#linuxcnc-devel Logs

Jan 29 2018

12:27 AM seb_kuzminsky: ffs
09:32 AM cradek: skunkworks_: lowest friction lowest inertia kit encoder ever: http://timeguy.com/encoder.jpg (running on 0.25mm pivots in watch jewels)
09:33 AM cradek: it always settles with the setscrew on the encoder wheel pointing down
09:33 AM cradek: which is fine, it doesn't have to be in poise, since it will only turn a little bit
09:34 AM skunkworks_: very cool! is that one of the capacitive encoders?
09:35 AM cradek: no it's optical
09:35 AM skunkworks_: Good.. :)
09:35 AM cradek: heh yeah
09:35 AM skunkworks_: how does that fit into the de-spooler?
09:36 AM cradek: it's part of the tension measurer
09:37 AM cradek: winding tension is about 1 cN, breaking force is about 2cN
09:45 AM skunkworks_: that is just crazy
09:46 AM seb_kuzminsky: how will you measure the tension?
09:50 AM cradek: 1cN is about 1 gram in gravity, so I think the machine just has to keep a 1g weight on the wire, and this will read its position (use pid to keep it in place)
09:51 AM cradek: I can't decide whether the unspooling motor needs a separate encoder
10:02 AM skunkworks_: I see this all being done in hal
10:02 AM cradek: yep
10:49 AM KimK: cradek: I agree with skunkworks, very cool! Is that a US Digital encoder? Which series/model? What shaft diameter and length did you use? How many lines or counts?
10:52 AM cradek: it's 500 lines I think. It's one of the HEDS models, I swiped it off something.
10:54 AM cradek: the wheel has a 1/4 inch hole, so I had to make a bushing to adapt it to the new 1.2mm shaft (aluminum disc that's about as thick as the wheel's set screw diameter)
10:54 AM cradek: I wanted to add as little mass as possible
10:56 AM cradek: the jewels are friction fit plate jewels, super easy to use, ream a hole undersize, press 'em in
10:58 AM cradek: the shaft is just blue steel wire, you can get it in various diameters from any watch parts house, ends turned down to square pivots
10:59 AM cradek: that extra length on the left is where the not-yet-fully-designed mechanism will attach...
11:03 AM cradek: https://www.esslinger.com/staff-and-pivot-watch-and-clock-wire-rod-assortment-blue-steel-55-1-93mm-12-pieces/
11:03 AM cradek: super useful to have around
06:22 PM skunkworks: seb_kuzminsky, you mentioned using hal_float_t for in/out/avgarray/avgtotal https://pastebin.com/QhdMzYrX
06:22 PM skunkworks: I don't see any reference to those in the docs
06:26 PM seb_kuzminsky: skunkworks: it's in hal.h, and should be used anywhere you need a floating-point type in hal
06:29 PM seb_kuzminsky: it's documented in the "hal_type_t" manpage, but that appears to be missing from our debs?
06:30 PM seb_kuzminsky: or maybe it's in the -dev package?
06:30 PM seb_kuzminsky: ah yes, it's in linuxcnc-dev/linuxcnc-uspace-dev
06:30 PM seb_kuzminsky: which even kind of makes sense
06:31 PM seb_kuzminsky: i see that the halcompile/comp docs incorrectly use float
06:33 PM skunkworks: lots of components use it also'
06:35 PM seb_kuzminsky: yeah :-/
07:00 PM cradek: I have a brass rod .81mm diameter. I want 1 gram of brass. What length do I cut? My number feels wrong.
07:05 PM jepler: .81mm? What's that, 20ga wire?
07:05 PM cradek: I think it's 1/32 in rod
07:05 PM cradek: but yeah it's a lot like a wire
07:06 PM jepler: I get around 4.5g/meter and that you should take 222mm to get around 1g
07:06 PM jepler: You have: (.81mm/2) ^ 2 * pi * (8.73 g/cm3)
07:06 PM jepler: You want: g/m
07:06 PM jepler: * 4.4985667
07:06 PM jepler: You have: gram / ((.81mm/2) ^ 2 * pi) / (8.73 g/cm3)
07:06 PM jepler: You want: mm
07:06 PM jepler: * 222.29303
07:07 PM cradek: thanks for checking me
07:07 PM cradek: a gram is more than I thought
07:20 PM cradek: http://timeguy.com/1gram.jpg
07:21 PM cradek: I will have to consider the angle, because this is more tension than if it was pulling straight down
07:24 PM cradek: pushing this wire through a .09mm jewel hole is much much harder than threading a needle
09:09 PM jepler: I think I see a wire in that picture..
09:11 PM cradek: in person it looks like spider silk, but if you get just the right reflection you can see the copper color
09:13 PM jepler: this prompts me to wonder if you can spritz it with something (deionized water, pure alcohol) that would form droplets, to see it better..
09:14 PM cradek: yeah, it might work just like spider silk
09:19 PM cradek: did you see http://timeguy.com/1meter-0_015.jpg ?
09:20 PM cradek: resistance of 1m of wire shows the diameter is correct (first thing I checked)
09:31 PM jepler: no, I missed it. glad it's right!
09:33 PM jepler: hm, seems like you have to hope the enamel is no thicker than in the original wire too, or you blow your available volume for the winding that way
09:35 PM cradek: yeah, there are a lot of unknowns
09:35 PM cradek: I'm nowhere near trying one yet
09:36 PM cradek: I bet the enamel is both thinner and better than the original
09:36 PM jepler: how'd the soldering go for that test piece? any trouble?
09:36 PM cradek: it was hard to get the enamel to melt. I had to turn the iron up most of the way
09:37 PM cradek: I'm not sure how I should be unenameling it
09:37 PM cradek: I tried a flame but the wire kinda just disappears
09:37 PM cradek: a few times I got it to turn red and not disappear, and that gave a clean looking wire
09:38 PM cradek: it would be sad to wind it and then burn the end up making it too short to terminate
09:44 PM Tom_itx: chemical melt?
09:44 PM Tom_itx: tolulene or something
09:44 PM jepler: hmm this source mentions spot welding, solvents, and silver metal bath as methods to get through the enamel https://www.hitachi-metals.co.jp/e/products/auto/el/pdf/MagnetWire_en.pdf
09:44 PM jepler: not specific about solvents, it would seem to depend on the actual coating/enamel used
09:45 PM cradek: I think it's polyeurethane
09:46 PM cradek: 3UEW/155, fwiw
09:46 PM cradek: 155(C) is I think max temperature
09:49 PM jepler: seems most hydrocarbon flames are well higher than the melting temperature of copper https://en.wikipedia.org/wiki/Adiabatic_flame_temperature#Common_flame_temperatures
09:50 PM jepler: dip solder pots would be 220-260 for lead-tin alloys and 350-400 for lead-free. that should take the coating right off without damaging the copper.
09:51 PM cradek: so there might also be a setting of my temperature-controlled soldering iron that works?
09:52 PM cradek: if so I could use one setting to unenamel and a lower setting to actually solder on the plastic forms
09:53 PM jepler: https://www.amazon.com/Yescom-Capacity-Soldering-Desoldering-Welding/dp/B00RLDQCC4/ can't possibly be a good product at that price
09:54 PM cradek: the internet says the plastic is diallyl phthalate, but the internet also says that melts at -70C
09:55 PM cradek: jepler: I bet the temperature regulation on that device is not very good, to say the least
09:57 PM jepler: inexplicably the second page of results on amazon includes the book "The Republic (Penguin Classics)" by Plato
09:57 PM jepler: what a world
10:00 PM cradek: g'night
10:01 PM jepler: see you
10:06 PM jepler: what's this using, a coil stolen from an electric stove? https://hackaday.com/2009/06/26/home-made-solder-pot/