#hazzy Logs
Jul 03 2020
#hazzy Calendar
12:04 AM TurBoss: <TurBoss "VectorImage_2020-07-03_062941.jp"> ok so on mode 2 pins from 16 to 19 become mpg inputs?
12:05 AM TurBoss: i need to change some wires
12:15 AM Joco[m]: looks like it
12:53 AM TurBoss: @joco_nz:matrix.org: got all the inputs plus analog working
12:54 AM TurBoss: the encoder has 3 wires A B and common
12:55 AM TurBoss: common goes GND or +5
01:02 AM Joco[m]: so - looking at the digram I posted earlier: https://matrix-client.matrix.org/_matrix/media/r0/download/matrix.org/PDYwOqrVPNeULmHomhNoIyWU
01:03 AM Joco[m]: I reckon you need to feed 5v to common so that A B can feed a voltage to the mesa when they trigger
01:03 AM TurBoss: ok
01:03 AM Joco[m]: I would check this
01:04 AM Joco[m]: by putting VDC to common and use a volt meter to see if you get a voltage off A/B when you turn the knob
01:04 AM Joco[m]: but that's just me. I'm paranoid about electricity
01:06 AM Joco[m]: my reasoning for the circuit is this
01:06 AM * Joco[m] uploaded an image: image.png (21KB) < https://matrix.org/_matrix/media/r0/download/matrix.org/bOpsJFLWiEdTKNBqlomaVzIa >
01:06 AM Joco[m]: the only thing that makes sense is that A/B get a positive voltage when the MPG is turned. Else this make no sense to me.
01:07 AM TurBoss: ok the fire test
01:07 AM Joco[m]: the only other thing I can think of is there is a 4th 5V "wire" that you are missing on the encoder
01:08 AM Joco[m]: Alternative is that the computer encoder is for a sourcing sensor not a sinking one
01:08 AM TurBoss: it works
01:08 AM Joco[m]: and the mesa is sinking.
01:08 AM TurBoss: it counts
01:08 AM Joco[m]: cool
01:09 AM TurBoss: hurra
01:09 AM TurBoss: no mafic amoke
01:09 AM TurBoss: magic
01:09 AM Joco[m]: yah - I'm not completely useless after all - lol
01:09 AM TurBoss: :)
01:09 AM TurBoss: thanks
01:10 AM TurBoss: do i still need w4 w5 w6 in left hand?
01:41 AM Joco[m]: nope
01:42 AM Joco[m]: they relate to the spindle encoder. Think that works differently to MPG
01:46 AM TurBoss: ok
01:54 AM Joco[m]: hmmm. Been tuning small mills cutting. Have external dimension cuts quite accurate. To within 0.02mm. Just done some square pocket cuts. interestingly the Y axis is bang on (i.e. within 0.02mm) but the X is 0.1mm too large. Its very consistent.
01:55 AM TurBoss: cool
01:55 AM Joco[m]: I did a test where I modelled the pocket length at 36.9mm aiming for a 37mm length and after milling its bang on 37mm (to within 0.02mm). Wonder why its out. :-/
01:56 AM Joco[m]: I can totally manage the quirk. The fact it is so consistent allows that. Just not sure why its out is all. Something must be flexing somewhere.
01:57 AM TurBoss: tools can flex
01:57 AM Joco[m]: External dimensions more important to me anyway. In the vast bulk of cases an internal pocket that is within 0.1mm will be fine.
01:57 AM TurBoss: at job they do a seccobd free pass
01:58 AM TurBoss: finish one
01:58 AM Joco[m]: the cut is in ali and I'm doing two 0.2mm finishing passes so don't think that is the issue
01:58 AM TurBoss: because of tool blend
01:58 AM TurBoss: tool diameter is 16 here
01:59 AM Joco[m]: 8mm tool
01:59 AM TurBoss: and it blends
01:59 AM TurBoss: lol
01:59 AM TurBoss: 16 or 18
01:59 AM * TurBoss uploaded an image: VectorImage_2020-07-03_085004.jpg (239KB) < https://matrix.org/_matrix/media/r0/download/jauriarts.org/XKZXyOvmQfUsLAIJBfwmemXW >
01:59 AM Joco[m]: blends or bends?
01:59 AM TurBoss: flexs
01:59 AM TurBoss: bends
01:59 AM TurBoss: sorry
02:00 AM Joco[m]: yeah - though so. You must be pushing it hard for something that thick to bend
02:00 AM TurBoss: aero aluminium
02:00 AM Joco[m]: but I had thought of that but an 8mm tool isn't going to flex on a 0.2mm cut whith a chip load of 0.2mm
02:00 AM TurBoss: most strong
02:00 AM Joco[m]: 0.02mm
02:01 AM Joco[m]: * but I had thought of that but an 8mm tool isn't going to flex on a 0.2mm cut whith a chip load of 0.02mm
02:01 AM TurBoss: duno
02:01 AM TurBoss: i cant help
02:01 AM TurBoss: thats the operator who knoes
02:01 AM Joco[m]: Is ok. Will ping Lcvette when he's backn on deck
02:01 AM TurBoss: oki
02:01 AM Joco[m]: he might have some sage advice. Which could be "live with it" - lol
03:16 AM TurBoss: jog wheel begin to work very nice
03:20 AM Joco[m]: good stuff
03:27 AM TurBoss: Joco: do you know if I can set feed override from hal?
03:27 AM TurBoss: or do I need a componet to do that
03:39 AM TurBoss: ```
03:39 AM TurBoss: halui.feed-override.counts
03:46 AM Joco[m]: should be able to
03:47 AM Joco[m]: people hook these things up to external components so all get hooked up to HAL
04:06 AM Joco[m]: TurBoss: https://www.forum.linuxcnc.org/24-hal-components/30794-overrides-hal-example
04:20 AM * TurBoss sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/UZoJnVkfmllCNpnIsvvnplOY >
04:20 AM TurBoss: the problem i have is in the last line
04:24 AM TurBoss: got it
04:24 AM TurBoss: hurray
04:26 AM Joco[m]: \o?
04:26 AM TurBoss: now i need to adjust the scale
04:27 AM Joco[m]: * \o/
05:11 AM JT-Cave: morning
05:17 AM TurBoss: hello JT
05:34 AM TurBoss: it works but only in fractions of 10
05:34 AM * TurBoss sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/nznDfizfdjOHYtwBMRvkpSlU >
05:55 AM hazzy-m: Morning
06:22 AM TurBoss: Hello!
07:07 AM * JT[m] sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/GjuefguAVQRrNSgqsgjEqrqJ >
07:07 AM JT[m]: That's my MPG on the BP mill
07:39 AM TurBoss: JT: what about halui feedoverride?
08:02 AM JT[m]: feedoverride for MPG?
08:04 AM TurBoss: a potentiometer for feed
08:12 AM Lcvette: o/
08:12 AM Lcvette: still working on it turboss?
08:13 AM TurBoss: o/
08:13 AM TurBoss: got everything except the poti
08:13 AM TurBoss: the feed override goes 10 by 10
08:14 AM TurBoss: 12, 20, 30 , etc
08:14 AM TurBoss: jog wheel axis select and jg steps are ok
08:16 AM Lcvette: is 10 not ok?
08:16 AM Lcvette: can you scale it?
08:18 AM TurBoss: how
08:18 AM TurBoss: I need 1 by 1
08:18 AM TurBoss: 1%, 2%
08:23 AM Lcvette: need 1x1
08:25 AM * JT[m] uploaded an image: VectorImage_2020-07-03_081427.jpg (7360KB) < https://matrix.org/_matrix/media/r0/download/matrix.org/GPVDvspQzQCSOSJPJdhhekEn >
08:25 AM JT[m]: Note to self, the next place you drive Rusty to is to get new tires on the front
08:25 AM Lcvette: TurBoss: can i see what you have?
08:26 AM Lcvette: JT: plenty of life left there!
08:26 AM Lcvette: :D
08:28 AM TurBoss: surf
08:29 AM Lcvette: i see
08:29 AM Lcvette: so you have not added a scale to it?
08:30 AM * Lcvette sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/TmvuxdJWPrVnUYoxjFClhkHL >
08:30 AM Lcvette: TurBoss: ^^^
08:30 AM Lcvette: this is the thing for the pot
08:31 AM Lcvette: what is the resolution on the input of the mesa?
08:31 AM Lcvette: do you know?
08:31 AM TurBoss: ok
08:32 AM Lcvette: * this is the thing for the pot?
08:33 AM Lcvette: maybe ask pcw?
08:34 AM JT[m]: Lcvette: yea but the cracked tread is a bit of a concern...
08:35 AM Lcvette: @pcw_home or @pcw_mesa
08:35 AM TurBoss: 8bit
08:35 AM Lcvette: probably better safe than sorry
08:36 AM JT[m]: yea
08:37 AM JT[m]: hmm, I wonder if the CHNC has feed over ride knob
08:38 AM TurBoss: por value goes from 0 to 24
08:38 AM Lcvette: TurBoss: what about a slider?
08:39 AM Lcvette: have to have mechanical knob?
08:39 AM TurBoss: its almost ready
08:39 AM Lcvette: ok
08:39 AM TurBoss: yes
08:40 AM TurBoss: :p
08:41 AM Lcvette: https://forum.linuxcnc.org/24-hal-components/36336-physical-feed-override-knob?start=30
08:41 AM Lcvette: did you find this?
08:43 AM Lcvette: TurBoss: ^^^
08:43 AM TurBoss: hurray!
08:43 AM * Lcvette sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/PvJkcHvBPIvYDUtGeVomZUqd >
08:44 AM Lcvette: TurBoss: helpful?
08:44 AM Lcvette: :D
08:44 AM Lcvette: \o/
08:44 AM Lcvette: hurray!!
08:44 AM Lcvette: glorious victory awaits!
08:44 AM TurBoss: yes
08:46 AM Lcvette: there is more in that thread
08:55 AM JT[m]: the CHNC has some sort of selector switch for feed over ride in 10% increments
09:07 AM * TurBoss goes arround
09:07 AM TurBoss: latter
09:49 AM Lcvette: you get it working turboss?
10:19 AM TurBoss: nop
10:50 AM Lcvette: TurBoss: /o\
11:07 AM TurBoss: xD
11:08 AM Lcvette: and now?
11:08 AM Lcvette: is working?
11:08 AM Lcvette: :o
11:08 AM hazzy-m: TurBoss: https://github.com/KurtJacobson/RF45-config/blob/master/hal-comps/pot.comp
11:10 AM hazzy-m: Simple comp you could modify for scaling
11:11 AM TurBoss: i got it working just with jal
11:11 AM TurBoss: but it changes steps by 10
11:11 AM TurBoss: so i just need to set a correct scale
11:12 AM TurBoss: ill paste the code whe ill arraive home
01:05 PM TurBoss: 100jch: hello
01:05 PM TurBoss: sure ask
03:06 PM TurBoss: Great! I have the pot wonking !!
03:28 PM Joco[m]: Wonking? ROFL
03:31 PM Joco[m]: Lcvette: you got a moment to do a bit of explaining/teaching?
04:08 PM Joco[m]: Lcvette: for when you return: have dialed in back lash comp on mill in x and y. Out side cuts and measurements in x are good (within 0.02mm) but inside pocket cuts in x are 0.1mm oversized. Not getting my head around why that would be. Any wisdom?
06:23 PM JT-Cave: backlash can only be compensated for on one side of the backlash...
06:24 PM JT-Cave: if you have ANY backlash your parts will not be any more accurate than the amount of backlash even with compensation
08:07 PM TurBoss: hello
08:12 PM Joco[m]: thanks JT-Cave . So if I have a total backlash of 0.1 I can either compensate in one direction of X and accept 0.1 in the other direction on X or. Halve the error (i.e. compensate for 0.05) and accept that error in both directions in X.
08:12 PM Joco[m]: Eventually I will hunt down where the mechanical issue is. As time allows.
08:13 PM Joco[m]: Probably slop in the ball nut. A repack this slightly larger balls would probably solve the bulk of it.
08:13 PM Joco[m]: * Probably slop in the ball nut. A repack with slightly larger balls would probably solve the bulk of it.
09:23 PM TurBoss: aparently an android tablet with a wireless keyboard + mouse is not useless
10:54 PM Joco[m]: what you using for ?
11:29 PM Lcvette: JT-Cave: i don't follow your explanation about backlash being compensated for on one side of the backlash
11:34 PM Lcvette: I have not played too terribly much with it in linuxcnc, but with mach3 its working in both directions and you 100% can get parts to tolerances better than the backlash amount. I do it on my G0704 which runs mach 3 with backlash comp. It had almost 0.003" of backlash and i can get internal and external pockets working 0.001"
11:35 PM Lcvette: Joco: it sounds like your steps per unit is off
11:36 PM Lcvette: And ghb fixing with backlash achieves correct external tolerance but overwhelms inside pocket dimensions
11:37 PM Lcvette: * And fixing with backlash comp achieves correct external tolerance but overtravels inside pocket dimensions
11:38 PM Lcvette: Also are you doing circular pockets or square pockets