#linuxcnc-devel | Logs for 2014-06-05

[08:36:39] <cradek> argh, he cross-posted
[08:36:47] <cradek> well at least I didn't respond to both lists
[08:36:51] <cradek> that's obnoxious
[08:44:11] <pcw_home> Not going to have a lot of luck driving a step/dir drive with PWM
[08:48:09] <cradek> well yeah, but explaining that could clog the devel list for days
[08:48:32] <pcw_home> Im answering on users list
[08:48:44] <cradek> my hero
[08:50:53] <cradek> he's not going to be successful reading a 2400ppr encoder with a parport
[08:51:36] <archivist> beyond effin slow rotation rate
[08:52:11] <archivist> just the magnetics of a step will probably exceed that :)
[08:56:47] <pcw_home> well maybe up to 300 RPM or so
[08:57:29] <pcw_home> I guess I woudl also set the microstepping up as high as it can go
[08:58:23] <pcw_home> so the PID can (somewhat lumpily) pull the motor around
[08:58:59] <cradek> I'd back up and reconsider what I was doing and why...
[08:59:07] <cradek> I'm not sure what his goal is
[08:59:17] <pcw_home> Well theres that...
[08:59:20] <cradek> then next, I'd buy appropriate hardware
[09:00:37] <pcw_home> I think this kind of closed loop has some usefulness if you have a linear scale
[09:01:02] <cradek> yeah
[09:01:09] <cradek> but not with a parport
[09:01:48] <cradek> I guess index homing is nice too, which you could get with rotary encoders
[09:02:29] <skunkworks_> lots of linear scales have index pulses...
[09:02:48] <skunkworks_> sometimes at each end..
[09:03:25] <pcw_home> I think someone before had this same setup, a rotary table with a step motor and high res encoder
[09:15:53] <skunkworks_> was I supposed to do anything more with the ja4-cba4?
[09:16:31] <skunkworks_> seems to me I was supposed to test something
[09:33:13] <pcw_home> Crap didnt fix the reply -to so the stepgen mail went to the developers list
[09:47:34] <cradek> pcw_home: weird. I get a Reply-To: header that lists his private email and the emc-users list. I wonder if that is a user preference setting for the list?
[09:48:07] <cradek> if you have that Reply-To: but your mailer used the From: instead, it's busted
[09:50:37] <cradek> wait, it couldn't have been From:
[09:50:47] <cradek> I guess I don't know what your mailer did...
[09:54:28] <pcw_home> His email header looked like this
[09:54:30] <pcw_home> Date: Thu, 5 Jun 2014 01:47:43 -0700 (PDT)
[09:54:31] <pcw_home> From: ELHIMA Moustapha <melhima@yahoo.com>
[09:54:33] <pcw_home> To: "emc-users@lists.sourceforge.net" <emc-users@lists.sourceforge.net>,
[09:54:34] <pcw_home> emc2 <emc-developers@lists.sourceforge.net>
[09:54:36] <pcw_home> Subject: [Emc-users] Step motor in closed loop
[09:54:49] <pcw_home> no reply to
[09:55:20] <cradek> are you sure no reply-to? why would you have generated a reply to the lists at all then?
[09:55:33] <pcw_home> I just should have been more careful before i hit send
[09:55:55] <pcw_home> I dont know, normal list mails have the expected reply-to field
[09:56:12] <cradek> very odd
[09:56:25] <pcw_home> like:
[09:56:27] <pcw_home> Date: Thu, 05 Jun 2014 11:14:58 +0300
[09:56:28] <pcw_home> From: Marius Alksnys <marius.alksnys@gmail.com>
[09:56:30] <pcw_home> Reply-To: "Enhanced Machine Controller (EMC)"
[09:56:31] <pcw_home> <emc-users@lists.sourceforge.net>
[09:56:33] <pcw_home> To: emc-users@lists.sourceforge.net
[09:56:34] <pcw_home> Subject: Re: [Emc-users] How to use Mesa 7i90HD ?
[09:57:11] <cradek> your mailer must be (sometimes) hiding headers from you?
[09:57:35] <cradek> what mailer is it? (I see it doesn't generate a X-Mailer header)
[09:57:43] <pcw_home> pine
[09:58:24] <cradek> huh, I'd expect pine to really work right...
[10:01:12] <archivist> I only see one incoming from that user, I see duplicates from other lists, is the list being "clever"
[10:05:51] <cradek> andy's bike engine started!
[10:07:33] <cradek> http://bodgesoc.blogspot.com/
[10:08:06] <pcw_home> I saw that, He must be excited
[10:08:58] <cradek> cool coil winder setup. but I am surprised the inertia of that big roll doesn't break the wire. maybe there's a trick to it I can't see in the video.
[10:10:49] <archivist> there should be a tension device in between
[10:11:36] <archivist> better not mention the coil winder out in the garden unloved
[10:16:02] <pcw_home> normally theres a spring arm with a pully/or felt for small wire, maybe with enough length the wire stretch helps
[10:17:10] <archivist> also one often pulls off the end of the reel rather than rotate it specially with smaller wire sizes
[10:17:25] <pcw_home> * pcw had a job winding transformers for a while a long long time ago
[10:18:20] <archivist> we got some stuff ex transformer co when it shut down, and one of the worker came to us to be a clock maker
[10:19:00] <archivist> cradek, did you ever get the Bulova coils rewound
[10:19:21] <pcw_home> I hated the big stuff (hand guided and bruised my thumbs)
[10:20:48] <archivist> it is a fight between you and the wire :)
[10:21:08] <cradek> archivist: nope.
[10:21:57] <skunkworks> cradek, did you ever score small enough wire?
[10:22:35] <archivist> I remember searching all we won from the tx co
[10:23:29] <cradek> no, that is sure the main problem.
[10:24:08] <archivist> perhaps time to use a larger wire size to see if you can get away with it
[10:24:15] <cradek> the rest would be the "fun" part...
[10:25:12] <cradek> I truly doubt that would work. They didn't use the small wire for fun.
[10:26:54] <archivist> if you can get the right number of turns on I think it should
[10:27:23] <archivist> best of luck counting the old ones off though
[10:28:16] <skunkworks> same wire size - fill it up same amount ;)
[10:28:27] <skunkworks> logger[psha],
[10:28:27] <logger[psha]> skunkworks: Log stored at http://psha.org.ru/irc/%23linuxcnc-devel/2014-06-05.html
[10:28:42] <archivist> turns ratio matters with coils
[10:29:14] <seb_kuzminsky> another openssl vulnerability, everybody upgrade again
[10:29:15] <skunkworks> what does this coil do?
[10:29:34] <mozmck> how small is the wire?
[10:30:15] <cradek> it is #54
[10:30:17] <archivist> the tuning fork drive coil which is also part of the oscillator
[10:30:45] <mozmck> I'll see what I have laying around. I guess it's coated copper?
[10:31:03] <cradek> archivist: there's a 4th power in effect - if you double the diameter of the wire, a full bobbin gives 1/16 the resistance
[10:31:41] <cradek> mozmck: yes it appears to be copper coated with something
[10:31:48] <mozmck> ok
[10:32:50] <cradek> jmk found a manufacturer who said 54AWG is $3500/lb but it's extremely hard to find at all, he didn't know of any in the US
[10:33:08] <archivist> I was only thinking going up in size by a few percent
[10:33:32] <cradek> #50 can be found with not much trouble
[10:35:16] <cradek> I guess #50 and smaller are so fragile you probably won't find someone to respool some for you - you have to buy whatever is on the roll they have from the manufacturer
[10:35:34] <skunkworks> you can get nichrome in that gauge... ;)
[10:35:42] <mozmck> ok, probably don't have it, but we have some odd stuff from surplus
[10:39:14] <cradek> hmmm http://www.planetengineers.com/product.asp?pid=1235
[10:40:34] <skunkworks> yikes
[10:43:18] <mozmck> ouch!
[10:43:41] <mozmck> Dad suggested getting some out of speakers.
[10:43:44] <skunkworks_> seems like something someone would have laying around...
[10:44:32] <archivist> speaker wire is much larger
[10:44:46] <mozmck> all of them?
[10:44:52] <archivist> this is really fine tiny stuff
[10:44:59] <skunkworks_> .016mm
[10:45:16] <cradek> this is smaller than you think, and then still smaller than you think after you think the first time
[10:45:28] <skunkworks_> ^that is mm... ;)
[10:45:32] <archivist> I have even taken a clock motor to bits and measured
[10:45:48] <cradek> so .0006 inches diameter
[10:46:15] <skunkworks_> (and that was rounded up...)
[10:46:21] <archivist> effin small
[10:47:17] <pcw_home> I thought my 1 mill Tungsten was small
[10:47:18] <mozmck> wow!
[10:48:21] <skunkworks_> breath on it and it breaks
[10:50:39] <mozmck> how would you wind something like that?
[10:52:21] <cradek> well it's even worse than you think because the coil form is in the way, so you can't just spin it like on andy's setup. the wire has to come in endwise and turn the corner.
[10:52:38] <mozmck> what is the coil for?
[10:52:55] <cradek> it keeps the tuning fork vibrating in a wristwatch
[10:53:14] <cradek> magnets are attached to the ends of the tuning fork, and they go inside the coil
[10:53:20] <cradek> coils (two of them)
[10:53:40] <cradek> http://www.amw.at/Accutron/214id.jpg
[10:54:04] <skunkworks> so why do they fail?
[10:54:32] <cradek> I'm sure a servo-driven spool unwinding would be part of the solution
[10:55:09] <cradek> skunkworks: unknown. some have obvious damage (moisture, or cracked shellac) but some will fail when you drop them. they're 50+ years old.
[10:55:28] <cradek> in another 50 years they'll probably all be dead
[10:55:32] <skunkworks> ah. It looks the the coils are on spools - do the spools come out?
[10:55:39] <cradek> nope the green part is one piece
[10:55:52] <cradek> well there's a green part on each side of the watch, connected by a few wires
[10:55:54] <skunkworks> ah
[10:55:59] <cradek> each side is one chunk of plastic
[10:56:21] <mozmck> interesting.
[10:57:52] <skunkworks> would there be a market for fixing them? (for the 2k investment in wire....)
[10:59:39] <skunkworks> how long did they last on a battery?
[11:02:28] <skunkworks> cradek, it does say that you can buy the wire in different sizes.. .6lbs seems like quite a bit.. I wonder if you can get an oz or less
[11:05:06] <skunkworks> https://oldfathertime.com/accutron_repair.htm
[11:05:16] <skunkworks> Please note that coils are in very short supply. For that reason, we only install them in watches receiving a "Full Restoration". Luckily coils are needed in only a small fraction of the Accutrons we receive for repair.
[11:05:26] <skunkworks> looks like they don't even re-wind them..
[11:06:54] <cradek> skunkworks: about a year
[11:07:57] <cradek> skunkworks: I'm pretty certain nobody rewinds them. legend has it, the original machinery that made them was destroyed by bulova in a weird PR blitz
[11:09:01] <cradek> the market is hard to guess. they're definitely an interesting watch to collectors.
[11:09:21] <skunkworks> father time seems to be full of themselfs..
[11:10:32] <pcw_home> Drive a 50 Ga coil with a short pulse...
[11:12:52] <cradek> skunkworks: I've never dealt with them. it would be fun to sell them coils. I am sure they need them.
[11:13:52] <skunkworks> they say they are in 'short supply'
[11:13:56] <pcw_home> what type of plastic are they wound on?
[11:14:26] <cradek> pcw_home: I don't know...
[11:16:07] <skunkworks> swelled?
[11:22:20] <cradek> http://img.timezone.com/img/articles/horologium631672882451976629/ElectronicsSectionBoth.jpg
[11:22:52] <cradek> one side has two separate windings, a sense and a drive
[11:22:59] <cradek> the other just drives
[11:25:13] <skunkworks> wow - that looks like a lot of turns...
[11:25:27] <cradek> in 1960 the transistor in these watches was often the first one regular people owned
[11:26:54] <skunkworks> it is suprisingly small
[11:28:42] <cradek> skunkworks: someone on the internet says 8000 turns
[11:29:02] <skunkworks> if you spin the watch bits - you could use a small tube to guide the wire past the lumpy parts..
[11:29:14] <cradek> yep
[11:29:23] <kwallace> BTW, here is my version: http://wallacecompany.com/tmp/IMG_2155-1a.JPG
[11:29:38] <cradek> kwallace: ooh that's a nice one
[11:30:11] <cradek> kwallace: I have only one accutron clock, but it's fugly...
[11:30:25] <cradek> kwallace: does it work?
[11:30:56] <cradek> (they must have balanced that huge second hand very carefully...)
[11:32:28] <kwallace> My dad was into Accutrons when they first came out. This thread reminded me that I had the clock. It doesn't have a battery in it so I don't know if it works.
[11:34:39] <kwallace> One battery contact has "124X" printed on it. I wonder if that is a battery type?
[11:34:48] <pcw_home> Silicon transistor?
[11:35:10] <cradek> kwallace: I bet 214, not 124?
[11:35:19] <cradek> kwallace: correct modern battery is 387s
[11:35:53] <cradek> kwallace: (the original batteries did say 214 on them)
[11:38:52] <cradek> pcw_home: originally Ge PNP, later Si NPN
[11:39:44] <kwallace> It's "124X". Thanks for the 387s, I'll order one.
[11:40:19] <cradek> kwallace: huh, I don't know what 124X would mean. what is the date code at the bottom of the back of the case? (M or N followed by a number)
[11:46:07] <cradek> while the 214 is the venerable original, I think my favorite model is the 221, which had worm gears (!) to make the train tiny, and buzzes at 440Hz, which is concert pitch
[11:46:13] <cradek> http://electric-watches.co.uk/wp/wp-content/uploads/Bulova-Accutron-2210-2.jpg
[11:46:42] <cradek> you can see a worm in this photo
[11:47:37] <cradek> these were mostly in women's watches but a few men's models, unfortunately they have no second hand
[11:48:55] <kwallace> http://wallacecompany.com/tmp/IMG_2159-1a.JPG http://wallacecompany.com/tmp/IMG_2159-2a.JPG
[11:49:49] <cradek> cool, you have the plastic ring from your old battery. sometimes the 387s is hard to find, so it's good to keep those.
[11:50:29] <cradek> maybe the clocks don't have the date code - I don't see it - it would be under BULOVA on the back, or at the bottom opposite it
[11:51:53] <kwallace> It looks like there was at one time a 1" x 1.5" label on the back.
[11:53:26] <cradek> date code, N3 = 1973: http://www.hknebel.de/Uhren/Bulova22102B.jpg
[12:00:02] <archivist> did they go without the seconds hand due to drive freq not being related to seconds
[12:05:13] <cradek> I think it was just impossible with the worm gear reduction
[12:07:49] <cradek> I bet it was hard enough to get any hands on it - there's nothing in the center of the movement except battery
[12:13:21] <kwallace> It looks like I should add a series diode to the battery.
[12:14:10] <cradek> not unless you have trouble that correct setup doesn't fix
[12:16:33] <kwallace> It just seems to be safer to correct the battery voltage than to monkey with the movement.
[12:17:25] <cradek> a correctly set up (phased) indexing mechanism is maximally tolerant to amplitude variation. it's not really monkeying...
[12:17:38] <cradek> the vast majority have no problem tolerating the amplitude with a modern cell
[12:18:04] <cradek> and for a desk clock you'll need less headroom because you won't be banging it on stuff
[12:18:25] <kwallace> Okay, I'll try a naked battery first.
[12:25:12] <archivist> I am monkeying with a mechanical clock at the moment, got the escapement running on a lot less weight but the strike is being a bit more intransigent
[13:00:55] <thews-work> steinkuehler?
[13:23:28] <skunkworks> I have not seen him on irc...
[13:39:29] <kwallace2> This one is not that far from me: http://www.ebay.com/itm/261324488434 Too bad there is no room in my shop.
[13:40:13] <cradek> you must have two square feet free...
[13:42:06] <kwallace2> There is barely room to walk. Plus I am on a gadget diet.
[13:45:19] <kwallace2> Which reminds me, I have a friend trying to get rid of a nice 10" x 30" engine lathe, which I plan to put out side for now.
[13:50:23] <skunkworks> awww - look at the cute little shaper...
[13:52:09] <archivist> dont remember seeing them this side of the pond, the lathes came over in qty though
[14:46:19] <cradek> pcw_home: aaaaand he responded to you on -devel
[18:01:59] <kwallace2> Would Robert Ellenberg happen to be here?
[18:18:03] <andypugh> If I fiddle with the UART bit fo Hostmot2, should I do it in the 7i90 branch or as a bugfix in 2.5? or in Master?
[18:18:19] <andypugh> I don’t think I can actually test it.
[18:18:49] <andypugh> Which points at 7i90 so I fix a problem for a user.
[18:18:49] <skunkworks_> kwallace2: no.. best would be to ask a question on the list...
[18:19:00] <andypugh> But it really belongs in 2,6
[18:19:02] <skunkworks_> kwallace2: do you see an issue?
[18:20:00] <kwallace2> skunkworks_ , Thank you. I just wanted to ask him a question on git.
[18:20:54] <skunkworks_> heh - I could probably help...
[18:20:59] <skunkworks_> or anyone on here...
[18:26:51] <kwallace2> I'm having fits with using a submodule. I tried a commit in the subm and I don't see it in git log. I have the edits archived so I haven't lost anything. I just haven't taken the time to get proficient with git.
[18:28:39] <andypugh> cradek: seb_kuzminsky Where would you like to see a change to Hostmot2 UART checking made?
[18:28:49] <kwallace2> Hello andypugh, I enjoyed your magneto test video.
[18:29:18] <andypugh> It was very loud.
[18:29:46] <kwallace2> The muffler isn't done yet?
[18:34:14] <seb_kuzminsky> hi andypugh, nice progress on the ne'er-a-bike
[18:34:43] <andypugh> kwallace2: No, during the test the exhaust port was completely open.
[18:34:59] <andypugh> I have a silence made, but it hangs under the engine.
[18:36:23] <seb_kuzminsky> the uart part of hostmot2 isn't unique to the 7i90, right? it could be found any anyio board? if so the fix belongs in 2.6
[18:36:36] <andypugh> OK.
[18:36:58] <andypugh> It will get tested sooner in 7i90 is all
[18:37:21] <seb_kuzminsky> hmm, right
[18:37:48] <seb_kuzminsky> how about this: (this may be too much git-fu, but it seems sensible to me)
[18:37:55] <seb_kuzminsky> you do the uart fix on a feature branch off 2.6
[18:38:06] <seb_kuzminsky> i rebase the 7i90 branch on top of 2.6
[18:38:25] <seb_kuzminsky> we merge those feature branches in an throw-away integration branch, which the user can test
[18:38:31] <andypugh> OK
[18:38:38] <seb_kuzminsky> when your uart fix works, we merge it into 2.6
[18:38:59] <seb_kuzminsky> the 7i90 driver will go in master when it's ready (it messes with the 7i43 driver in a way i dont feel comfortable with for 2.6)
[18:42:09] <micges> seb_kuzminsky: 7i90 driver is working, what is the problem that it wasn't merger to master after fest?
[18:48:06] <seb_kuzminsky> the 7i90 driver that kim and i wrote at fest last summer was basically a copy of the 7i43 driver with some minor tweaks
[18:48:25] <seb_kuzminsky> there was much overlap between the 7i90 driver and the 7i43 driver, mostly in the epp driver code
[18:48:36] <seb_kuzminsky> i wanted to pull that out of both drivers and make it shared between them
[18:49:01] <seb_kuzminsky> i did that and requested a code review, and got some feedback that i haven't addressed yet
[18:49:16] <seb_kuzminsky> there's not much work remaining to be done on it, i just havent found the time yet
[18:51:47] <micges> maybe you could merge just 7i90 driver without shared epp code to 2.6?
[18:52:08] <seb_kuzminsky> that's a good idea
[18:58:02] <andypugh> pcw_home: What is the actual UART Instance stride? 0x40?
[18:59:22] <andypugh> Regmap says 0x10.
[18:59:57] <andypugh> So, it actually is 0x10 but IDROM states 0x40? That’s the problem?
[19:05:00] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.6 9078c06 06linuxcnc 10docs/html/gcode.html 10docs/src/gcode/gcode.txt Merge remote-tracking branch 'origin/v2.5_branch' into 2.6 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=9078c06
[19:11:04] <KGB-linuxcnc> 03andy pugh 05uart_step 2f3ce1f 06linuxcnc 10src/hal/drivers/mesa-hostmot2/uart.c The Mesa IDROM format only supports a limited number of instance strides. * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2f3ce1f
[19:11:44] <andypugh> It’s almost the smallest possible patch, when I looked at how to do it.
[19:13:02] <seb_kuzminsky> weird
[19:13:14] <seb_kuzminsky> why do sserial and uart clash? (i dont know how either subsystem works)
[19:14:11] <andypugh> The IDROM format says that each module is either (effectively) a “big stride” or a “small stride” as I understand it.
[19:15:01] <andypugh> And the two strides are defined separately. But there are only two.
[19:15:07] <andypugh> This is my understanding anyway.
[19:43:43] <andypugh> Night all
[20:00:34] <KGB-linuxcnc> 03Kim Kirwan 052.6 c7bd914 06linuxcnc 10(8 files in 5 dirs) add driver for the 7i90 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c7bd914
[20:00:42] <seb_kuzminsky> i shoulda probably pushed that about a year ago
[20:00:51] <seb_kuzminsky> thanks for prodding me, micges
[20:00:54] <seb_kuzminsky> oh, he left
[20:20:51] <KGB-linuxcnc> 05hm2-7i90 420171e 06linuxcnc 04. branch deleted * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=420171e
[20:21:56] <KGB-linuxcnc> 03Sebastian Kuzminsky 05master 478931a 06linuxcnc Merge remote-tracking branch 'origin/2.6' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=478931a
[22:52:56] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hm2-epp c20d9e3 06linuxcnc 10src/Makefile 10src/Makefile.inc.in 03src/hal/drivers/epp.c 03src/hal/drivers/epp.h add epp driver * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c20d9e3
[22:52:56] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hm2-epp 8bc0445 06linuxcnc 10configs/by_interface/mesa/hm2-servo/hm2-servo.hal 10configs/by_interface/mesa/hm2-stepper/hm2-stepper.hal 10src/hal/drivers/mesa-hostmot2/hm2_7i43.c 10src/hal/drivers/mesa-hostmot2/hm2_7i43.h hm2_7i43: use the new shared epp driver * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8bc0445
[22:52:56] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hm2-epp 16db57b 06linuxcnc 10src/hal/drivers/mesa-hostmot2/hm2_7i90.c 10src/hal/drivers/mesa-hostmot2/hm2_7i90.h hm2_7i90: use the new shared epp driver * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=16db57b
[22:53:01] <KGB-linuxcnc> 05hm2-7i90-3 462f809 06linuxcnc 04. branch deleted * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=462f809