#emc-devel | Logs for 2007-08-15

[00:24:54] <jmkasunich_> jmkasunich_ is now known as jmkasunich
[00:29:54] <jmkasunich> cradek: a little off the top, eh?
[02:07:10] <cradek> the mill is reassembled and is working!
[02:07:24] <cradek> well, as well as it ever did, I think
[02:08:10] <cradek> something seems fragile - I keep getting error 4000 (?) and if I try to MDI it sometimes crashes (junk on the screen) but maybe that's when I forgot to enter a decimal point or something
[02:08:22] <cradek> (no manuals yet, so I'm kind of flying blind)
[02:09:20] <cradek> but this is a great accomplishment - the gearbox is back together, the phase converter works, etc etc
[02:10:43] <steves_logging> Congrats! I will look up the error code. They are a bit mapped.
[02:10:54] <cradek> ah thanks!
[02:12:21] <cradek> I had the Z belt off, so I probably moved its index - wonder if it's hitting a limit sometimes
[02:15:34] <steves_logging> code 4000 maps to "drive fault", stand by while I list the bits MSB to LSB
[02:15:55] <steves_logging> 8000 electronics fault
[02:16:02] <steves_logging> 4000 drive fault
[02:16:12] <steves_logging> 2000 internal comm error
[02:16:27] <steves_logging> 1000 front panel not sneding valid key codes
[02:16:55] <steves_logging> 0800 part program check sum error in stored program
[02:17:11] <steves_logging> 0400 tool table check sum error in stored table
[02:17:18] <steves_logging> 0200 ram failure
[02:17:25] <steves_logging> 0100 rom failure
[02:17:41] <steves_logging> 0080 system fatal error, whatever that is....
[02:18:01] <steves_logging> 0040 communications error
[02:18:18] <steves_logging> 0020 EAF board communications error
[02:18:28] <steves_logging> 0010 commanded move would exceed machine limits
[02:18:53] <steves_logging> 0008 spindle not running for a non-G0 move
[02:19:23] <steves_logging> 0004 cutter comp error #2, indeterminate intersection
[02:19:40] <steves_logging> 0002 cutter comp error #1 no intersection found
[02:19:56] <steves_logging> 0001 program error dound by parser
[02:19:59] <steves_logging> finis
[02:20:15] <cradek> thank you! that is surely the most important data in the book!
[02:20:46] <steves_logging> that and the debug monitor stuff, minimal as it is
[02:21:10] <cradek> I haven't hooked up a terminal yet
[02:21:21] <cradek> (but there is a DB9 cable on it)
[02:21:44] <steves_logging> debug monitor talks to a connector directly off the MC68000 board, no external connector
[02:22:26] <steves_logging> debug monitor only knows world of motion control as seen by MC68000
[02:22:40] <steves_logging> the Z8000 parser offers no such support
[02:23:51] <cradek> I wonder if "drive fault" means my power is inadequate
[02:24:35] <cradek> strange that it rapids fine, but faults when homing or when it's just sitting there
[02:24:45] <steves_logging> Did you say that Z axis belt was not installed? Drive fault usually means that motion was commanded and not detected by encoders.
[02:25:05] <steves_logging> basically following error
[02:25:17] <cradek> no, it's all back together
[02:25:25] <cradek> seems to move smoothly
[02:26:18] <cradek> I need to collect more data before we try to debug it :-)
[02:26:24] <cradek> maybe I can figure out the pattern
[02:26:31] <steves_logging> re-reading your earlier comments, getting "junk" on the display is NOT good
[02:26:41] <cradek> that did happen once
[02:26:51] <cradek> after I did an MDI command
[02:27:02] <cradek> I think it executed the command and then went to the junk
[02:27:19] <steves_logging> well a drive fault should relate to a single axis
[02:28:21] <steves_logging> miss-alignment of the Z axis microswitches if a possibility since that stuff got futzed with
[02:29:54] <cradek> they looked pretty solid, I checked them out when I had that apart
[02:30:07] <cradek> it hits all 3 nicely
[02:30:33] <steves_logging> there is a timing/gearing relationship between the microswitches and the index pulse from the Z axis encoder
[02:31:07] <cradek> I surely screwed that up
[02:31:39] <cradek> I should have marked it somehow but I didn't...
[02:31:54] <cradek> I did think about it as I was taking the belt off though - that counts doesn't it?
[02:32:09] <steves_logging> repeat after me... "my bad, my bad"
[02:32:55] <cradek> but you don't know how minor that seems after all I had apart and after 1000 successful applications of prybar/hammer
[02:33:20] <steves_logging> somewhere in the manual are re-timing instructions that cover X and Y for sure, and for Z one would hope
[02:33:51] <steves_logging> what you resorted to the the bigger hammer approach.... 8-)
[02:34:18] <cradek> none of the pulleys have holes for a gear puller - what a huge pain
[02:34:43] <steves_logging> figures...
[02:35:03] <cradek> in fact most of it was a huge pain - that mill will be in that garage until the end of time
[02:36:06] <steves_logging> FYI, the axis drive pulleys are a possible source of backlash on this machine, it seems the taper lock can become loose
[02:36:33] <cradek> interesting
[02:36:45] <steves_logging> and hard to figure out....
[02:36:48] <cradek> it has quite a bit in X - about .005
[02:37:05] <cradek> I haven't tried to debug that at all yet
[02:37:21] <steves_logging> oops, not good on backlash if it is in the ballscrew
[02:37:56] <steves_logging> there an option, present on everyexample I have seen, for backlash compensation
[02:38:16] <steves_logging> it is a pair of pots connected to an A/D on the MC68000 board
[02:38:20] <cradek> both of his machines had about the same, .003-.005 on X and Y both
[02:38:28] <cradek> I figured it was normal
[02:38:36] <steves_logging> the debug monitor can be used to read the value, and set it if the pots are not fitted
[02:38:47] <cradek> hmm, maybe those are wrong and making it worse?
[02:39:00] <cradek> no, I pulled the table, duh
[02:39:05] <cradek> so it's real
[02:39:13] <cradek> hi jmkasunich
[02:39:50] <cradek> I did see two mystery pots in the bottom section
[02:39:54] <steves_logging> not good news on backlash, the one I mentioned doing a successful bearing bore has less than 0.001"
[02:40:28] <cradek> I should measure it near the end of travel too
[02:41:24] <steves_logging> ball screws tend to wear the balls and the nut more than the screw, so the backlash is more evenly distributed than is typical for an Acme screw
[02:41:45] <cradek> oh, ok
[02:42:11] <steves_logging> bad in that it is everywhere, good in that correction applies more evenly
[02:42:34] <steves_logging> of course EMC can map it
[02:42:55] <steves_logging> BOSS assumes constant backlash correction
[02:44:10] <steves_logging> .... so now that the deed is done, how did you get it through the door?
[02:44:50] <cradek> I took the whole gearbox/transmission apart and removed those three castings, then set the spindle motor on its side
[02:44:57] <jmkasunich> hi
[02:45:16] <cradek> then pulled it up a 1/4" steel ramp using rollers and a comealong anchored to a very conveniently placed huge tree
[02:45:27] <cradek> then, "simply" reassembled it
[02:45:33] <jmkasunich> simply
[02:46:50] <steves_logging> all in a good days work
[02:47:00] <cradek> yep
[02:47:13] <cradek> I started at sun-up. I'm not used to that kind of a day's work
[02:47:26] <steves_logging> now you know how to replace the vari-speed belt when the time comes
[02:47:37] <cradek> yep, no problem
[02:48:14] <cradek> I also see how to mount the spindle encoder (easy)
[02:53:31] <jmkasunich> cradek: does the machine have a quick change spindle?
[02:54:03] <cradek> jmkasunich: yes
[02:54:09] <jmkasunich> nice
[02:54:11] <cradek> it's regular 30 taper
[02:54:18] <jmkasunich> and you have toolholders to match?
[02:54:26] <cradek> yes quite a few, and a very nice collet set
[02:54:46] <cradek> 1/4 to 7/8
[02:55:37] <cradek> and a tap head (that I don't yet know how to use)
[02:55:48] <steves_logging> keep an eye out for 1/8" and 3/16", you may find them usufull
[02:55:53] <steves_logging> usefull
[02:55:58] <cradek> yes 1/8 would be nice
[02:56:01] <jmkasunich> tap heads are nice
[02:56:08] <jmkasunich> and easy to use, if its what I'm thinkning of
[02:56:15] <jmkasunich> makes rigid tapping a breeze
[02:56:23] <jmkasunich> well, not rigid, but thats the point
[02:56:27] <jmkasunich> makes tapping a breeze
[02:56:37] <steves_logging> internally reversing, or just compression/extension
[02:56:48] <cradek> I need a jacobs chuck and whatever crazy tapered things necessary to hook it up
[02:57:02] <cradek> steves_logging: I don't even know that, I just threw it in the box
[02:57:06] <jmkasunich> hold out for a keyless chuck
[02:57:35] <steves_logging> self reversing tapper will have a "reaction arm" or at least a place to install it
[02:57:53] <cradek> this one seems to have no arm
[02:58:05] <jmkasunich> you need to take some photos
[02:58:09] <jmkasunich> machine, tooling, etc
[02:58:21] <cradek> I have a couple
[02:58:44] <cradek> I should have photoed the big move, it was a sight to see, but I was busy moving
[02:58:50] <cradek> brb
[02:58:56] <steves_logging> life goes on...
[03:00:43] <steves_logging> without photos, that is...
[03:03:08] <cradek> http://timeguy.com/cradek/emc/bridgeport
[03:04:44] <jepler> cradek: does it have a name yet?
[03:05:09] <cradek> if so, it would be "R2 unit"
[03:07:29] <steves_logging> so... you did manage to "split" the head, great
[03:07:49] <cradek> yeah I kept splitting until it was short enough
[03:07:57] <cradek> each piece could be lifted by two people, which was nice
[03:09:15] <jmkasunich> nice machine
[03:09:32] <jmkasunich> seems like that airmotor wants to be replaced by a stepper ;-)
[03:09:37] <steves_logging> FYI, in case you don't alreay realize, that air duct on the top of the rear cabinet is not stock 8-)
[03:09:55] <cradek> yep I noticed that. There are two of them.
[03:10:17] <cradek> jmkasunich: as far as I can tell so far, that and the brake are the only air operated parts
[03:10:31] <steves_logging> lube pump?
[03:10:38] <cradek> I think it's electric
[03:10:48] <steves_logging> OK, could not remember
[03:10:55] <cradek> I should be sure (and make sure it's working)
[03:11:15] <cradek> I think the air motor spins fast - it's reduced by two worms - a stepper might kind of suck
[03:11:16] <steves_logging> yes and check for clogged metering widgets on all feed lines
[03:11:38] <cradek> right, check the metering widgets (??)
[03:12:08] <jmkasunich> the lube pump pressurizes the oil lines, but there is a little orifice at each lube point that actually meters out the oil
[03:12:11] <steves_logging> lube pump has small metering orifices so that lube feeds equally to all points
[03:12:24] <cradek> oh ok
[03:12:56] <cradek> I can't get to the upper ones (that you can see in the disassembled photo) - I'll have to look for evidence of them working and hope for the best
[03:13:00] <steves_logging> trash clogging the metering units is the main cause of ballscrew failure
[03:13:52] <steves_logging> X axis nut spins and tends to sling off the oil
[03:14:32] <jmkasunich> * jmkasunich is easily amused
[03:14:58] <jmkasunich> (watching my computer play the game "color lines" by itself)
[03:28:37] <jmkasunich> woo hoo! 1344 points... not bad for something I coded up in a couple of evenings
[03:32:40] <cradek> what is color lines?
[03:32:41] <steves_logging> so JMK, what is it that you are trying to avoid having to do?
[03:32:52] <jmkasunich> I'm avoiding working on 5i20 stuff
[03:33:16] <jmkasunich> cradek: if you have an ubuntu system up, its in the games menu called "five or more"
[03:33:34] <jmkasunich> I've wanted to code an AI to play it since I first saw it (but that was on a doze box)
[03:33:57] <jmkasunich> I hacked the gnome version to allow call an AI function to calculate the next move
[03:34:05] <jmkasunich> and have been messing with it a bit
[03:34:45] <jmkasunich> (open source is cool ;-)
[03:34:54] <cradek> fun.
[03:35:19] <jmkasunich> at this point, I'd say the AI is at least 90% as good as I am
[03:35:22] <jmkasunich> maybe better
[03:39:50] <cradek> oh ... you don't have to move in a straight line.
[03:39:55] <cradek> that helps.
[03:39:58] <jmkasunich> lol
[12:54:52] <jepler_> jepler_ is now known as jepler
[13:21:00] <skunkworks> cradek: does the bridgeport have split ball nuts to get rid of backlash?
[13:26:04] <SWPadnos> mine doesn't, but I put the ballscrews in myself
[13:29:21] <skunkworks> I have no experience with bridgeport equipment
[13:30:51] <skunkworks> work here bought import knee mills. At home we have a large manual brown and sharp mill.
[13:35:02] <skunkworks> the brown and sharp is a 50 taper head
[14:16:47] <jepler> "wait for external digital input (#) to (rise / fall / be high / be low) (with timeout)" -> stores 1 in #5678 if timeout reached, 0 otherwise
[14:17:22] <jepler> "read external digital input (#)" -> finishes all motion, then stores current value of digital input in #5678
[14:17:36] <jepler> "read external analog input (#)" -> finishes all motion, then stores current value of analog input in #5678
[14:18:29] <jepler> I wonder if these are good primitives to add to gcode
[14:20:02] <jepler> everything would happen through new canon calls, so that sai or the axis preview can return whatever values seem like good ones
[14:28:29] <alex_joni> hmm.. I don't quite understand the first one
[14:28:50] <alex_joni> it simply pauses the g-code and waits / timeouts?
[14:33:03] <jepler> the prototype for the canon call might be 'int WAIT(int input_number, int wait_type /* rising, falling, etc */, double timeout);'
[14:33:06] <jepler> so it's the job
[14:33:14] <jepler> of the canon implementor to do the timeout
[14:33:21] <jepler> (in axis, for preview, the "timeout" would take no time)
[15:01:33] <alex_joni> and the input number would be one of the ones synced to motion?
[15:01:40] <alex_joni> sounds good to me..
[15:07:00] <jepler> WAIT() would finish all motion, then wait for the level or edge trigger specified
[15:08:24] <SWPadnos> what if the condition is true after the motion stops? (ie, looking for high on a logic input, what if it's already high)
[15:29:06] <jepler> I imagine 4 wait types, so that you have the choice of edge or level trigger
[15:32:39] <steves_logging> steves_logging is now known as steve_stallings
[15:33:59] <steve_stallings> skunkworks - the Bridgeport and most other commercial machines have anti-backlash ball nuts that rely on two slightly offset tracks in the nut fitted with selected balls to preload properly on the precision ground screw
[15:34:59] <steve_stallings> the "cheap" rebuilds are just polish the screw and fit new balls
[15:37:26] <skunkworks> steve_stallings: interesting. I think our big K&T has 2 nuts that are shimmed for backlash.
[15:37:32] <steve_stallings> but for the fact that I have enough trouble putting the balls back into a non-preloaded nut, I might try a "cheap" rebuild myself
[15:37:36] <skunkworks> (preload)
[15:38:04] <steve_stallings> as the machines get larger it is more likely that there will be two independent nuts
[15:38:18] <skunkworks> I would think that for as popular as the bridgeports are - rebuilt ones would be inexpensive.
[15:38:42] <skunkworks> 'inexpensive'
[15:38:44] <SWPadnos> depends on whether that's BP CNC or BP manual machines, I Think
[15:38:52] <steve_stallings> relative to other machines, maybe, but still a few hundred dollars I would guess
[15:40:03] <steve_stallings> the $700 kits for X and Y to fit manual conversions are a lower grade screw than the factory ones and do not involve the spinning X axis nut
[15:40:38] <SWPadnos> yay for not involving the spinning X axis nut :)
[15:41:05] <steve_stallings> well, it does limit X axis speed due to screw whipping
[15:41:32] <steve_stallings> I like the Bridgeport mechanicals on the Boss machines
[15:41:38] <SWPadnos> true, though I'm not sure the manual machines could really handle really high speeds anyway
[15:42:45] <SWPadnos> if you keep near the center of travel, you actually have 3 points of contact for the screw, so the unsupported length is ~18 inches, which isn't bad for a 1.25 or 1.375 diameter screw
[15:42:59] <SWPadnos> (well, you have 3 supports anyway, but you get what I mean ...)
[15:43:39] <steve_stallings> but are the screws in the $700 kits that large, I thought they were 1" at the most
[15:43:56] <SWPadnos> the ones I got are that large, and precision ground as well
[15:44:10] <SWPadnos> at ~$600 or thereabouts (new old stock, chinese screws)
[15:44:16] <steve_stallings> ah, from the guy that sells on eBay?
[15:44:26] <SWPadnos> yeah - Machine Tools Direct, I think
[15:44:29] <steve_stallings> those are supposed to be nice
[15:44:36] <steve_stallings> yes, that is the guy
[15:45:01] <SWPadnos> mine are definitely good
[15:52:39] <steve_stallings> steve_stallings is now known as steves_logging
[16:38:28] <cradek_> cradek_ is now known as cradek
[17:13:14] <steves_logging> steves_logging is now known as steve_stallings