#linuxcnc Logs

Nov 12 2023

01:59 AM Deejay: moin
02:16 AM jpa-: solarwind: continuously in one direction or without backlash in opposite directions?
02:56 PM solarwind: I'm trying to understand how my heidenhain glass scales work
02:56 PM solarwind: The read head outputs sinusoidal 1Vpp signals
02:57 PM solarwind: but I'm not sure how the glass is actually etched.
02:57 PM solarwind: The "signal period" is 20µm, but what does that mean exactly? The DRO is letting me see steps down to 0.1µm which is insane
02:58 PM solarwind: and it seems to be a real 0.1µm "resolution"
02:59 PM solarwind: so what does the glass actually look like? Is it an "analog" mechanism or a digital one like quadrature encoders with discrete slots?
03:04 PM Tom_L: https://www.heidenhain.us/resources-and-news/types-of-encoders-and-applications/
03:07 PM solarwind: So why output a sine wave?
03:07 PM Tom_L: phase differential?
03:07 PM solarwind: well sine/cos output
03:07 PM solarwind: why not just stick with binary?
03:08 PM Tom_L: go ahead
03:08 PM solarwind: No I mean why did they do that? What's the advantage? You're going digital -> analog -> digital
03:08 PM Tom_L: sine gives infinite position
03:09 PM solarwind: interpolating that doesn't seem like an advantage
03:09 PM solarwind: I don't know what the real resolution is
03:09 PM Tom_L: look how a resolver works
03:09 PM solarwind: a resolver is different
03:10 PM solarwind: it's fundamentally analog
03:10 PM solarwind: optical on/off is binary
03:12 PM CaptHindsight: it would be difficult to etch 100nm lines onto glass scales
03:13 PM solarwind: You don't _need_ to
03:13 PM solarwind: why have the read head do the analog modulation?
03:13 PM solarwind: You can do it much better in the DRO itself
03:14 PM CaptHindsight: https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcQATVxwcDXsUChGZnS5gEI9n7srOt1niI8lkrkzBt3HT5aSBx6vQM9xOXGWAnJJNnZV_YM&usqp=CAU
03:14 PM CaptHindsight: let me find some good pics
03:16 PM CaptHindsight: https://sci-hub.se/10.1533/9780857099297.1.230
03:16 PM solarwind: Thanks, that looks like a proper scientific paper
03:17 PM solarwind: I'm so sick of amateur blogspam
03:17 PM CaptHindsight: https://www.newport.com/n/diffraction-grating-physics
03:19 PM CaptHindsight: https://www.newport.com/n/the-grating-equation
03:21 PM CaptHindsight: https://www.renishaw.com/en/how-renishaw-optical-encoders-work--36979
03:21 PM CaptHindsight: for comparisons between types ^^
03:23 PM solarwind: excellent, thanks
03:23 PM CaptHindsight: https://www.mdpi.com/1424-8220/23/4/1987
03:23 PM solarwind: I don't know how you find this stuff
03:23 PM CaptHindsight: this one shows the chemical etching process
03:23 PM CaptHindsight: grasshopper "how do you not?"
03:24 PM CaptHindsight: I spend 1-20 hours a day doing research, so I have Google pretty well trained
03:26 PM CaptHindsight: 99% is tech, 1% is cats and silly stuff
03:45 PM pcw-home: Almost all high resolution encoders use interpolation of some type
03:50 PM pcw-home: For simple quadrature it gets impractical to get the detector grating close enough to the wheel/scale in increase the line count beyond a certain point
03:51 PM pcw-home: Heidenhain typically have 5X or > interpolation
04:07 PM Connor: pcw-home: Question with Mesa 7i76e, what's the max steps in can do? I.E. I had my ClearPATH setup for a rather large step per revolution, and had to drop it down because LinuxCNC the servo would rebound.. Go past the mark and go back the other way. Not sure how else to describe it. Was really strange.
04:10 PM JT-Shop: nice ride today
04:12 PM JT-Shop: Connor, I thought I had a tool tip to say what the units is but somehow I placed it in status tip
04:12 PM JT-Shop: I'll change that in the morning
04:12 PM Connor: JT-Shop, I wasn't seeing it. It was at the very bottom.
04:13 PM JT-Shop: yup I just looked and it's in the wrong place
04:13 PM Connor: Ahh. ok
04:13 PM JT-Shop: it's not in your focus but should be
04:13 PM JT-Shop: thanks for the feedback
04:14 PM Connor: Yea, coming from PNPConf vs MesaCT I was just plugging in the same values.. So, I used Inches per Minute for velocity, instead of Inches per Second.
04:14 PM Roguish: pcw-home..... what's the best rt kernel now? for master and bookworm
04:24 PM pcw-home: I think the included 6.1 one is OK. I just tried the 2.9.1 ISO on a slowish (Intel 3150) CPU with Realtek MAC and it was fine
04:25 PM JT-Shop: is it possible to port the code for the 5i25T back to 2.9?
05:11 PM JT-Shop: is there a short way to say "per second per second" for acceleration?
05:12 PM Tom_L: per 2nd sq
05:12 PM JT-Shop: hmm
05:13 PM Tom_L: per 2nd/sq
05:13 PM Tom_L: c8 or bluewing ride?
05:13 PM Tom_L: i assume C8
05:14 PM JT-Shop: mm/sec.2
05:14 PM JT-Shop: today C8
05:14 PM XXCoder: ^2 is better
05:14 PM XXCoder: mmm/s^2
05:14 PM XXCoder: -m
05:14 PM JT-Shop: right
05:15 PM JT-Shop: mm/s^2 or in/s^2
05:15 PM JT-Shop: that will work
05:15 PM XXCoder: yeah standard is good
05:15 PM JT-Shop: going to add place holder text in acceleration and velocity line edits
05:23 PM Tom_L: yeah, sq is more for woodworkers
05:25 PM JT-Shop: at least give a hint before they start typing
05:36 PM JT-Shop: the place holder text has to change based on the units selected :)
05:38 PM Tom_L: maybe i opened a can of validation worms here
05:39 PM JT-Shop: Connor, opened that can
05:39 PM Tom_L: ahh
05:48 PM JT-Shop: and that's good to know that improvements are needed
05:52 PM * JT-Shop calls it a night
05:53 PM Tom_L: dark enough for sure
06:07 PM Tom_L: trying to see how you would calculate for a rotary axis with the calc
06:30 PM pcw-home: "is it possible to port the code for the 5i25T back to 2.9?" yes, but not sure if the release manager would approve since its not bug fix ( though it's not very invasive, only adding a bit of code to recognize the card )
07:30 PM Connor: Why isn't it in 2.9?
07:31 PM Connor: and yea, I kinda caused can of worms on mesact :) I was plugging in the same exact values from pnpconf into it, only to find that velocity is in units per second vs units per minute like in pnpconf. which caused things to be really strange.
08:07 PM LWK: i though the 5i25T was in 2.9.1 stable release but not in the debain linuxcnc-uspace (2.9.0~pre1)
08:12 PM pcw-home: possibly If someone other I merged it
08:13 PM XXCoder: andy video. a rarity
08:15 PM pcw-home: LWK you are right, I did merge 5I25T into 2.9, just forgot
08:17 PM pcw-home: Done Sept 21, shows how bad my memory is...
09:01 PM roycroft: what did we do in the olden days?
09:01 PM * roycroft cannot remember
09:02 PM roycroft: i was doing some work on my hifi power amplifier, replacing some capacitors, and found that an insulator was missing
09:02 PM roycroft: this amp is 50ish years old, and most parts are not available
09:03 PM roycroft: but 15 minutes later i had fired up my plastic squirter and made a replacement the fit perfctly
09:04 PM roycroft: i'm think that if i could not have done that i'd have had to find a piece of plastic and spend a couple hours carving it up to shape, and it would have looked like crap
09:04 PM roycroft: not that a plastic squirter makes really attractive parts
09:04 PM roycroft: but it's better than what i'd have whittled
09:19 PM Tom_L: were the originals bakelite?
09:21 PM roycroft: no
09:22 PM roycroft: the heat sink mounts were, but these insulators were not
09:22 PM roycroft: they were probably abs
09:24 PM roycroft: although they are a bit shinier than abs usually looks