#emc-devel | Logs for 2008-07-16

[01:23:14] <skunkworks> jmkasunich: I think you are right - eagle causes people to make crappy schematics.. Sure couldn't be me..
[01:23:16] <skunkworks> http://www.electronicsam.com/images/KandT/servostart/newschem.png
[01:23:55] <jmkasunich> ;-)
[01:24:42] <jmkasunich> TL082 as a comparator?
[01:24:57] <jmkasunich> op-amps make crappy comparators
[01:25:23] <jepler> hi skunkworks, jmkasunich
[01:25:31] <skunkworks> Hi.
[01:25:31] <jmkasunich> hi jepler
[01:25:42] <skunkworks> how was the trip?
[01:25:55] <jepler> is "cam" the right term for the thing that protrudes from the table in order to activate a switch at a certain point in travel?
[01:26:08] <jmkasunich> yeah
[01:26:30] <jepler> ok, I wasn't sure and the wikipedia article on cams is all about their use to turn rotary into linear motion
[01:26:30] <jmkasunich> or at least, that is one perfectly acceptable name for it (I'm sure there are others)
[01:27:22] <skunkworks> I really can't think of any other name at the moment.
[01:27:41] <jepler> I've jerry-rigged (that term's racist, right?) a limit switch on my machine, and wondered what anyone thought: http://emergent.unpy.net/files/sandbox/limit-switch-plan.jpg
[01:28:09] <jmkasunich> skunkworks: if you replaced the AND gate with a NAND, you could use the remaining unused sections as inverters and reduce the total chip count
[01:28:25] <jepler> er, "gantry bridge" probably isn't the right term either .. the cam is on one of the "legs" of the gantry
[01:28:35] <jepler> jmkasunich: nah he needs to put a CPLD in there instead
[01:28:40] <skunkworks> so - lm392 would be better..
[01:28:41] <jmkasunich> bah
[01:28:46] <skunkworks> heh
[01:28:51] <jmkasunich> lm211 would be my choice (faster)
[01:28:56] <skunkworks> ah
[01:29:33] <skunkworks> I was trying to decide if I was going to try to do 2 servo amps on 1 board...
[01:29:35] <jmkasunich> the combination of round cam and angled switch leaf will lead to not-very-precise switching
[01:29:46] <skunkworks> with that - I have only 2 extra or gates.
[01:29:51] <skunkworks> the rest is used.
[01:30:14] <jmkasunich> the "let the cam move past the switch" school makes sense for machines where the switch might get smashed
[01:30:39] <jmkasunich> but for your machine, I bet you'd get better accuracy if you put the switch at end-of-travel and head right at it
[01:31:37] <skunkworks> jmkasunich: why not 2 diodes instead of a or?
[01:31:51] <jmkasunich> could work
[01:31:52] <jepler> I could easily mont it that way..
[01:32:19] <jepler> in fact I did that initially but I know it's the "wrong" way :-P
[01:32:48] <jmkasunich> skunkworks: speaking of diode or... you could use the spare gates to combine the overcurrent trip signal and the enable signal
[01:33:19] <jmkasunich> keep in mind that real comparators (as opposed to opamps) rarely have internal pullups
[01:34:29] <jmkasunich> do you have a source of 4000 series chips (they're getting rare)
[01:35:34] <jepler> is everything operating on the unregulated(?) 15V supply? or is there a regulated supply I don't spot?
[01:35:55] <jmkasunich> yeah, I noticed that too - 4000 series is the only thing that will run on high voltage
[01:36:04] <jmkasunich> personally I'd make 5V for the logic side
[01:36:57] <skunkworks> there is a 15 volt 3 leg regulator.. But I don't remember why I picked that.. (it was all cmos)
[01:37:27] <jmkasunich> the gate drive supply needs to be enough to fully turn on the fets
[01:37:36] <jmkasunich> but 12V should be plenty for that
[01:37:44] <skunkworks> maybe I thought the low side of the of the driver ic only used...
[01:37:49] <skunkworks> right
[01:38:05] <jmkasunich> I dunno what the upper limit of 4000 series is, but I bet 15V is close to the edge
[01:38:14] <skunkworks> I think it is 15v
[01:38:59] <skunkworks> I have to do a 'lot' of cleanup of the schematic
[01:40:26] <skunkworks> I didn't have problems getting cmos last time.. maybe I need to do some research.
[01:42:13] <skunkworks> jepler: love what you did with zenbot
[01:42:49] <jmkasunich> cmos does have some advantages - large logic swings and slow propagation delays improve noise immunity, and relatively low output slew rates reduce crosstalk, etc
[01:44:51] <jepler> skunkworks: with the new table I can actually cut circuit boards now .. still working out the kinks, though, and I'm quickly tiring of the dremel
[01:46:57] <skunkworks> Neat! :)
[01:48:52] <jepler> so next I'll blow a couple hundred dollars on a spindle/motor combination that chris found on ebay
[01:49:03] <skunkworks> did he like it?
[01:49:08] <jepler> I don't think he's actually used it yet
[01:49:22] <jmkasunich> link?
[01:49:25] <jepler> if he'll let me I'll borrow his before I buy one
[01:49:25] <skunkworks> heh - I know it got a few good reviews on cnczone
[01:49:53] <jepler> jmkasunich: http://stores.ebay.com/Wolfgang-Engineering_W0QQ_trksidZp1638Q2em118Q2el1247
[01:52:11] <jmkasunich> hmm, the video at http://www.youtube.com/watch?v=nzSjIV9tr3E is not as impressive as the speaker thinks it is
[01:52:23] <jepler> heh
[01:52:36] <jmkasunich> looks like 3/4 of a thou
[01:53:06] <jmkasunich> not bad really, but not super
[01:53:16] <jepler> I measure about 12 thou on my dremel using a similar test
[01:53:24] <jmkasunich> ouch
[01:53:30] <jepler> so it's literally an order of magnitude bettre
[01:54:43] <jmkasunich> hmm, in the longer vid he says he couldn't measure any runout even with a 0.0001 indicator
[01:54:54] <jepler> which was posted first?
[01:54:54] <jmkasunich> I wonder if the runout in the first vid was actually machine flex?
[01:55:18] <jmkasunich> I think the runout one, the long one is still playing
[01:55:57] <jmkasunich> yeah, runout vid was oct, longer one was dec
[01:57:10] <jepler> also for comparison, sherline claims "Runout at spindle nose: .0003" or less, Runout at 3" out from spindle nose: .0005" or less"
[01:58:11] <jepler> (I'm measuring at about 1")
[02:03:02] <jmkasunich> * jmkasunich compares $169 to who knows how many hours if I make my own
[02:03:55] <jepler> don't forget the cost of the motor if you don't have a suitable one
[02:04:21] <jmkasunich> I was planning to use a dremel in round 1 with my homemade spindle
[02:04:35] <jmkasunich> dremel runout is fine if all it's driving is a belt pulley
[02:04:41] <jepler> yeah I suppose
[02:05:56] <jmkasunich> my spindle idea is based on this: http://jmkasunich.com/pics/fast-spindle-2028.jpg
[02:08:10] <jepler> 11krpm? I'd take over 30 if I could get it .. (1krpm = 1 inch/min is apparently the formula cradek uses to compute feed for pcb milling -- http://timeguy.com/cradek/cnc/pcb)
[02:08:35] <jepler> (what is that in sfm?)
[02:08:43] <jmkasunich> speed depends strictly on motor and pulley - I'd probably use a bigger pulley on the motor end for more speed
[02:08:53] <jepler> yeah
[02:08:55] <jmkasunich> SFM depends on cutter diameter
[02:09:44] <jmkasunich> heh, with a 0.005 cutter, even 20K rpm is only 26 sfpm
[02:09:46] <jepler> use .01 inch
[02:09:55] <jmkasunich> 52 - still crawling
[02:09:59] <jepler> yeah something in that neighborhood
[02:12:41] <jmkasunich> until I solve backlash issues a spindle would just be a waste of $$
[02:17:52] <jmkasunich> wtf? is there DRM for pdf files?
[02:18:04] <jepler> how big's your backlash? I'm using BACKLASH = 0.0055 (X) and 0.0045 (Y) and backlash now seems to be only a minor issue
[02:18:21] <jepler> yes, the PDF standard includes DRM features
[02:18:25] <jmkasunich> http://www.thk.com/documents/us_pdf/products/generalB/LMGuide/en_B_085.pdf
[02:18:31] <jmkasunich> can't seem to save it locally
[02:19:06] <jmkasunich> I think my lash is about twice that
[02:19:07] <jepler> I agree; evince greys out the "save" option (ubuntu 6.06)
[02:19:14] <jepler> wget would probably save it
[02:20:03] <jepler> jmkasunich: to fix that: launch gconf-editor, navigate to the item /apps/evince/override_restrictions and turn it on
[02:20:57] <jmkasunich> thanks ;-)
[02:20:59] <jepler> or execute
[02:21:04] <jepler> gconftool --type bit --set /apps/evince/override_restrictions 1
[02:22:06] <jepler> time to tend to the real world .. talk to you later
[02:22:24] <jmkasunich> later
[03:01:21] <cradek> wish my spindle was as simple as R8 to make stuff for
[03:01:32] <jmkasunich> the QC?
[03:01:36] <cradek> yeah
[03:01:47] <jmkasunich> what kind of stuff do you have in mind?
[03:02:05] <cradek> R8 is a drawbar thread, a diameter, a bunch of who-cares, and then a short taper that can be cut with a compound
[03:02:30] <cradek> I was just looking at your photo of the spindle and how easy that part is compared to what I'd have to go through
[03:03:03] <jmkasunich> for a spindle, I'd be tempted to clamp around the nose of the quill and forget the spindle
[03:03:10] <cradek> guess I could hold it in a 1" ER collet or something
[03:03:34] <cradek> the mill has the habit of willy-nilly pulling the quill all the way up...
[03:03:42] <jmkasunich> triangular piece of 1/2" plate, three holes, one for quill, one for spindle, one for motor
[03:03:51] <jmkasunich> oh
[03:03:55] <jmkasunich> that would be bad
[03:04:21] <cradek> if using emc, I could add and honor another limit switch
[03:04:37] <jmkasunich> but putting an accessory spindle assembly in the main spindle and then accidentally turning the main one on would also be bad
[03:04:51] <cradek> it has a neutral - I use it when I mount an indicator etc
[03:05:05] <cradek> (no brake though, then)
[03:06:39] <cradek> fwiw, I agree, smash into the switch head-on. if it's at the end of travel so it's protected, that's best
[03:07:45] <jmkasunich> I don't think I'd use the word "smash"
[03:08:09] <cradek> heh
[11:47:08] <cradek_> cradek_ is now known as cradek
[12:55:52] <skunkworks_> jmkasunich: reading back on the discussion last night - it finally dawned on me what you ment about diodes and or-ing the overcurrent trip signal and the enable signal.. Duh.. that would get rid of the 2 diodes and 'I think' use up all the logic gates on all the chips.. puting 2 drives on 1 board. Neat.
[20:30:34] <CIA-34> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/examples/gcode.lyx: Added more example discriptions
[20:33:01] <CIA-34> EMC: 03bigjohnt 07TRUNK * 10emc2/nc_files/lathe-g76.ngc: Added example with more descriptive name