not even millivolt pulses
07:25 PM xxcoder: if what I recall was right, it only works at lower rpm
07:26 PM roycroft: i have tried at low rpms
07:26 PM xxcoder: but then that blue one is only 500 hz and I had remembered it being much higher so dunno
07:26 PM roycroft: low enough that i can count them
07:26 PM xxcoder: odd. it was working last I knew
07:27 PM roycroft: as in i can see the wrench ridges on the collet and count them as it spins
07:27 PM roycroft: hmm
07:27 PM roycroft: ok
07:27 PM roycroft: maybe my fluke just doesn't like what it's sending
07:27 PM roycroft: i'll try connecting it to the mesa board
07:28 PM roycroft: although that introduces another variable, as i've never configured an encoder circuit with linuxcnc
07:28 PM xxcoder: ok. I guess it wasnt too high on voltage so safe enough to test
07:28 PM xxcoder: any olloscope?
07:28 PM roycroft: yeah, i can use my 'scope
07:28 PM roycroft: that might be more forgiving than the mesa board
07:28 PM roycroft: if it still works
07:29 PM roycroft: it's another thing i haven't used since i moved, but it is not buried, so i can easly get to it
07:29 PM xxcoder: yeah. hope it does but dunno
07:29 PM roycroft: it's a tektronics 465b
07:29 PM roycroft: just about the most bullet-proof 'scope ever made
07:30 PM roycroft: it takes a few minutes to warm up, but it's always worked
07:30 PM roycroft: i guess that means i should go plug it in so that few minutes' clock can start
07:31 PM roycroft: i may not be aiming it at the magnets
07:31 PM roycroft: so it could be as simple as that
07:31 PM roycroft: but with the 'scope i can pass a magnet in front of it and look for a blip
07:31 PM xxcoder: yeah easy enough to test if it even works in first place, then see if higher hz works
07:36 PM lcnc-relay: <.le_potato@> 6061 is 100$/kg
07:36 PM lcnc-relay: <.le_potato@> nice prices
07:43 PM lcnc-relay: <.le_potato@> on aliexpress 25$/kg
07:56 PM Tom_L: it's probably the easiest to machine
08:01 PM _unreal_: I dont know if I stated or not but I finally FINALLY found my computers that drive my machines
08:01 PM _unreal_: daughter miss labled them as books and stuffed animals.
08:01 PM _unreal_: we are talking laptops in laptop bags
08:03 PM _unreal_: box feels like a book and stuffed animal
08:04 PM roycroft: so i don't know what to do now
08:04 PM roycroft: the 'scope is fine
08:05 PM roycroft: but when i probe the hall sensor i get essentially white noise
08:05 PM Tom_L: is it open collector output?
08:05 PM Tom_L: maybe it needs a pullup
08:05 PM roycroft: when i probe the "dc" power suppply output i get several 60hz sine wave-ish things
08:05 PM roycroft: maybe
08:06 PM roycroft: i got out a function generator, and when i plug the 'scope into that i get nice clean sin waves and square waves and sawtooth waves
08:06 PM roycroft: so the 'scope seems ok
08:06 PM Tom_L: put a 10k on it and see what happens
08:06 PM roycroft: my concern is that when i eliminate the hall sensor i get that crap from the power supply
08:06 PM xxcoder: yeah dc shouldnt be that noisy
08:07 PM roycroft: i should be getting square waves at worst
08:07 PM Tom_L: smps tend to make some
08:07 PM roycroft: but a straight line would be better
08:07 PM roycroft: so i think i'm going to start making dinner
08:07 PM roycroft: because this project is turning into a headache, and that's not what i need
08:07 PM xxcoder: had sushi. yum
08:07 PM roycroft: and yes, i thought about a pullup resistor
08:08 PM roycroft: and i might try that, but if i'm not feeding something approaching dc to the hall sensor i don't expect good results
08:08 PM Tom_L: 10k wouldn't hurt anything i doubt
08:08 PM Tom_L: 1k may be more reasonable if it needs one
08:08 PM xxcoder: I wonder if 24v psu I got was utter crap
08:11 PM Tom_L: if you're worried about that, plug a couple 9v batteries together and try it at 18v
08:11 PM Tom_L: or 9 for that matter
08:11 PM roycroft: i'm trying to feed 12v to the hall sensor
08:11 PM Tom_L: iirc it had a pretty broad range
08:11 PM roycroft: with a voltage reducer coming from the 24v psu
08:12 PM roycroft: i can use a bench psu for it and set it to whatever voltage i want
08:12 PM xxcoder: the dc-dc convertor? I recall it have that
08:12 PM roycroft: yes
08:12 PM roycroft: and that may be the problem
08:12 PM xxcoder: it should accept quite wide range of input volt.
08:12 PM xxcoder: but maybe?
08:12 PM Tom_L: just suggesting a battery would be pure dc
08:14 PM xxcoder: yeah should be able to send battery 9v * 2 into dc-dc see if output is clean
08:15 PM roycroft: or just use a high quality bench power supply that has clean dc output
08:15 PM roycroft: and verify the output with the 'scope first
08:15 PM Tom_L: any number of valid tests
08:15 PM roycroft: getting a 9v battery would require going outside and going into the shop
08:15 PM xxcoder: indeed. I somehow always forget you have nice electrics stuff
08:16 PM roycroft: and i do not want to do that - although it's cooled off, it's still more comfortable in the house than outside :)
08:16 PM Tom_L: i use what's handy
08:16 PM roycroft: i have some nice electronics gear
08:16 PM roycroft: but not a lot
08:16 PM Tom_L: some have troble thinking outside the box sometimes
08:16 PM roycroft: yes
08:16 PM lcnc-relay: <.le_potato@> https://vi.aliexpress.com/item/1005002538659888.html
08:16 PM lcnc-relay: <.le_potato@> is dis good?
08:16 PM Tom_L: not suggesting anything...
08:16 PM roycroft: aliexpress does not exist for me any more
08:16 PM roycroft: i will not pay tariff roulette
08:17 PM lcnc-relay: <.le_potato@> why
08:17 PM lcnc-relay: <.le_potato@> ah
08:17 PM lcnc-relay: <.le_potato@> i pay 30% tariff...
08:17 PM roycroft: i only purchase things shipped from the usa
08:17 PM Tom_L: Sorry, the page you requested can not be found:(
08:17 PM xxcoder: looks like some kind of tool length sensor
08:17 PM xxcoder: also not found after it redirects to .us domain
08:17 PM roycroft: well if i knew what the tariff would be i would be able to make a decision
08:17 PM lcnc-relay: <.le_potato@> https://prnt.sc/jK5I_PAx2mcx
08:17 PM roycroft: but in this country it's tariff roulette, literally
08:17 PM roycroft: it could be 30% the day it leaves china
08:18 PM lcnc-relay: <.le_potato@> how much is the tariff?
08:18 PM roycroft: and it could be 140% the day the tariff collector comes to my door
08:18 PM xxcoder: who knows
08:18 PM roycroft: LITERALLY
08:18 PM Tom_L: 3 left hand fingers
08:18 PM xxcoder: RNG tariffs
08:18 PM roycroft: i am not making up numbers
08:18 PM roycroft: that's the range it's been on imports from china since the current regime took over
08:18 PM roycroft: and it goes up and down like a yo-yo
08:19 PM roycroft: so my rule is to only buy things shipped to me from this country
08:19 PM roycroft: we don't have a blue state tariff yet ...
08:19 PM Tom_L: let's not open that can o worms
08:20 PM lcnc-relay: <.le_potato@> why toolsensor need air?
08:20 PM roycroft: my point is that no person in the us can make a reasonable decision about purchasing any foreign-made goods that are not shipped domestically
08:20 PM roycroft: and i'll leave it at that
08:20 PM roycroft: aliexpress is dead to me :)
08:20 PM xxcoder: potato to keep pad clear
08:21 PM lcnc-relay: <.le_potato@> really
08:21 PM xxcoder: chip could mistakenly lengthen the tool
08:21 PM lcnc-relay: <skunkworks8841@> it homes.. I sped up the home speed and it fixed the bouncing problem so far
08:21 PM Tom_L: did you use the debounce comp or hardware?
08:21 PM xxcoder: or rather, "lengthen" the tool. mismeasures can cause serious problems.
08:22 PM Tom_L: resistor/cap thing
08:22 PM Tom_L: i've almost forgotten now
08:22 PM * roycroft goes off to heat up the oven and let the dough start rising
08:23 PM Tom_L: https://www.picotech.com/library/articles/blog/what-is-switch-bounce-how-to-implement-debounce
08:23 PM Tom_L: comp would be easier
08:24 PM Tom_L: iirc typical bounce is around 20ms
08:25 PM roycroft: maybe i'll pull out that bench ps while the dough is rising and see if it cleans things up
08:25 PM lcnc-relay: <.le_potato@> NO or NC doesnt matter?
08:26 PM Tom_L: usually not
08:26 PM lcnc-relay: <.le_potato@> NO better?
08:26 PM xxcoder: if dc-dc was bad part, then its sad. I paid little bit more so I had nice case and all lol
08:26 PM Tom_L: you can invert the signal in the config
08:26 PM lcnc-relay: <.le_potato@> idk on mesa board how input works
08:26 PM lcnc-relay: <.le_potato@> like is it have pullup or pulldown resistor
08:26 PM lcnc-relay: <.le_potato@> or can only switch to ground?
08:27 PM Tom_L: http://linuxcnc.org/docs/stable/html/drivers/hostmot2.html#_gpio
08:28 PM lcnc-relay: <.le_potato@> In this high-impedance state the I/O pin floats (weakly pulled high),
08:28 PM lcnc-relay: <.le_potato@> so i need NO
08:29 PM Tom_L: i don't trust those and if anything will add a pullup/down
08:29 PM Tom_L: or anything that has an open collector output will float
08:31 PM lcnc-relay: <.le_potato@> then idk
08:31 PM Tom_L: i wasn't able to see your link
08:31 PM lcnc-relay: <.le_potato@> i sent screenshot
08:31 PM lcnc-relay: <.le_potato@> there is lot of chinese tool setter
08:31 PM Tom_L: for example on my touch probe i didn't need anything
08:32 PM Tom_L: iirc it's NO
08:33 PM Tom_L: net touch-probe hm2_[HOSTMOT2](BOARD).0.gpio.028.in_not => motion.probe-input
08:33 PM Tom_L: in_not inverted it
08:34 PM lcnc-relay: <.le_potato@> https://prnt.sc/_TJ6opKNb3hu
08:34 PM lcnc-relay: <.le_potato@> https://prnt.sc/c9WBD-ETJ5ge
08:34 PM lcnc-relay: <.le_potato@> there is some good stuff in china
08:35 PM pcw--home: If you are talking about the isolated inputs on a 7I95/7I96 those have a common pin tha need to go to V+ for NPN sensors (or contact closure to ground) or GND for PNP sensors (or contacct closure to V+)
08:37 PM pcw--home: I would not normally use bare FPGA pins for switches and what not as a misconnect to 12 or 24V would be bad
08:37 PM lcnc-relay: <.le_potato@> use opto?
08:37 PM pcw--home: there are optos on the 7I96/7I95
08:39 PM pcw--home: If you use bare FPGA pins, it's best to have a series resistor to protect the input clamp diodes
08:45 PM roycroft: well i set the 'scope to 1v per division, and used a 10x attenuating probe, and when i set the ps to 11.73v (it's touchy and i just wanted to get close) my flat line went up ~1.2 divisions and stayed dead flat, so i think the 'scope is working well and calibrated, and that the psu is clean
08:45 PM roycroft: i'll check after dinner to see if the hall sensor still makes white noise with clean power
08:46 PM xxcoder: tested the 24v-12v dc-dc?
08:46 PM roycroft: no
08:46 PM roycroft: and right now i don't care about that
08:46 PM roycroft: nor the 24v psu, since i've already decided to swap it out for a 48v psu
08:47 PM xxcoder: the spindle one? thought you was using the 24v for movement system for this
08:47 PM roycroft: i'm focusing on one thing at a time, because one thing is the most my brain can handle at any given time these days
08:47 PM lcnc-relay: <_sebastianm@> good evening
08:47 PM roycroft: no, i already have 48v for motion control
08:47 PM lcnc-relay: <_sebastianm@> is anyone aware of a linuxcnc "feature" that gives me a Z-softlimit depending on the current tool length?
08:47 PM lcnc-relay: I would love to have a mechanism that prevents me cutting into my spoilboard by accident
08:47 PM lcnc-relay: like... "'....dont go deeper than 60mm below my TLS...."
08:47 PM xxcoder: oh! no wonder I was confused lol. ok
08:48 PM roycroft: i've never connected the 24v psu to any of the drivers - i knew from the start that would not work well
08:48 PM Tom_L: sebastianm, use the tool table to set each tool length
08:49 PM xxcoder: so the psu that you got with machine was 48v also
08:49 PM Tom_L: i don't now of a 'softlimit'
08:49 PM xxcoder: ? or was it 24v and you ditched it?
08:49 PM roycroft: no
08:49 PM Tom_L: per se
08:49 PM roycroft: the one i got with it was 24v
08:49 PM roycroft: that's what i got from you
08:49 PM roycroft: i already had some 48v psus
08:49 PM xxcoder: hmm probably why I had so much trouble with tuning motion system
08:49 PM roycroft: so i used one of mine for motion control
08:49 PM roycroft: and was going to use yours for the spindle
08:49 PM roycroft: which is what i thought it was for, for some reason
08:50 PM xxcoder: ok
08:50 PM Tom_L: sebastianm, if your offsets are set correctly and the z work offset is set it shouldn't happen
08:50 PM roycroft: and i've no idea why i thought that
08:51 PM xxcoder: probably crossed wire in brain ;)
08:51 PM roycroft: that's not unusual for me these days
08:51 PM Tom_L: sebastianm, maybe setting the Z- home limit would be an option but a long shot
08:52 PM lcnc-relay: <skunkworks8841@> https://photos.app.goo.gl/5tiLGY9tCnjGBaaA8
08:53 PM Tom_L: sebatianm, MIN_LIMIT = for each axis
08:54 PM Tom_L: probably not the solution you were looking for
08:55 PM Tom_L: skunkworks, 2 Y axis drivers?
08:56 PM Tom_L: nice tight fit
08:57 PM lcnc-relay: <.le_potato@> for touch probe, i meassure Z on aluminium
08:57 PM lcnc-relay: <.le_potato@> then i change tool and go to tool setter?
09:07 PM lcnc-relay: <_sebastianm@> Tom_L: but that is not depending on the tool, right? with a very long tool I need to have the limit "earlier" than with a very short bit
09:10 PM Tom_L: that's why you have tool length offsets
09:10 PM Tom_L: most machines with tool changers preset the tools
09:11 PM Tom_L: ones that don't usually use a tool touchoff of some sort
09:11 PM Tom_L: zeroing them one at a time
09:12 PM Tom_L: that's what the H word is for on a tool change
09:13 PM Tom_L: http://linuxcnc.org/docs/stable/html/gcode/m-code.html#mcode:m6
09:13 PM Tom_L: http://linuxcnc.org/docs/stable/html/gcode/g-code.html#gcode:g43
09:16 PM Tom_L: G43 H1 (set tool offsets using the values from tool 1 in the tool table)
09:18 PM lcnc-relay: <skunkworks8841@> Tom_L: yes X1,x2
10:16 PM roycroft: yikes
10:16 PM roycroft: the cram fire is now 90,000 acres
10:16 PM roycroft: it was 20,000 acres on wednesday afternoon
10:29 PM xxcoder: that sucks
10:36 PM roycroft: new uri tuchman video
10:36 PM roycroft: he's making a makita knife
10:37 PM fluffywolf: new applied science video!
10:46 PM fluffywolf: today's video... he seems to have built a high-resolution x-ray backscatter imager in his garage.