#emc | Logs for 2010-06-21

Back
[00:00:11] <madsci44> i never do them that way i always run a seperate 24 volt supply for the contactor coils
[00:00:33] <madsci44> maybe a current limiter in series with the switch if that happens
[00:01:54] <alex|lappy> madsci44, that is what i was thinking
[00:02:15] <alex|lappy> in the form of a 15A fuse (the rated current of the switch)
[00:02:21] <madsci44> yeah now i see what your saying about a fuse
[00:02:44] <alex|lappy> I mean, it won't stop the machine, but I will have to replace it before i can start it again
[00:03:14] <andypugh> If the problem is that the switch is stuck then it _will_ stop the machine as the relay will then drop out
[00:03:59] <madsci44> you know what
[00:04:07] <madsci44> you shoudl use a seperate relay contact for the latch
[00:04:19] <madsci44> vs a seperate contact for the main transformer
[00:04:33] <madsci44> that load dump resistor is much less important
[00:04:46] <madsci44> or add a relay
[00:05:05] <andypugh> That's normal, yes. The NO "start" switch is only "live" when the contactor is open, it is normally on a NC contact.
[00:05:08] <madsci44> replace load dump with a bleeder
[00:05:31] <madsci44> then you never have to worry about swich issue
[00:06:00] <alex|lappy> sorry, I am not all up on the electronics
[00:06:06] <alex|lappy> what exactly does a bleeder consist of?
[00:06:42] <alex|lappy> and, "the latch"?
[00:08:45] <madsci44> well latch is like where the switch initially closes the contact and then the contact maintains the power after until its iterrupted
[00:08:54] <madsci44> interrupted
[00:08:54] <alex|lappy> ah, k
[00:08:55] <alex|lappy> yeah
[00:09:04] <SWPLinux> pcw_home: is JTAG necessary to change the product ID on a 5i22?
[00:09:34] <madsci44> so the idea is instead of sharig the same relay pole for both the main power transformer AND the switch on latch, use a seperate contact for the latch
[00:09:51] <SWPLinux> we just tested the 5i22-1.5 with hostmot, but the one I have has the 5i20 ID (so we had to mess with the IDs to get it detected)
[00:10:29] <madsci44> bleeder resistor is just one you leave in circuit at the output that will drain the capacitors after its been off for a few seconds
[00:11:13] <andypugh> If you do the maths, that bleed resistor needs to be big and high-power if the caps are big.
[00:11:38] <andypugh> And then it wastes power all the time, and get hot.
[00:11:39] <madsci44> load dump resistor is like a hefty low-ohm high watt power resistor that drops capacitor charges as quickly as possible,
[00:11:40] <alex|lappy> hmm, i ordered a 8ohm 50 watt one already
[00:12:08] <madsci44> but your load will probably do that anyway
[00:12:16] <andypugh> OK, you are mounting that on the NC terminals then, rather than permanently in circuit?
[00:12:40] <alex|lappy> andypugh, what, the resistor?
[00:12:44] <andypugh> Yes
[00:12:47] <madsci44> i have to get cigs before the store closes but i can draw you a version if you like
[00:12:49] <alex|lappy> that is on the DC side
[00:12:57] <madsci44> when i return
[00:12:57] <alex|lappy> madsci44, i would really apreciate that
[00:13:02] <madsci44> cool
[00:13:27] <andypugh> So, what conects the load-dump to the caps when you turn off?
[00:13:40] <alex|lappy> andypugh, the dpdt relay
[00:14:10] <alex|lappy> it is using one of it's poles for it's own relay and the main relay, the second one is to close the dump circuit
[00:14:16] <andypugh> OK, just wondering. I was thinking of doing that with my drive, bit wasn't sure if it was "normal"
[00:14:19] <alex|lappy> the second one being the second relay
[00:18:13] <andypugh> Yeah, I just found the diagram.
[00:19:43] <alex|lappy> again btw, you guys are crazy helpful
[00:31:40] <Endeavour> Hello.
[00:32:32] <elmo40> jthornton: what repo are you using with 10.04 to get that SMP rtai kernel?
[00:32:48] <Endeavour> What's SMP?
[00:33:23] <Jymmm> THREE CHARACTERS OF THE ENGLISH LANGUAGE
[00:33:30] <Jymmm> doh stupid caps lock
[00:34:05] <elmo40> I love the Brother machines. FAST stuff! http://www.youtube.com/watch?v=L5aC5Kpvibc
[00:34:25] <elmo40> Endeavour: http://en.wikipedia.org/wiki/Symmetric_multiprocessing
[00:34:31] <cradek> elmo40: linuxcnc.org/mozmck/
[00:35:02] <Endeavour> elmo40: I see.
[00:35:22] <alex|lappy> elmo40, 0.o
[00:35:27] <Endeavour> I got EMC2 running on 10.04 Ubuntu following Cradek's instructions. I have terrible latency though.
[00:36:06] <elmo40> alex|lappy: what?
[00:36:10] <elmo40> fast machine, eh?
[00:36:22] <Endeavour> Mine?
[00:36:25] <elmo40> tool changes are great since they don't use CAT style holders.
[00:36:33] <alex|lappy> elmo40, that is fucking onreal
[00:36:35] <alex|lappy> unreal
[00:36:40] <elmo40> LOL
[00:37:27] <alex|lappy> does one even bother with putting your hand on the estop with that thing?
[00:37:39] <alex|lappy> it is not like you could halt it if it decided to kill itself
[00:44:54] <Valen> i swear it looks like its knocking the block over as it moves
[00:46:05] <Valen> its so violent
[00:46:52] <mIreland> hello all. We have worked out most of the mechanical issues on our Kasuga mill; soon it will be time to hack through the 5i20 configuration. I'll be turning up with questions.
[00:48:00] <Valen> mIreland: use the demo config ;->
[00:48:29] <mIreland> I was surprised to find that the limit switches go right to the servo amps. I guess I can use the demo config without lim switches then.
[00:49:06] <mIreland> or is there a demo config made for this machine? that wd be a neat surprise as well
[00:49:49] <SWPLinux> you would normally want limits to cause amps to get disabled (at least in the direction of travel toward that limit)
[00:50:42] <SWPLinux> you probably also want to notify emc that the limit was hit, so either a second contact on the switch, or parallel the output (if it's a safe voltage) to both the amp and the 5i20 (via an isolated I/O card)
[00:51:01] <Valen> mIreland: whats the machine?
[00:51:09] <Valen> 3 axis?
[00:53:06] <mIreland> 3 axis knee mill. I seem to have left the shop without my notes; I will have better answers later
[00:53:27] <Valen> theres a 3 axis servo demo
[00:54:01] <Valen> wiki.linuxcnc.org/cgi-bin/emcinfo.pl?JakeAndRussells its basically all we are using
[00:54:08] <Valen> with few changes ;->
[00:55:41] <madsci44> i am bak
[00:56:00] <mIreland> original westamp servo drives. We are fitting them with encoders to avoid the resolver converter issue
[00:57:47] <Valen> sounds cool
[01:00:04] <andypugh> You can convert 3x resolvers with one Aduino...
[01:00:55] <andypugh> http://wiki.linuxcnc.org/emcinfo.pl?ResolverToQuadratureConverter
[01:01:51] <andypugh> Not that that has been extensively tested yet, but I have spun a 500w servo using it (in conjunction with some other stuff)
[01:03:43] <Jymmm> SWPadnos: you're such a quitter!!!
[01:04:29] <Valen> andypugh: looking pretty cool
[01:05:02] <elmo40> Valen: I wouldn't call it 'violent', it is still a machine. with these mechanical limits any machine could perform at that speed.
[01:05:25] <andypugh> Valen: ? What's looking cool?
[01:05:35] <Valen> the resolver converter
[01:05:53] <andypugh> Have you seen the video?
[01:06:00] <Valen> elmo40: the point i was making was the acceleration is very very fast
[01:09:39] <madsci44> alex|lappy: your relay has only the two poles?
[01:10:03] <Endeavour> Hello Valen
[01:10:18] <Valen> sup
[01:10:20] <StonedSlacker> Okay, so I was playing around with one of those segmented bar led displays trying to figure out the polarity. Using my old computer power supply, the 5V rail, I was able to quickly figure out what was what, but when I tried with the 3V it trips it's internal breaker. Why?
[01:11:14] <StonedSlacker> It just gives a dim flash then the PS dies.
[01:12:13] <elmo40> someone made an ER collet for the MT taper of the spindle on a lathe http://img8.imageshack.us/img8/2955/ercolletholder2.jpg
[01:12:29] <elmo40> kinda rough finish, though
[01:13:36] <andypugh> I have one a bit like that. It was £10 from eBay
[01:13:54] <alex|lappy> madsci44, yes
[01:14:09] <andypugh> I have another one too, but that is a combination M39x4 and ER16 nose for my rotary table.
[01:14:20] <andypugh> I did make that one myself.
[01:14:24] <madsci44> do you have another relay like it? or thats all you got to work with?
[01:14:39] <alex|lappy> madsci44, I can purchase another
[01:14:43] <alex|lappy> they are v. spendy though
[01:14:52] <madsci44> ok
[01:14:59] <alex|lappy> but you know, i don't wanna die
[01:15:08] <alex|lappy> so if I need to spend another $40 or so
[01:15:16] <alex|lappy> so be it :p
[01:15:17] <madsci44> yeah - you can get good deals for contactors on ebay too sometimes
[01:17:39] <elmo40> another pic of that ER collet http://img4.imageshack.us/ifs/3696/img216/2/ercolletholder4.jpg
[01:19:02] <alex|lappy> what taper is that elmo40?
[01:19:16] <elmo40> not sure. MT3?
[01:19:24] <elmo40> whatever the MT taper is on the spindle of a lathe
[01:19:26] <elmo40> mt4
[01:19:47] <alex|lappy> you know, I could really use something like that
[01:20:19] <alex|lappy> I have an mt4 taper on my lathe, and a full set of er40 collets
[01:20:26] <alex|lappy> and no collet chuck
[01:21:46] <andypugh> Make that your first CNC project then. MT tapers are a breeze with CNC :-)
[01:22:01] <alex|lappy> andypugh, I actually have done a bunch of CNC
[01:22:20] <alex|lappy> i am in my final couple of terms as a machine tech student
[01:22:26] <andypugh> I meant the first project on the machine you are building.
[01:22:36] <alex|lappy> hmmm
[01:23:06] <alex|lappy> interpolate the taper on XZ and YZ planes
[01:23:17] <alex|lappy> ooh, that would be fun to write by hand actually
[01:23:20] <alex|lappy> kinda challenging
[01:23:22] <andypugh> (Assuming you are building a CNC lathe. Which is a fair assumption as nearly everyone in the world does. (Unless my sample is somehow skewed))
[01:23:46] <alex|lappy> andypugh, I actually don't have my own, but I do have 3 I can use at school
[01:23:51] <alex|lappy> oh, 4 now~
[01:23:56] <andypugh> it's not hard. You do it all with one G1
[01:24:18] <alex|lappy> 2 brand new haas TL-1s, a mori seki and some crappy rhino training lathe
[01:24:26] <andypugh> On a lathe. Probably more of a challenge to spiral mill it.
[01:24:53] <alex|lappy> oh yeah, easy code on a lathe
[01:25:51] <elmo40> andypugh: your sample is skewed, I built a mill instead of a lathe
[01:25:58] <elmo40> retrofitting a lathe, though ;)
[01:26:11] <alex|lappy> I have been thinking of doing my lathe
[01:26:26] <alex|lappy> it would be easy, especially after having chosen most of the components once already for my mill
[01:26:34] <alex|lappy> well, relatively easy
[01:26:46] <alex|lappy> and I would be able to get rid of the loud as shit gearbox
[01:26:47] <andypugh> There you are. I understand sampling, I spent most of today not writing some code for one of the UK's major market research companies.
[01:28:03] <andypugh> Hmm. The company pays 50p per survey. My code chooses who gets sent the surveys. I ought to hard-code it so I always get one :-)
[01:29:05] <alex|lappy> you could make a few pounds a day at that rate
[01:29:32] <alex|lappy> buy yourself a nice shiny new CNC in about..... 50,000 days!
[01:29:37] <alex|lappy> you can leave all this EMC stuff behind
[01:29:37] <andypugh> Exactly!
[01:29:56] <andypugh> I am not sure I want a CNC that actually works.
[01:30:01] <alex|lappy> lulz
[01:35:36] <andypugh> Night all
[03:00:24] <StonedSlacker> Well, since nobody had anything to say about my last question I take it I just have something screwed up. I thought maybe it was something predictable for people who know more about electronics, using old computer power supplies in particular. Guess not.
[03:03:48] <Endeavour> Hello.
[03:37:18] <Endeavour> Anyone around tonight?
[03:37:36] <mIreland> howdy.
[03:37:59] <mIreland> what are you working on, Endvr?
[03:38:16] <Endeavour> Just ordered parts to build a dedicated box for my CNC Router.
[03:38:45] <mIreland> marvelous. what are you using for electronics?
[03:39:00] <Endeavour> Just some stuff from Probotix. Going to eventually get a Gecko.
[03:39:05] <Endeavour> What about yourself?
[03:39:25] <mIreland> Mesa boards, Atom motherboard.
[03:39:41] <Endeavour> I got an Atom D510
[03:39:49] <mIreland> doing a retrofit on a Kasuga mill
[03:40:25] <mIreland> Atom boards rock for this sort of application. Steppers or Swervos?
[03:41:31] <Endeavour> Steppers.
[03:41:37] <Endeavour> I still haven't quite figured out EMC2.
[03:41:42] <madsci44> hilo
[03:41:42] <Endeavour> Or other software, for that matter.
[03:41:54] <Endeavour> Hopefully I'll get it down next week.
[03:42:11] <mIreland> any CNC or other machining experience?
[03:42:21] <Endeavour> Nothing substantial.
[03:42:45] <madsci44> dija figure out your latency issue Endeavour? (cool nick)
[03:43:40] <mIreland> u had a latency issue with an Atom board? I've gotten smokin results every time
[03:43:57] <madsci44> i could be mixing up people
[03:44:00] <Endeavour> madsci44: No.
[03:44:09] <madsci44> i have been a scatterbrain lately
[03:44:11] <Endeavour> mIreland: I just ordered the Atom.
[03:44:21] <Endeavour> My current machine gives me about 80000ns latency.
[03:44:37] <Endeavour> Core i7 processor, Gigabyte P55AU
[03:44:48] <madsci44> have you been through all the disable power management/acpi in bios, remove uneeded kernel modules etc?
[03:44:49] <Endeavour> NVidia GTS 250 1GB Card
[03:45:15] <mIreland> I see. I think your issue will be fixed by the new board. I bet it was cheap, too.
[03:45:29] <Endeavour> madsci44: No. Few people have had much in the way of suggestions. I've been just kind of randomly messing with stuff. I'm not proficient with Linux.
[03:45:51] <madsci44> there are lots of things you can try its mostly hit and miss trial and error
[03:46:14] <mIreland> You want to turn off anything you don't need in the bios setup. Also, use the simplest video card you can get away with.
[03:46:16] <Endeavour> mIreland: The new machine was about $250.
[03:46:18] <madsci44> i had oen system where disablign usb in bios affected it quite a bit
[03:46:36] <Endeavour> I suspect in part that part of the problem is the graphics card. Atom should take care of it I guess.
[03:46:51] <madsci44> what graphics ?
[03:47:06] <Endeavour> NVidia GTS 250 1GB
[03:47:45] <madsci44> onboard?
[03:47:57] <Endeavour> No. That's the current card.
[03:48:03] <Endeavour> The graphics are built in on the Atom board.
[03:49:00] <mIreland> ouch. is it a whole system, or did you just get a motherboard? for $65 I've made whole systems, just scavenge the old Power supply and find RAM
[03:49:50] <madsci44> best thing is to first go into bios and disable everything you dont use, particularly ACPI and Auto Power Mangement stuff, perhipherals - like USB if you dont need it to test with shut it off in bios
[03:49:50] <Endeavour> $80 for the Atom board, $40 for the case, $45 for the memory, $20 for a DVD drive, and $35 for a hard drive.
[03:50:41] <Endeavour> madsci44: I might try that out. Realistically I didn't really want to use my primary desktop for control anyway. A small, portable box is nicer to have.
[03:50:53] <madsci44> yeah
[03:51:04] <mIreland> My Hardinge uses a 4Gb thumbdrive for it's HD. You won't need a DVD player past installation, so borrow one from another system temporarily
[03:51:30] <mIreland> Better yet, you can install from another thumbdrive if you set the bios correctly.
[03:52:16] <madsci44> yeah i did something like that with a UDMA compact flash card - 4 gig also
[03:52:51] <madsci44> same idea cept it runs off ide instead of usb
[03:52:53] <Endeavour> mIreland: What do you use for your CAM software?
[03:53:40] <mIreland> My hardinge is a lathe. Simple to code for. I use my feeble brain for CAM software.
[03:53:54] <mIreland> But then, I have a few years of exp.
[03:55:03] <mIreland> free open source CAM software has been a weak spot for the EMC community. I haven't checked in a while, but there was not a good answer last time I checked.
[03:57:11] <mIreland> For a router, you will mostly be doing cookie cutter progs. A simple CAM package will do. If you are good a at trig you can do it by hand as I do
[03:59:54] <Endeavour> Vectric stuff any good?
[04:00:32] <mIreland> If it is I'd like to know. I've posted the same question as you in the past.
[04:01:11] <madsci44> no idea
[04:01:21] <Endeavour> Hm
[04:01:26] <mIreland> G code is falling down simple. You will want to be able to verify your progs by inspection anyway.
[04:02:22] <mIreland> Curves are tricky, but a well written spreadsheet would cut right to the correct answers.
[04:02:59] <mIreland> Cutter comp is confusing at first; when you get it you can punch the numbers in right off your blueprints.
[04:04:05] <mIreland> Axis itself is the best G code verifier I've seen; you can tell by looking if you have programmed a crash.
[04:07:01] <madsci44> has anyone tried the dedicated core on smp thing ?
[04:07:31] <SWPLinux> yes
[04:07:32] <madsci44> im wondering if the difference is large
[04:07:46] <SWPLinux> sometimes is doesn't help, sometimes it helps a little, sometimes it helps a lot
[04:08:03] <madsci44> :) which means i HAVE to try it hehe
[04:08:21] <SWPLinux> yep :)
[04:08:28] <mIreland> dedicated core? could you give me a quick explanation?
[04:08:35] <madsci44> and report my results of course
[04:08:50] <mIreland> or just say rtfm...
[04:08:55] <SWPLinux> on a core2 duo machine I tested, it helped a lot to load down the non-RT core with a CPU hog
[04:08:59] <Valen> madsci44: yes
[04:09:23] <madsci44> you can set the setup so that in a multicore cpu one of the processors is dedicated strictly to realtime tasks, while the other runs the normal operating system, ui and so on
[04:09:23] <Valen> SWPLinux: except for that guy who tried it and got a latency of like 150K up from 8k
[04:09:35] <SWPLinux> this was my hog: while true ; do echo "nothing" > /dev/null ; done
[04:09:42] <madsci44> the idea is that it wont have to split its time doing other things adding latency i guess
[04:09:51] <SWPLinux> oh right, I left out the "sucks more" case :)
[04:10:08] <SWPLinux> actually, that's not quite the right description
[04:10:38] <madsci44> ok
[04:10:39] <SWPLinux> you can tell the Linux kernel not to use certain CPU cores for normal scheduling
[04:11:01] <mIreland> should amount to close to the same thing...
[04:11:11] <SWPLinux> but you can also tell the kernel that you want to use one of those cores, which we do with the RT tasks
[04:12:01] <SWPLinux> well, yes and no. we use the highest numbered core for our stuff, but some architectures route all interrupts to the first core, so there's some RT stuff that may have to go through the "shared" core
[04:12:34] <SWPLinux> it varies by CPU and chipset, I bet don't know for sure, and can't test my theory :) )
[04:12:45] <SWPLinux> err
[04:12:59] <SWPLinux> insert period in the right spot and that makes a lot more sense
[04:13:10] <mIreland> Could you just run RT on the second core? Whichever one gets the HW interrupts would be the one to avoid, rite?
[04:13:36] <SWPLinux> except that the RT scheduler uses a hardware interrupt
[04:13:40] <SWPLinux> the clock
[04:14:29] <SWPLinux> in practice, there won't be additional tasks on the isolated core, but I don't know if you can prevent someone from running a task which asks to be put there
[04:16:09] <mIreland> zounds like a chipset needs to be developed to handle our application; we need the clock HW interrupt but shunt all others to another core. over my head, but fun to think about.
[04:16:43] <SWPLinux> might as well design a high performance CPU that has reasonable interrupt response :)
[04:17:54] <mIreland> I'm always leery of running to much extra stuff on my industrial controller anyway. KISS(toopid).
[04:17:57] <Endeavour> Would be nice to have an embedded operating system to take care of everything without needing a whole computer to run the machine.
[04:18:33] <mIreland> I bet some of these wizrdz have done things like that...
[04:18:34] <SWPLinux> that's more or less what RTAI is
[04:18:57] <SWPLinux> you might get slightly better interrupt response from QNX, I don't know
[04:19:25] <Endeavour> SWPLinux: Sort of. You still require a whole computer to run RTAI.
[04:19:25] <madsci44> years ago, when I was up on what was what, ( I have been out of it for a while and these days i am quite lost) I used rt systems with linux for cluster computing research
[04:19:59] <SWPLinux> embedded OS and embedded PC are two different things
[04:20:17] <madsci44> what i liked about the setup then was the realtime which had a base period somewhat like EMC, was triggered externally
[04:20:28] <SWPLinux> you actually don't need a whole PC to run Linux, if you build your own kernel
[04:20:33] <madsci44> to an interrupt
[04:21:00] <mIreland> That is what the clock interrupt is used for now, rite?
[04:21:04] <SWPLinux> madsci44: yeah, adding hardware interrupt threads to HAL is one of my long-term golas
[04:21:19] <madsci44> that way things could be synchronized to external hardware - or other machines etc
[04:21:23] <SWPLinux> it's a timer tick, which fires off the RT scheduler (more or less)
[04:21:25] <SWPLinux> sure
[04:21:35] <SWPLinux> err, goals
[04:21:37] <madsci44> encoders and so on
[04:21:45] <SWPLinux> no, I wouldn't use it for that
[04:21:54] <SWPLinux> PC CPUs aren't microcontrollers
[04:22:23] <madsci44> but like you dont ahve the variability with a fixed base period to smoothly handle high step rates
[04:22:31] <madsci44> its not a microcontroller in either case
[04:22:39] <SWPLinux> there are so many things to change at each context switch, it's nearly impossible to get the interrupt response as good as a $3 AVR (for example)
[04:22:52] <mIreland> So the advantage is syncro to outside machines. Onboard clock cannot be syncronized in that way I suppose
[04:22:52] <madsci44> thats true
[04:22:55] <SWPLinux> you still don't really with variable timing
[04:23:12] <SWPLinux> think about generating slightly different rates for 9 motors
[04:23:13] <madsci44> but even with latency the addition of quantizing makes it much worse at high output rates
[04:23:42] <SWPLinux> sometimes you need 10 cycles between steps
[04:23:48] <SWPLinux> other times 100000
[04:23:56] <mIreland> quantizing effects for outside triggered clock, rite?
[04:23:57] <SWPLinux> (or many many more)
[04:24:17] <madsci44> well then you ahve less issues with it
[04:24:18] <SWPLinux> I'm talking about varying timing so the next step is at the exact right time
[04:24:32] <madsci44> if you are down to stepping every odd base period - then your phase is totally blind
[04:24:41] <madsci44> you have to hope the mechanics will follow it
[04:24:51] <SWPLinux> trouble is, we have 9 independent motion axes, and possibly other things (like spindle) that might want steps generated
[04:24:51] <madsci44> and then latency makes it worse
[04:25:03] <madsci44> well you dont do them independantly
[04:25:21] <madsci44> you tell the controller when all the axis hit these positions wake me
[04:25:34] <SWPLinux> EMC is the controller
[04:25:45] <madsci44> well its a controller
[04:25:45] <mIreland> I get it. 9 axes is so much you mite want multiple computers; they wd require syncro...
[04:26:04] <SWPLinux> no, emc handles 9 axes fine
[04:26:23] <SWPLinux> I was responding to the idea of using variable clock timing to hit exact step times
[04:26:45] <SWPLinux> it's impractical on a PC, or really on any serially executing processor
[04:26:48] <madsci44> or you have an external handle the really critical microsecond synchronization stuff and still benefit from EMC's realtime interaction with it, trajectory planner, and interpereter, and user interface
[04:26:55] <SWPLinux> an FPGA does the job very nicely though
[04:27:07] <SWPLinux> yeah, like a Mesa card
[04:27:09] <madsci44> theres nothing contradictory there i think - its like best of both worlds in my view
[04:27:15] <madsci44> yes like that
[04:27:25] <SWPLinux> encoder counters, PWM, step generation, etc. in hardware, all else in software
[04:27:45] <SWPLinux> in any case, it's time for me to go to sleep. night guys
[04:27:54] <madsci44> gnite
[04:28:32] <mIreland> are Mesa FPGAs currently able to do that? or is this the scenario where outside triggers are useful?
[04:29:19] <madsci44> I have not used them myself so im not the best person to ask - but yes they are available and thats the general idea as i understand it
[04:30:33] <madsci44> I have my own controllers that are hybrid fpga/microcontroller cores, which im trying to do same thing, but what im trying to do i need to improve external interaction, such as via interrupts or some other method than step pins
[04:35:20] <mIreland> so you need HAL to service outside clock interrupts, and the module has not been created. do I understand?
[04:35:34] <madsci44> will be a while i guess before i catch up on everything to do it though
[04:35:34] <madsci44> its amazing how far one can fall behind it seems in only a few years
[04:36:54] <madsci44> right
[04:36:55] <madsci44> outside internal clock interrupts
[04:37:22] <mIreland> sounds like you are the sort of super ninja I'm always in awe of. If I knew what you still remember I'd be farther along than I am.
[04:39:35] <mIreland> FPGA can emulate microcontroller, rite? is this what you have done?
[04:42:02] <mIreland> Could a big enough Mesa FPGA board run EMC by itself if properly programmed?
[04:42:44] <madsci44> man i dont think so
[04:42:44] <madsci44> i used to write drivers in 2.4 and so many things - last couple years i have been doing mostly electronics design and machinery
[04:42:44] <madsci44> software went on the back burner, now its 2.6 and im lost hah,
[04:42:46] <madsci44> they can yes but its better to use a good controller for that in my oppinion
[04:42:47] <madsci44> the nice thing about fpga to me, is reconfigurable circuitry, (cpld's were even cooler but they did not become as popular in high densities)
[04:42:49] <madsci44> so you can have like the equivalent of a dedicated logic circuit for instant output translation or routing like a circuit rather than like a processor
[04:42:50] <madsci44> no
[04:45:07] <madsci44> the purpose isnt so much to run software (with fpga's strictly speaking its not really the point) its more like to run state machines
[04:45:07] <madsci44> the densities are high enough now to create circuitry to behave as simple microprocessors, like say an 8051
[04:45:38] <madsci44> but not so that you could run someting like emc from the fpga, the emc runs on the computer, the fpga adds hardware subsystems to interact with sensors and drivers and so on
[04:46:42] <mIreland> thank you; it would have taken my a while to sort all that out. clearly an 8051 is not going to cut it, but in principle i wasn't quite wrong. I have been partial to Mesa FPGAs but my knowledge is quite limited.
[04:46:47] <madsci44> its like where you have - say an encoder synchronized stepper motor for example
[04:47:40] <mIreland> FPGA will handle all the syncro; computer/EMC supplies the instructions..
[04:53:11] <madsci44> yeek - my internet connection is misbehaving
[04:53:26] <mIreland> i was afraid i'd lost you. you are telling me good stuff...
[04:53:57] <mIreland> tell me if you know, who wrote HostMot2?
[04:53:57] <madsci44> last part i typed was like if you wanted to do floating point math? you are better off with a processor , and if you want to do it fast, probably best to have something with a math core,
[04:53:58] <madsci44> <madsci44>
[04:54:11] <madsci44> so thats where the fpga is weak pc is perfect
[04:54:11] <madsci44> <madsci44> but if you want to react to an input with microsecond resolution, then fpga or cpld is good,
[04:54:11] <madsci44> <madsci44> if your reaction depends on complex calculations , then the idea is have the fpga handle the microsecond response stuff, and interact with the pc to do the complex stuff
[04:54:16] <madsci44> not sure if that all got through
[04:54:25] <madsci44> umm not offhand can peek at the source
[04:55:11] <mIreland> no worries. I keep finding out that there are extra smarties in this chatroom; figured it mite be one of them
[04:55:49] <mIreland> I can look at the source when my curiosity gets that bad:)
[04:56:35] <madsci44> it says opyright (C) 2007-2008 Sebastian Kuzminsky
[04:56:51] <madsci44> nah im mostly new to emc
[04:56:57] <madsci44> i am still finding my way around
[04:57:14] <madsci44> lots of the developers hang out here though
[04:57:28] <madsci44> and definitely lots of smart people
[04:57:31] <mIreland> thanx. not sure why i wanted to know that, but it is the driver I always use.
[04:58:11] <madsci44> i still dont know half of whos who lol
[04:58:34] <mIreland> so, hostmot2 is a linux driver, rite? where does the programming for the FPGA come from?
[04:59:39] <madsci44> well usually the fpga either comes with code on it, or loads it from perhaps an eprom on board, or, in most cases, the driver sends the program to the fpga to configure it
[05:00:28] <madsci44> the fpga configuration itself is usually written in tools for that platform, in languages like vhdl or verilog or other systems provided by chip manufacturers and third party tools
[05:01:07] <mIreland> you anticipated my next question.
[05:01:12] <madsci44> so then you create a configuration - it results in a binary file that, when sent to the chip will configure it as desired
[05:02:05] <madsci44> the driver can also send information on the fly like to parameterize it, or set registers that the configuration reads for choosing functions, etc
[05:03:29] <madsci44> generally the configuration is like the wiring of the circuit. the driver sends it to the device, so then its wired up, and from there on it functions as configured
[05:03:38] <mIreland> Cool. so I just find out what hardware language Mesa FPGAs are programmed in and then climb another learning curve mountain. I love this hobby, I really do.
[05:03:52] <madsci44> :)
[05:04:08] <madsci44> if you have familiarity with digital logic its very easy to learn
[05:05:00] <madsci44> most these days even have visual tools for drawing schematics which take that and figure out the configuration for you
[05:05:39] <mIreland> I have some. Now I need a driving reason to smash books against my forehead! So far I cannot find a reason to program my own, but I do like to grok the whole system when I can.
[05:05:39] <madsci44> but in most cases you can get tighter use of resources doing the verilog yourself
[05:07:39] <mIreland> weird bugs and races would be my concern if I were to take that on... as I recall there were bugs worked out of the HM2 driver after it was deployed. I won't be able to outdo it's author, im sure.
[05:08:44] <mIreland> this is a great discussion for me. for the most part, you are solidifying the things I thought I knew.
[05:10:00] <madsci44> well im generalizing alot to try and get the concept across so probably lots of particulars one could nitpick my descriptions
[05:10:08] <madsci44> i have a stupid way of describing things sometimes
[05:10:32] <mIreland> I'm not trying to learn to program, here, just looking for a light in darkness.
[05:11:51] <mIreland> If I could shift the dicussion a bit, I'd like to ask: In what ways has the advancement to 2.6 stymied you? I would expect most of your knowledge to transfer easily.
[05:12:29] <madsci44> i just havent been following things for a while and lots has changed
[05:12:33] <mIreland> this mite be to linux specific a query, and I bet there is a lot to say. Simple would be fine, I am just looking for hints.
[05:12:49] <madsci44> driver infrastructure is all different, im not familliar with the calls or the structures
[05:13:14] <madsci44> things like - git have come along - different ways of handling repositories and patches
[05:13:55] <mIreland> o. little things like that. I have to admit, the different package management schemes since old Slackware have made my head spin more than once.
[05:13:55] <madsci44> realtime interfaces are different i started with rt-linux and now everything is rtai
[05:14:10] <madsci44> yeah even that - i gave up trying to follow debian policy
[05:14:24] <madsci44> it changes too much i just dont have time to keep up on all of it and get work done at the same time
[05:14:39] <madsci44> google has gone to the dogs
[05:14:51] <madsci44> its much harder to find useful info it seems to me lately
[05:15:08] <madsci44> (easier to buy stuff tho oddly enough :) )
[05:15:43] <madsci44> emc uses TK/tcl and pytho for alot of their UI stuff which i was never into
[05:15:56] <mIreland> not so odd, really. It makes all of our knowledge more valuable if it is hard to find.
[05:16:12] <madsci44> yeah thats why i like this irc thing
[05:16:19] <mIreland> ah yes Python... I've wondered often why we have such a focus on that.
[05:16:24] <madsci44> even just a pointer to a direction is so helpful
[05:16:46] <madsci44> yeah i wonder that too
[05:16:50] <mIreland> again, thanx. that is just what you are doing
[05:17:07] <madsci44> I think its because some involved are comfortable in it so thats what they use
[05:17:42] <madsci44> i think also TCL/TK is considered a light weight toolikit for making GUI's , and its fairly stable so since python has good interface to it
[05:18:02] <madsci44> maybe some strategy to the combination
[05:18:17] <mIreland> EMC uses it so much I've put it on my list of things to learn. Now that I'm working up another machine, it mite be time to take the plunge. more useful than VHDL for me, i bet
[05:18:39] <madsci44> i have never cared for interpereted languages to do complex tasks - i LOVE shell stuff for what its designed for but to me thats different
[05:19:25] <madsci44> python is huge lately it seems everywhere
[05:19:34] <mIreland> shell wizardry is one of the funnest things about nix in general. I'm just a baby, but I can dazzle folks who are starting out.
[05:20:06] <madsci44> yeah i love it - its so powerful and at the sametime always learning new tricks
[05:20:43] <mIreland> The blizzard of different languages is frustrating if you can't predict what will last. I like skills which can't be made obsolete
[05:21:32] <mIreland> a good reason to do VHDL after all I suppose. That one wont fall away I don't think
[05:22:41] <mIreland> Shell trix are the way to instruct noobs; you can never predict what Gooey stuff they will have installed, but the shell is semi universal
[05:24:47] <mIreland> so tell me, what sort of machine tool are you running, anyway? you seem to have some unusual requirements
[05:31:15] <mIreland> Okay, madsci44 fell away... Is there anybody else actually watching rite now?
[05:32:47] <archivist> would that be right now
[05:36:03] <mIreland> :) I was about to start humming softly to myself...
[05:43:46] <madsci44> ack
[05:43:51] <madsci44> no wonder it got so quiet hah
[05:44:01] <madsci44> (my internet again)
[05:46:20] <mIreland> welcome back. I was curious, what sort of machine tools are you controlling?
[05:46:36] <madsci44> so far mostly gantry type systems
[05:46:44] <madsci44> linear grinder
[05:46:51] <madsci44> stuff I build
[05:47:00] <madsci44> I have to do my lathe and mill soon tho
[05:47:22] <madsci44> how bout you?
[05:48:16] <mIreland> so far I've done my Hardinge Chucker only; had an abortive attempt at a stepper based engraver but I destroyed it and got sidetracked. Currently doing a Kasuga knee mill with my uncle
[05:48:35] <madsci44> cool
[05:49:08] <mIreland> My Chucker performs better than anything I've worked with in industry(admittedly, my experience there is limited, too)
[05:49:16] <madsci44> i keep looking at my mill when i walk by it thinking how should i mount motors, how should i drive quill etc
[05:49:32] <madsci44> (and where can i get parts heh)
[05:49:59] <madsci44> wow thats cool tho - its a nice machine to begin with
[05:50:06] <mIreland> Ooh, manual to CNC conversion. Very ambitious. Do you favor servos or steppers?
[05:50:28] <madsci44> well i love ac servos, absent that i prefer steppers
[05:50:45] <madsci44> for the budget i can make them perform very well with my driver designs
[05:51:49] <mIreland> Chucker and Kasuga are both servo designs; now that I think about it the Kasuga might be AC servos, but I'm not sure. No docs.
[05:52:06] <madsci44> i think most industry stuff is ac servo
[05:53:21] <mIreland> Why is AC preferable to DC?
[05:53:24] <madsci44> i run my steppers close loop anyway so i get alot out of them
[05:54:01] <mIreland> so you have linear scales for F/B?
[05:54:36] <madsci44> ac has better control of shaft position, due to sinusoidial drive and encoders, also often higher power density for the same frame size, near constant torque for a wide velocity range down to zero
[05:55:24] <madsci44> i have one setup with linear scales and rotary encoders together most else of what i have built is fine with just rotaries
[05:57:05] <madsci44> i think in alot of what i have done i would stick with the steppers tho
[05:57:34] <madsci44> they have just the right power curve for the feedrate ranges on my systems
[05:58:06] <mIreland> Pretty sure my chucker has DC servos, or at least I thought it did. Radek did such a good job helping me that I never actually had to find out. they buzz; now that I think about it they are probably AC servos.
[05:58:17] <madsci44> in most cases (sometimes depends on the motor) if I profile them with my drive I can get them near as quiet as servos
[05:58:46] <mIreland> say, where are you located?
[05:58:58] <madsci44> ontario, canada,
[05:59:27] <mIreland> I love it up there. I'm currently in San Diego.
[06:00:48] <madsci44> cool what has brought you all the way up here?
[06:02:27] <mIreland> just a road trip, when i was in the service. I didn't see much, but I remember the attitude of the people was very refreshing.
[06:02:44] <madsci44> cool
[06:03:43] <mIreland> I've been hoping to find someone near here who has done this before. I have never found anyone in person who knows this better than I; a little personal interaction can go a long way.
[06:04:05] <madsci44> yeah i hear that - same for me up here
[06:04:53] <madsci44> its almost getting to the point that average public thinks someone is strange because they want to "make things" heh
[06:05:09] <madsci44> maybe true i guess also tho hah
[06:05:11] <mIreland> I know there is at least one person here in SD who did the same model Hardinge that I did. I lost contact with him, though, and I can't seem to dredge up his webpage no matter how many specifics I enter
[06:05:30] <mIreland> SEO my foot
[06:06:54] <mIreland> Saw an interesting comic about mad scientists versus mad engineers; the mad scientist would only destroy *most* of the world and keep the rest as a control...
[06:07:56] <madsci44> hahaha
[06:08:14] <madsci44> well you need an objective frame of reference of course
[06:08:43] <mIreland> That's right. I was on the floor laughing.
[06:09:26] <madsci44> yeah it would be cool to collaborate locally with people with common interests
[06:09:38] <madsci44> i could see going alot further that way in most cases
[06:10:26] <mIreland> Any time I suck anyone into my orbit, even linux luddites, my productivity skyrockets.
[06:11:15] <madsci44> yeah i have noticed that happen to me sometimes too - like this external enthusiasm emerges
[06:12:30] <mIreland> sad that there is such resistance to linux. I've had very smart people walk on by. I just hold faith that this has true worth, and that the time I put in learning will pay off dividends one day.
[06:13:16] <madsci44> it is always a strong asset as a tool at the very least
[06:14:18] <mIreland> I don't live here in SD, I'm just visiting. I'm finding some warp smarties here as I go, and there is a ton of interest in retrofitting machines among folks who just want to run them
[06:15:08] <madsci44> yeah it seems to be quite a wave everywhere lately - i often wonder if a sort of new configuration of industry will eventually emerge out of it
[06:16:43] <mIreland> I hope so.rite now I'm working for room and board, but it is family. with the drawdown of industry, machines can be had for the asking if one is positioned right.
[06:16:57] <mIreland> that is a scary thing
[06:17:08] <madsci44> yeah
[06:17:11] <bootnecklad> Mathmatics Module 1, Higher Tier... Rrrretake
[06:17:37] <madsci44> 42!
[06:17:52] <bootnecklad> I wish... Got a B, teacher wants me to get an A*
[06:18:01] <bootnecklad> he also wants me to take maths for A level
[06:18:14] <bootnecklad> any
[06:18:15] <bootnecklad> BBL
[06:19:15] <mIreland> i iz soo jealous of all yall college students. Self taught may be solid, but it doesn't get credit short of proving one's skill at industry.
[06:20:01] <mIreland> Navy taught me good, but I've had to pull mighty hard at my own bootstraps since
[06:21:03] <madsci44> well who knows we might be lucky in the end
[06:21:34] <madsci44> its getting to where there is no industry - saturated and cleared out to china by all the monopolies
[06:22:35] <madsci44> now therese this interesting combination of creative unemployed people looking to invent a way to benefit from their skill, and new tools like emc2 and software controllers and gadgets like motor drivers made by small startups
[06:22:47] <mIreland> PHDs I know are telling me not to waste my money on skewl. But I'm sure of this: we will always have machines no matter how bad it gets.
[06:22:50] <madsci44> suddenly theres room to be creative and make it happen
[06:23:27] <madsci44> linux itself even - for the software world
[06:23:51] <madsci44> as long as we dont have to ask the chinese how they work.
[06:24:07] <mIreland> It's amazing, isn't it? to compare the rate at which this thing has developed to a real industrial machine development cycle, this is the most efficient thing I've ever seen\
[06:25:06] <madsci44> yes - i think in alot of ways its upsetting the perceived fundamental preaching that "bigger is more efficient" thinking in capitalizm
[06:25:36] <mIreland> Also, nobody slaps me for being a machinist and asking about how VHDL is used to wire FPGAs. I love it. I don't care if I never get rich, but it seems like a possibility.
[06:25:43] <fragalot> Beep
[06:25:46] <madsci44> its like an unwittingly collaborative distributed effort
[06:25:54] <madsci44> honk
[06:26:54] <mIreland> awoooga. yup, all you have to do is take a sharing attitude and you can soar.
[06:27:16] <fragalot> it's the freedom do to whatever you want that gets people going
[06:29:14] <bootnecklad> ^ ... planning a smal plotter!
[06:30:14] <madsci44> i guess thats what i like about automation and cnc, its a place where multidisciplinary ability can shine
[06:31:00] <mIreland> I think that our use of good ol IRC really keeps the riffraff out. Everybody knows that real development is done on modern tools, except us. :)
[06:31:19] <madsci44> lol
[06:32:23] <bootnecklad> why is there no interesting 'technology' news on BBC website
[06:32:23] <madsci44> its a strange world we live in thats for sure i think
[06:32:43] <madsci44> oh there is - but we call their idea of "technology" shopping
[06:33:08] <madsci44> ever notice that? "news" is always about gadgets and marketing
[06:34:22] <bootnecklad> hhmm yeah
[06:34:24] <bootnecklad> http://news.bbc.co.uk/1/hi/world/us_and_canada/10348685.stm
[06:34:25] <bootnecklad> LOL
[06:40:09] <mIreland> hell0 there Hereafter!
[06:40:26] <madsci44> neighbours?
[06:41:09] <mIreland> same house. that's my uncle, the one with the Kasuga mill
[06:41:12] <Hereafter> verry close
[06:41:23] <madsci44> cool
[06:42:04] <mIreland> how could you tell, anyhow?
[06:42:41] <madsci44> who me?
[06:42:48] <madsci44> you both have the same IP number
[06:43:26] <mIreland> I figured. very savvy.
[06:43:34] <fragalot> heh
[06:43:37] <madsci44> << doesnt get out much
[06:43:43] <fragalot> * fragalot pets madsci44
[06:45:08] <mIreland> hey there fragalot. What sort of machines are you working on?
[06:45:49] <fragalot> mIreland: small 3axis engraver mill, a merry go round, and an oldtimer car at the moment
[06:46:36] <Hereafter> what type of oldtimer car
[06:46:54] <fragalot> Trabant P601
[06:46:54] <mIreland> You could make a CNC bullard out of a merry go round, couldn't you...
[06:47:05] <fragalot> mIreland: lol
[06:48:06] <madsci44> its a giant indexing table
[06:48:38] <fragalot> well it's 8x8 meters
[06:49:51] <madsci44> thats like - amusement park size
[06:49:53] <fragalot> doesn't have an encoder though, the motor system is fairly crude. just has 2 .37kW motors, soft-start/stop controller
[06:50:03] <fragalot> madsci44: pretty much :)
[06:51:15] <fragalot> only unique thing apparently is that you can transport, deploy, and run it by yourself
[06:51:21] <fragalot> deploys in 10 minutes
[06:52:47] <mIreland> Cool! is it a money maker, or just an extra cool toy?
[06:53:29] <fragalot> it was a prototype we had to make, has been sold already but it comes back after every use because they don't have the spare room to put it
[06:54:06] <fragalot> some of the new laws around it are just mad.
[06:54:16] <mIreland> where are you located?
[06:54:32] <fragalot> railings can't be horizontal or vertical anymore, but need to be slanted 30 degrees and spaced less than 80mm apart.
[06:54:42] <fragalot> or need to be a square grid of same dimensions.
[06:54:53] <fragalot> stairs can't be higher than 110mm
[06:55:04] <fragalot> mIreland: belgium
[06:56:01] <mIreland> pleased to meet you. must be early in the morning there.
[06:56:29] <fragalot> 9am
[06:57:02] <mIreland> sorry to hear that safety paranoia is not only an American trend.
[06:57:24] <fragalot> it's ok to be concerned about safety, but those measured didn't really make much sense
[06:58:02] <fragalot> though as it's designed for kids, the small stairs make sense. no horizontal railings makes sense too so they can't climb it, but no vertical is a bit o.O
[06:59:16] <Jymmm> Rotary Phase Converter - $650 http://sfbay.craigslist.org/sby/tls/1802461626.html
[07:00:20] <mIreland> that seems like a lot for a rotary phase converter. you can make one from a big motor for cheap
[07:00:35] <Jymmm> http://sfbay.craigslist.org/sby/tls/1802423554.html
[07:02:37] <fragalot> * fragalot off to work!
[07:02:41] <fragalot> baibai
[07:02:47] <alex_joni> Jymmm: http://sfbay.craigslist.org/sby/tls/1800415259.html
[07:02:49] <mIreland> enjoy!
[07:02:51] <alex_joni> a bit pricey, but nice ;)
[07:03:20] <Jymmm> alex_joni: Located in Portland Oregon
[07:04:18] <mIreland> Very nice indeed. I like HAASs. Nothing for us to do to them, though.
[07:04:54] <mIreland> Although, come to think of it, we never had much trouble overflowing the storage on the HAAS mill I ran.
[07:05:11] <alex_joni> Jymmm: still across the pond for me.. so doesn't matter what state ;)
[07:05:20] <Jymmm> alex_joni: LOL
[07:06:07] <Jymmm> alex_joni: Well, I think my new used keyboard is the last toy for me for a while, unless I get real lucky.
[07:06:29] <Jymmm> I love this thing =)
[07:06:51] <alex_joni> heh, what kind?
[07:06:59] <Jymmm> let me get link...
[07:07:43] <Jymmm> http://homepage.mac.com/andrew.hake/Images/mbp17_images/andoru_apple_keyboard.jpg
[07:08:05] <alex_joni> nifty
[07:08:27] <Jymmm> Aluminum body, thin, low profile
[07:09:07] <Jymmm> and I love the feel to it too
[07:09:33] <Jymmm> alex_joni: Can't beat half price. Been wanting one for 2 years
[07:11:18] <Jymmm> alex_joni: It's hard to justify a new keyboard, when the old one is still working perfectly after 10+ years.
[07:13:09] <madsci44> that looks like apple style kinda
[07:13:23] <Jymmm> It's not kinda, it IS an apple mybd
[07:13:28] <Jymmm> kybd
[07:13:37] <madsci44> ah there i go.
[07:13:42] <madsci44> :)
[07:14:21] <madsci44> friend of mine got one an says same thing - feel is excellent
[07:14:54] <Jymmm> It's a love hate thing... either you'll love or hate it.
[07:23:19] <Valen> i have an acer thing i "liberated" from a client
[07:23:26] <Valen> wireless but really good on batteries
[07:24:22] <mIreland> So, I haven't checked in a while... is there any decent open source CAM software out there rite now?
[07:24:49] <alex_joni> mIreland: heekscnc
[07:24:57] <alex_joni> closest to something functional ;)
[07:25:06] <alex_joni> mIreland: check out #cam
[07:25:11] <mIreland> cool! that one looked very promising when it started out
[07:25:30] <alex_joni> http://code.google.com/p/heekscnc/
[07:27:10] <mIreland> 'closest to functional' is a pretty lukewarm assessment, but as I recall there was nothing that could make that claim before.
[07:27:35] <alex_joni> I think some people are already using it
[07:27:44] <alex_joni> but I don't know how pre-beta it is ;)
[07:27:47] <alex_joni> better ask in #cam
[07:29:15] <mIreland> I will, thx for the pointer.
[07:39:33] <mIreland> so, we've been wondering about the merits of resolvers versus encoders. Our Kasuga mill came with resolvers. any good reason not to switch to encoders?
[07:40:49] <mIreland> fcjyhyh7 kitten says hello. down, kitty.
[07:42:34] <mIreland> My first reaction was to yank the resolvers for encoders because of the expense of converters. I am interested to know if there is any good reason not to make that change.
[07:48:02] <mIreland> somebody mentioned an arduino configuration which will convert all 3. that would seem to make converting the resolver outputs a competitive choice, pricewise. this makes me want to choose by technical merits.
[07:49:23] <Valen> encoders are probably the simpler option otherwise you need to work out the resolution of both setups
[07:49:36] <Valen> and also look at any jitter or noise in the resolver system
[07:52:25] <mIreland> simple is good, simple guy that I am. I had used jon elsons converter boards on my last project, but they were not cheap.
[07:54:27] <mIreland> awesome. I am running out of questions. I bet I'll have some after we dig into the wiring tomorrow, though. the dana/summit toolchanger is a bit intimidating.
[07:55:15] <mIreland> I understand that some guys emulate all the logic, and some just use what comes in the machine.
[08:15:49] <madsci44> wow vishay is really blowing up their numbers on mosfets
[08:19:19] <madsci44> so called "continuous current" has a footnote - for <= 1% duty cycle
[08:19:25] <madsci44> duh?
[08:20:02] <celeron55> that sure doesn't sound like very continuous 8)
[08:21:56] <Valen> heh
[08:22:07] <madsci44> yeah really
[08:22:37] <madsci44> who is good for mosfets these days?
[08:23:01] <Valen> I like irf
[08:23:04] <alex|lappy> ur mom!
[08:23:10] <madsci44> LOL
[08:23:19] <alex|lappy> sorry, it just came out
[08:23:29] <madsci44> didnt irf sell out to vishay? i used to always get irf
[08:23:43] <madsci44> welcome back alex|lappy
[08:24:49] <Valen> dunno
[08:26:26] <alex|lappy> madsci44, only for a moment
[08:30:16] <alex_joni> mIreland: on high end robots I sell we use only resolvers
[08:30:29] <alex_joni> they have one advantage over encoders: they are absolute position inside one turn
[08:30:42] <alex_joni> making it (with a bit of work) not needed to home each time
[08:30:54] <alex_joni> on shutdown you save current position and motor turn position
[08:31:23] <alex_joni> on the next powerup you compare rotor position, if it matches, your position isn't compromised and you can go on without homing
[08:31:24] <Valen> you can get absolute encoders too if you wan that
[08:31:41] <alex_joni> Valen: yeah, but those are usually ssi or some other form of serial bus
[08:31:56] <alex_joni> which means it's a bit of a pita to integrate into emc2
[08:32:54] <Valen> true that
[08:38:57] <alex|lappy> hmmm, so if I attached a rubber wheel to a rotary encoder, and put that wheel in contact with the table of my mill, could I make a super gheto linear encoder?
[08:39:23] <mIreland> i don't see why not.
[08:39:38] <alex|lappy> I wonder how accurate that would be
[08:39:43] <alex|lappy> kinda fun to test
[08:39:53] <alex|lappy> would make a cheap +-.001 DRO probably
[08:40:07] <mIreland> probably not very, but you could test and find out by experiment
[08:42:32] <mIreland> if it works to that accuracy you have something useful. The backlash on our current project will certainly be more than .001.
[08:43:26] <alex|lappy> mIreland, I was just thinking all of a sudden of mounting one on my manual lathe
[08:43:40] <alex|lappy> $100 DRO, just hook it up to an arduino and count the pulses
[08:43:59] <Valen> you'll find its linear error will likely be large
[08:44:08] <mIreland> brilliant. you would have to worry about chips and oil, of course
[08:44:18] <alex|lappy> bah
[08:44:31] <alex|lappy> my machines are always flawlessly clean ;)
[08:46:32] <mIreland> I say try it! it mite be a pretty good method, as long as it counts the same going and coming. a rubber wheel could act weird, tho.
[08:47:20] <mk0> rubber wheel cannot be applied
[08:48:43] <mIreland> all surfaces in my shop would be too slick, and the rubber would die after a while as well.
[08:49:18] <mk0> yep
[08:49:57] <mk0> theory of strength of materials
[08:50:30] <mIreland> Still, there could be uses. a run of electric tape down the back of the mill table and a hard wheel, perhaps. like the old concrete wheels on rubber road concept
[08:52:25] <madsci44> better would be someting like piano wire under tension and three small precision discs, two to wrap it around the middle one which drives the wheel
[08:53:13] <madsci44> then just use index points and a pickup along the travel to keep it calibrated
[08:53:22] <alex|lappy> i guess my thought was something you could whipe the oil off with a towel real quick, mag base it on and use it for a couple of minutes
[08:54:02] <alex|lappy> instant portable dro.
[08:54:13] <mIreland> accuracy proportional to circumference.
[08:54:32] <madsci44> yes and easy to get accurate since they are simple lathe project
[08:55:09] <mIreland> and intriguing notion. a hard wheel would seem like a requirement, any way you choose.
[08:55:38] <madsci44> but its arbitrary because you calibrate it mathematically i the arduino or controller or whatever
[08:55:41] <madsci44> in
[08:55:58] <mIreland> otherwise, radius effectively changes
[08:56:36] <alex|lappy> anyway, it is 2am
[08:56:41] <alex|lappy> so uh, good night!
[08:56:44] <madsci44> the index points always are correcting any changes, since over a couple inches or so travel you wont get that much variation
[08:57:03] <madsci44> nite
[08:57:28] <mIreland> how would one make the index points?
[08:57:47] <alex|lappy> mIreland, microswitches?
[08:58:08] <alex|lappy> although that is less temporary
[08:58:22] <madsci44> or an optical sensor and fingers to interrupt
[08:58:31] <madsci44> you just want something with a sharp rise and fall
[08:58:35] <alex|lappy> hmm, oh oh
[08:58:41] <alex|lappy> a track
[08:58:46] <alex|lappy> magnet backed
[08:58:53] <alex|lappy> that you stick on the side of the table
[08:58:59] <madsci44> sure or even drilled holes along a flat strip
[08:59:08] <alex|lappy> yeah
[08:59:31] <alex|lappy> you could maybe do a bit of electrical tape and a knurled bit on the end of the encoder
[08:59:43] <mIreland> could be a good idea. versatile and inexpensive
[09:00:06] <alex|lappy> after you knurl just take a light pass over the top to i give a nice concentric standard dimension
[09:00:59] <mIreland> this could be made and tested in an afternoon. If it holds to .001 you have invented somethng useful
[09:01:12] <madsci44> you can hold better than that im certain
[09:01:28] <alex|lappy> .001 was being cautious
[09:01:57] <alex|lappy> I just don't have the DRO or CNC machine at my disposal right now to test it out
[09:02:04] <alex|lappy> I will on tuesday though...
[09:02:14] <mIreland> question is, will it wander after being run back and forth a bunch of times
[09:02:23] <madsci44> as long as you calibrate it right, it spins freely, the piano wire is held straight, and the controller is properly set to do the compensations accurately for whatever sort of rise and fall you have on the indexes
[09:02:25] <alex|lappy> mIreland, that is what the index points are for
[09:03:16] <madsci44> they are constantly correcting for any wander before it becomes enough to accumulate
[09:03:55] <mIreland> this idea should be developed. DROs do not grow on trees, like encoders do.
[09:04:58] <alex|lappy> mIreland, there really is no reason I can't actually try it early next week
[09:05:10] <alex|lappy> I probably will unless one of you guys get around to it first
[09:05:30] <alex|lappy> all i have to do is take apart one of my motors and pull the encoder out for an afternoon
[09:05:38] <alex|lappy> 6 screws, super easy
[09:06:18] <mIreland> we will be buying encoders for our project very soon. perhaps I will have to give it a whirl
[09:07:36] <madsci44> i have goofed with a few trial setups
[09:08:07] <alex|lappy> whoa, did i say iwas going to bed?
[09:08:10] <madsci44> trick is getting the disks on some sort of bearing, or bushing on a thin shaft, or like a needle shaft - diffeent ways to do that
[09:08:11] <alex|lappy> I think I did
[09:08:17] <madsci44> sorry go to bed hehe
[09:08:19] <alex|lappy> I am a liar
[09:08:28] <mIreland> nobody twisting yr arm...
[09:08:51] <alex|lappy> mIreland, these are the problems with being a nerd
[09:09:01] <mIreland> I am similarly afflicted
[09:09:08] <alex|lappy> ADVANTAGE: YOU ENJOY READING ABOUT CNC MACHINES FOR 10 HOURS STRAIGHT
[09:09:21] <alex|lappy> PROBLEM: YOU ENJOY READING ABOUT CNC MACHINES FOR 10 HOURS STRAIGH, EVEN THOUGH YOU STARTED AT 6PM
[09:09:38] <alex|lappy> there should be a warning label or something
[09:09:50] <mIreland> yes, amusement at stultifying things is a hazard for folks like us
[09:10:14] <mk0> marry!
[09:10:59] <mIreland> a good solution. nothing like a rolling pin to the head to induce drowsiness
[09:11:28] <mIreland> where are u alex|lappy? somewhere east and very late?
[09:11:53] <Valen> mIreland: whats the spend on your encoders looking like?
[09:12:03] <mk0> i'm east. 12 pm.
[09:12:15] <mk0> central europe almost
[09:12:17] <madsci44> you can machine the encoder drive wheel to 3.1416 inches , 1000 cpr us digital encoder, and you get 4000 quadrature counts per inch
[09:13:04] <madsci44> slight groove so the wire doesnt wander off the disc surface
[09:13:17] <mIreland> $85ish for new ones; hopefully we can beat that getting them second hand off ebay or from one of the surplus places around here
[09:14:06] <madsci44> maybe i can dig out the parts and send photos - its in a big box of junk labeled "shit ill never have time to finish"
[09:14:08] <madsci44> hehe
[09:14:08] <alex_joni> mk0: same here
[09:14:38] <alex|lappy> mIreland, nah, west coast USA, but I have school starting tomorrow
[09:14:56] <alex_joni> alex|lappy: school's overrated, you learn so much more in here
[09:15:15] <alex|lappy> alex_joni, well, school for me includes a machine shop and math
[09:15:18] <alex|lappy> that is really it
[09:15:21] <alex|lappy> sooooo I kinda love it
[09:15:24] <mIreland> Im in san diego. where abouts are you?
[09:15:31] <alex|lappy> Eugene Oregon
[09:15:39] <mIreland> oh now im jealous
[09:15:58] <alex|lappy> why, because we have the good herb >_>
[09:16:01] <alex|lappy> <_<
[09:16:03] <alex|lappy> ^_^
[09:16:34] <mIreland> well, my head is clear of that rite now, but I still like hippie chicks ;)
[09:17:14] <alex|lappy> they are lovely, no?
[09:18:10] <mIreland> that can depend; dreddies are a sign of mental instability, but that sometimes translates into some fun.
[09:18:50] <alex_joni> great pic http://upload.wikimedia.org/wikipedia/commons/1/1c/Fox_Hollow_Road,_Eugene.jpg
[09:21:38] <madsci44> i dont get out much
[09:22:01] <mIreland> me neither, but a guy can dream.
[09:22:20] <madsci44> is that ash alex_j?
[09:22:27] <madsci44> or frost?
[09:22:35] <alex|lappy> i think it is photoshop
[09:22:58] <madsci44> ah yes black and white
[09:23:17] <madsci44> neat
[09:24:34] <madsci44> the other trick to make it perfect
[09:25:00] <madsci44> is to have something flat along side the piano wire with a soft surface, just touching, so that it also does not vibrate
[09:26:12] <mIreland> one wheel should do, it seems to me. encoder fixed and wire moving with the slide.
[09:27:17] <mIreland> make the wire lay flat to the rubber track, with a loop around the wheel.
[09:27:47] <alex_joni> why not make something really fun?
[09:28:06] <alex_joni> take some casette tape and an old casette reader head
[09:28:07] <mIreland> what, this doesn't seem like fun to you?
[09:28:13] <alex_joni> use that to build a linear scale
[09:28:42] <alex_joni> you can modulate 'music' to write some sinuses that you read then back, and convert to position
[09:28:42] <mIreland> that is an interesting idea indeed. but you know, you can magnetize the steel as well...
[09:28:45] <madsci44> you dont want anything to inconsistantly deflect the piano wire in the travel,
[09:29:00] <alex_joni> mIreland: not sure how reliable that is
[09:29:06] <alex_joni> but yeah, you could
[09:29:24] <madsci44> there are commercial scales that use that concept - magnetic read heads
[09:29:53] <mIreland> I've been shocked to find a perfect imprint of my mag chuck on my ground parts while I sand their faces. I think it mite work pretty well.
[09:30:19] <alex_joni> how about a good laser mouse directly on the table?
[09:30:19] <madsci44> plastic stretches, shrinks,
[09:30:47] <madsci44> i found those to be inconsistant
[09:31:03] <mIreland> so no plastic. just magnetize the mill table.
[09:31:09] <madsci44> the use an imaging CCD and the way they work is sort of inherently imprecise for this sort of thing
[09:31:57] <madsci44> unless there are other types im not aware of
[09:32:33] <mIreland> The wheel idea is pretty unbeatable. with 4000 counts per rev at one inch, one could clearly get to within a tenth per count with a little bigger one.
[09:33:00] <madsci44> slip would be almost nil too as long as you avoid rubber or knurls or anything like that
[09:33:15] <madsci44> you can even suspend the middle disk without a bearing against the other two disks
[09:33:21] <madsci44> between and off center kinda
[09:33:34] <madsci44> the tension of the wire holds it together
[09:34:06] <madsci44> i really liked the optical mouse idea - that would have been so perfect
[09:34:18] <mIreland> very interesting idea. it brings to mind my experiments with rolamites
[09:34:51] <mIreland> once you get them tensioned correctly they move with almost no effort at all
[09:35:16] <madsci44> also cdrom read lasers - with the focus assembly can be good at creating fine pinpoint for indexing pickup
[09:36:53] <madsci44> rolamites?
[09:37:18] <mIreland> look it up; you won't believe it at first, no one does.
[09:38:03] <mIreland> I used shot tape measures for the springs; strong straight spring steel should work better.
[09:39:13] <mIreland> they make righteous linear slides, but there can be binding problems in practice.
[09:41:06] <mIreland> did you look? I hope you do...
[09:41:12] <madsci44> ah i see
[09:41:23] <madsci44> yes couple videos of different rigs
[09:42:00] <mIreland> knowledge you don't get in college.
[09:43:36] <mIreland> I was making my engraver with rolamite slides. tape measure steel is easy to get but not really good enough. I will revisit the design later
[09:44:23] <madsci44> yeah what i had in mind would be similar except only the middle disk and the encoder wheel
[09:44:54] <madsci44> you would want to do your own hub for the wheel and just buy the plastic disc (like 9 dollars from us digital) integrate the hub and center wheel
[09:45:46] <madsci44> the two other wheels each on either side would then have shafts to support accurately
[09:46:33] <madsci44> another cool thing
[09:46:42] <madsci44> if you spaced the index points randomly
[09:46:53] <madsci44> so the distances from one to the next were all different
[09:47:07] <madsci44> the machine would only have to traverse two and it could know where it is
[09:47:26] <madsci44> for initial reference
[09:47:35] <mIreland> interesting notion.
[09:50:48] <mIreland> index points need to hit very accurately. I'm not certain I can see a way to achieve that.
[09:51:11] <madsci44> it depends how you do your pickup
[09:51:35] <madsci44> if you take apart an old ball mouse you can get a phtotransistor and LED pair
[09:51:53] <madsci44> you can use an aperture to narrow the sensitivity range
[09:52:47] <madsci44> and a comparator to either take the peak, or both the rising and falling edges
[09:52:48] <mIreland> you could have the phototransistor sit at the bottom of a deep hole in the disc; very accurate alignment would be required for the lazer to reach the sensor...
[09:52:53] <madsci44> you can get it pretty sharp
[09:53:17] <madsci44> you would not use it on the disc
[09:53:41] <madsci44> for the disc pick up an EM1 or a HEDS from USDigital for 40 dollars - you cant beat that
[09:53:55] <madsci44> but for the index use the photo transistor
[09:55:43] <madsci44> oh!!!! i just caught what you were saying i think
[09:55:48] <madsci44> i misunderstood right?
[09:56:29] <madsci44> like a collimating hole sortof
[09:57:08] <madsci44> wonder if we could write a profiler in EMC :)
[09:57:09] <mIreland> yup. wont work for variable spaced indexes, but should amount to a very narrow aperture that hits once per rev
[09:57:48] <mIreland> profiler? me not get it
[09:58:23] <madsci44> once you have the disc and the rotary encoder, you have that output, then you have the index pulse output, when all that works
[09:58:57] <madsci44> you have to run through the travels several times while the system locaes the precise position of each index relative to the amount of travel of the wire
[09:59:23] <madsci44> if you do it repeatedly its like a statistical process to refine the known positions
[09:59:40] <madsci44> especially with varying the combinations of moves
[09:59:40] <mIreland> oh, for fixing any wanderings
[10:00:08] <madsci44> so then once the computer or controller has that "map" then during operation it uses that information to apply correction data
[10:00:28] <madsci44> knowing the ideal number of rotary pulses per linear index pair
[10:00:39] <mIreland> that sort of method can get very accurate indeed.
[10:00:46] <madsci44> and always correcting the current position according to the stored profile
[10:00:58] <madsci44> during use
[10:03:07] <mIreland> I like the varying distance to pulse idea, but it looks very complex. if the laser shines continuously and parallel to the direction of motion, you would get good index pulses once per rev in a very simple setup
[10:04:10] <mIreland> it would be like a laser pointer shining always on the wheel, but it could only be sensed when it fell to the bottom of the collimating hole
[10:04:31] <madsci44> its flat long strip with holes along it right? all the holes are in-line, if they are equally spaced that works but they will never be precisely spaced so the counts have to be profiled anyway
[10:05:10] <madsci44> so if they are going to be unequal distance anyway - make them unequal on purpose and use that to derive an initial reference to starting position
[10:07:03] <mIreland> hmm, yr rite there is a problem with my idea... it does not correlate to an absolute position on the slide...
[10:07:59] <madsci44> im a bit confused where the laser came from
[10:08:30] <madsci44> our images diverged somewhere here heh
[10:08:33] <mIreland> I was going a different direction; your idea is more interesting
[10:09:02] <madsci44> oh ok sorry if i missed that
[10:10:12] <madsci44> so as an example if you decide to use a flat piece of materiial for your index strip, you make a hole , or slit, or reflection or whaever technique
[10:10:28] <madsci44> say every inch and a half
[10:11:54] <madsci44> when you get to the first one make it 0.160 early, then at the 3 inch point, make that mark 0.050 late, then at the next mark, 4.5 inches make it 0.80 early and so on
[10:12:22] <madsci44> plan it only enough so each adjacent distance is different
[10:12:57] <madsci44> by a large enough margin to be decisive
[10:13:31] <madsci44> you put your linear pickup to read that
[10:14:01] <madsci44> its attached to the rotary assembly which reads the relative distance
[10:14:01] <mIreland> may as well make them closer; then you find true position with less travel; my question is how narrow(in pulses)can we make each hole? perhaps it doesn't matter, if we are detecting the *edge* of the holes, but then the pattern would be different forward and reverse... but yes, I see that one could code differences in the gaps and find absolute position in two pulse reads
[10:14:20] <madsci44> right exactly
[10:14:50] <madsci44> if your careful to check your optical pickup rise and fall with a scope and calibrate a good fast comparator to detect the edges
[10:15:02] <madsci44> it should be consistant enough,
[10:15:47] <madsci44> if your holes are more like 1/2 or 3/8 inch intervals, depending on your profiling algorithm, they can also help calibrate each other dynamically
[10:15:55] <mIreland> two such strips could code absolute position in a very short traversal
[10:16:17] <mIreland> using such inexpensive pickups, we can multiply at will
[10:16:40] <mIreland> a bit like a vernier scale
[10:16:44] <madsci44> yes or two rows of marks and two pickups - yes thats a good point too
[10:17:15] <madsci44> you want a coarse reference with a variance much greater than your worst case slip
[10:18:05] <madsci44> so that if you get any slip (on my test rigs? it took alot of travel to measure slip so its pretty minor) it will never be enough to be uncertain of which holes your traversing
[10:18:47] <mIreland> oh, you are speaking from experience...
[10:18:49] <madsci44> i hadnt thought of two or more rows thats an idea too
[10:19:08] <madsci44> if you have four rows then you can start encoding them in binary :)
[10:19:27] <madsci44> but now assembly and construction are getting more complex
[10:20:02] <mIreland> the random thing is smarter than the binary idea. it amounts to a resolution much finer
[10:20:04] <madsci44> to my thinking the elegance in the idea is getting very high precision with very little fuss from the benefit of analysis
[10:21:19] <mIreland> one row would of course suffice... as you say, the worst case slip is nearly nil and you would of course traverse at least an inch or two before cutting
[10:22:46] <mIreland> this is a very interesting idea. make the strips out of invar; you will end up with a serious instrument
[10:23:22] <madsci44> it would be cool to perfect a final version
[10:24:13] <mIreland> I imagine that stuff is expensive; I accidently own six feet of one inch bar. flat stock would be necessary, though.
[10:24:54] <madsci44> i would just use some cheap stainless
[10:25:27] <madsci44> channel to hold itself straight, and to provide a recess to hide the piano wire in
[10:26:07] <madsci44> not relaly hide - i mean protect and offer some soft parallel surface to dampen vibration
[10:26:51] <madsci44> the only kink in that whole idea is i dont know how long the wire will last
[10:28:07] <mIreland> make the channel fit the wire tightly and it should be impervious to chips; I bet the wire would last and last. put a strong spring for tension on one end to take up slack as it stretches.
[10:29:22] <madsci44> yeah i had a strong pull spring
[10:29:33] <mIreland> hmm, chips could be a problem at that. better cover it somehow so they don't get rolled around the wheel
[10:30:00] <madsci44> well i thought maybe house the wheels and encoder completely
[10:30:23] <madsci44> then you just need it so when you close the housing there are two fine holes for the wire to enter and exit
[10:30:42] <madsci44> and put wipers perhaps
[10:30:52] <mIreland> the larger the wheels the less the bend; also the greater accuracy
[10:31:03] <madsci44> if you locate it smart i dont think it would be a big worry
[10:31:14] <mIreland> this is a great idea.
[10:31:29] <madsci44> but then the less degrees rotation per linear unit of travel, so the lower resolution
[10:32:44] <mIreland> huh. I had that backwards in my mind. you are correct.
[10:32:51] <madsci44> i was going to try to make mine to fit inside my milling table, above the lead screw along the underside surface
[10:33:32] <mIreland> I bet there is room. we had our knee mill apart yesterday; there is a ton of space in there.
[10:33:56] <madsci44> yeah you dont need much if you build ti compact enough, and then you can affix the encoder to where the table nut is
[10:34:16] <madsci44> in the middle
[10:34:48] <madsci44> you can leave the piano wire hanging while its mounted, then when the table is on, just hook up the springs on each end
[10:35:02] <madsci44> have it planned so that it will line up straight tho
[10:35:27] <mIreland> genius. so, we want the wheel as small as possible without compromising the life of the piano wire. as long as the tension is held, you could use a more flexible fiber
[10:37:03] <mIreland> wow. I really may have to try this. it would be very much more accurate than an encoder on the ballscrew
[10:37:46] <mIreland> not sure where to fit it for the Y slide, however.
[10:38:54] <madsci44> in the knee
[10:39:08] <madsci44> (on mine at least)
[10:39:22] <Valen> you know you can get mylar linear encoders pretty cheap
[10:39:26] <madsci44> not sure if that works for everyone - but theres tons of room there underneath the y screw
[10:39:49] <madsci44> i was looking for them i couldnt find cheap ones
[10:40:39] <madsci44> i was pissed what they wanted for a strip of mylar
[10:41:24] <Valen> a very precisley printed strip of mylar
[10:41:37] <Valen> i seem to recall it being ~$an inch or something?
[10:41:39] <madsci44> they are usually photo processed arent they?
[10:41:46] <Valen> and $30 for the reader
[10:42:04] <madsci44> about 5 years ago i checked, they were like 50 dollars for a 40 inch length
[10:42:09] <madsci44> that was great
[10:42:26] <mIreland> well, that doesnt sound too bad. takes the fun out of it tho
[10:42:29] <Valen> thats pretty cheap
[10:42:31] <madsci44> then about 2 years ago i wanted to order some and they came to more like 150 each
[10:42:53] <madsci44> plus head
[10:42:57] <mIreland> okay now i want encoders again
[10:43:21] <Valen> they are linear scales too
[10:43:30] <Valen> so your measuring where stuff actually is ;->
[10:44:00] <mIreland> good point; no correcting to indexes
[10:45:27] <mIreland> still, this idea could have it's uses. one can make them as long as needed, provided enough tension can be supplied
[10:47:39] <madsci44> LIN-500-34-N 500 cycles per inch, from us digital 34 inch strip 142.80
[10:48:03] <madsci44> then you need a read head, and to mount it on something
[10:49:08] <mIreland> if slip turned out to be negligible, we could leave out the indexing bit and do this very cheaply and easily indeed.
[10:50:05] <madsci44> methinks we are clogging the channel :)
[10:50:20] <madsci44> sorry
[10:50:50] <mIreland> yes, and I have to force myself to sleep. goodnite!
[10:50:59] <madsci44> cya
[10:54:58] <Valen> you know you could just re-invent the resolver, as a tube full of ball bearings
[10:56:47] <jthornton> they all ready have them
[10:57:32] <Valen> they do, i'm just thinking for madsci44 and mireland its a pretty cheap thing
[10:58:22] <jthornton> ah yes I see they were re-inventing the cable encoder weren't they
[11:08:24] <madsci44> its fun to combine old stuff like that with new stuff like software to see what you can get out of it
[11:09:24] <madsci44> not for everyone heh
[11:30:28] <elmo40> ok, this is lame: http://www.youtube.com/watch?v=vOTrZqfEJEI
[11:36:45] <archivist> compared to EMC playing Daisy yes lame
[11:38:08] <jthornton> screensaver on random upped the latency to 6950 and had the cpu working at 50%!
[11:39:19] <archivist> video drivers have a lot to answer for :)
[11:42:11] <jthornton> they were pretty to look at :)
[12:57:34] <pcw_home> <SWPLinux> pcw_home: is JTAG necessary to change the product ID on a 5i22?
[12:57:35] <pcw_home> (you must have a very early card...)
[12:57:37] <pcw_home> You can do it through the bus interface but you need DOS, the 9054EEP program
[12:57:38] <pcw_home> and the batchfile for the 5I22-1.5 these are all in the 5I22 distribution zipfile
[13:08:54] <MarkusBec> MarkusBec is now known as markusandroid
[13:36:40] <SWPLinux> pcw_home: yes, this is one of those 5 I got when you first released the product
[13:36:59] <SWPLinux> pcw_home: is the source of 9054EEP available anywhere?
[13:39:08] <frallzor> bloody connector for the tool setter didnt fit my box. too thick material, so what does one do? Make a nice plate to screw to place =)
[13:39:24] <frallzor> http://pici.se/p/sHjiXmxCb/ pretty ok
[14:18:00] <SWPadnos_> SWPadnos_ is now known as SWPMobile
[14:32:04] <pcw_home> Yes in in the dist
[14:32:09] <pcw_home> its
[14:33:55] <pcw_home> 9056EEP is probably a better source model (as it uses the built in EEPROM access hardware instead of bit banging)
[14:34:15] <pcw_home> 9056eep is part of the 3x20 dist
[14:35:12] <SWPMobile> Ok, great. I'll try that if I can get some time on a computer with a PCI slot :)
[14:47:36] <Jymmm> SWPMobile: who da hell are you?
[14:47:56] <SWPMobile> I am me
[14:48:29] <Jymmm> impostor
[14:48:46] <SWPMobile> cool - I didn't realize that the message light on my phone is RGB
[14:49:03] <Jymmm> heh
[14:49:10] <SWPMobile> it blinked blue when you used this nick
[14:49:26] <Jymmm> SWPMobile: ah
[14:49:56] <Jymmm> droid?
[14:50:17] <SWPMobile> I should make a "mood lighting" app :)
[14:50:17] <SWPMobile> yep
[14:51:08] <Jymmm> SWPMobile: emc2mobile control
[14:51:37] <Jymmm> SWPMobile: use the accelometer to jog
[14:51:38] <SWPMobile> um, you can do that
[14:52:01] <SWPMobile> "CARRIER LOST, ARM REMOVED"
[14:52:24] <SWPMobile> the droid safety relay
[14:53:00] <Jymmm> SWPMobile: I forgot about that..... lose a gui, keep going anyway.
[14:53:23] <SWPMobile> yee-ha!
[14:53:33] <Jymmm> heh
[14:53:35] <SWPMobile> I'm working on that, BTW
[14:53:45] <Jymmm> SWPMobile: cool
[14:54:32] <SWPMobile> if I ever learn to type on my laptop, I may finish it
[14:54:32] <Jymmm> SWPMobile: So, what else besides flash is it lacking?
[14:55:14] <SWPMobile> the Droid has flash, just not 10 yet (that's in 2.2)
[14:55:28] <Jymmm> SWPMobile: So, what else besides flash is it lacking?
[14:55:39] <SWPMobile> .
[14:55:42] <SWPMobile> :P
[14:56:29] <Jymmm> SWPMobile: Gps?
[14:56:37] <SWPMobile> the email client is a little funny (like right now my empty trash folder says it has 14 messages in it)
[14:56:56] <SWPMobile> yep, real GPS, not just E911
[14:57:19] <Jymmm> SWPMobile: Cool. usb or BT tether?
[14:57:48] <SWPMobile> those are possible, but not built in
[14:57:55] <SWPMobile> and wifi
[14:58:04] <Jymmm> no wifi?
[14:58:35] <SWPMobile> it has all the radios, but not the tethering software built in
[14:58:42] <Jymmm> ah
[14:58:59] <SWPMobile> nice screen too, 800x480
[14:59:06] <Jymmm> SWPMobile: What connector, MiniUSB?
[14:59:35] <SWPMobile> micro
[14:59:44] <Jymmm> eh, ok
[14:59:56] <SWPMobile> has better longevity, strangely enough
[15:00:08] <Jymmm> how so?
[15:00:12] <SWPMobile> something like 10000 mating cycles
[15:00:20] <Jymmm> ah
[15:01:04] <SWPMobile> and of course I can charge it by plugging into a computer
[15:01:08] <archivist> thats a lot of mating cycles for consumer grade
[15:01:12] <Jymmm> SWPMobile: Except if you use a mini to micro adapter of course =)
[15:01:46] <SWPMobile> lasts from 1-3 days on a charge, depending on how many radios are on and stuff
[15:02:00] <SWPMobile> jymmm, well don't do that :)
[15:02:05] <Jymmm> archivist: Well, if you consider charging, data transfer, etc
[15:02:38] <SWPMobile> yeah, I looked it up for that reason
[15:02:54] <Jymmm> SWPMobile: Is it carrier locked?
[15:02:58] <SWPMobile> a few cycles/day = 1000/year
[15:03:15] <SWPMobile> well, it depends on what you know :)
[15:03:33] <SWPMobile> but yes, it more or less is because it's CDMA
[15:03:47] <Jymmm> SWPMobile: Does Verizon give you the subsidy code?
[15:03:47] <SWPMobile> the Milestone is the GSM version
[15:03:57] <Jymmm> SWPMobile: oh, that sucks
[15:04:06] <SWPMobile> I haven't asked
[15:04:26] <Jymmm> SWPMobile: Well, if it's CDMA, not a lot of use
[15:04:31] <SWPMobile> I actually planned to replace this phone when LTE is out
[15:04:42] <Jymmm> LTE?
[15:04:59] <SWPMobile> yep, GSM++
[15:05:20] <SWPMobile> a year or two out
[15:05:23] <Jymmm> SWPMobile: Ah, ok. How do you like the scribble texting?
[15:05:36] <SWPMobile> I haven;t
[15:05:40] <SWPMobile> err
[15:05:43] <SWPMobile> tried it
[15:05:53] <SWPMobile> gesture search is cool
[15:06:00] <Jymmm> SWPMobile: DUDE! You have to!!! It's really cool
[15:06:21] <SWPMobile> some day
[15:06:43] <SWPMobile> now, I think I'll go soak in the hot tub. bbl
[15:06:46] <Jymmm> SWPMobile: It works very well.
[15:06:56] <Jymmm> SWPMobile: Is it waterproof? LOL
[15:52:09] <StonedSlacker> I think I may finally have a question worthy of asking
[15:53:27] <StonedSlacker> What is responsible for generating g-code from images? Can it be modified? I would like to change it to do a g91XY move in place of z moves.
[15:54:12] <StonedSlacker> My etch-a-sketch has no z axis as I'm sure you're all aware
[15:55:10] <StonedSlacker> I think if I could replace negative z moves with a small v tick I could draw some pretty neat shit.
[16:08:16] <tom3p> StonedSlacker, you could hack the python source of the supplied utility ....src/emc/usr_intf/axis/scripts/image-to-gcode.py
[16:08:16] <tom3p> or just ignore the problem. i think you can tell emc you have an XYZ machine, yet only wire up the X&Y,
[16:08:16] <tom3p> Then the output would be compressed onto a single plane.. with extra marks where Z would have been higher.
[16:08:23] <frallzor> * frallzor yawns
[16:11:46] <SWPLinux> um, I don't know if that would work
[16:11:56] <tom3p> dang library blocked the freshmeat domain, can view the news
[16:12:02] <tom3p> SWPadnos, ?
[16:12:04] <SWPLinux> image-to-gcode does a raster of the entire image area
[16:12:31] <SWPLinux> if you have no Z, and you're using an etch-a-sketch, you'll get a lot of nice horizontal lines :)
[16:12:45] <tom3p> so he gets a raster :(
[16:12:53] <SWPLinux> yeah, that's what the code does
[16:13:21] <SWPLinux> if there were a waterline mode, it might work as you describe
[16:13:35] <SWPLinux> with some extra connecting lines between the contours
[16:14:40] <SWPLinux> but as you pointed out, the source code is included, and a raster mode might be cool for someone to add (hint hint) :)
[16:14:46] <SWPLinux> errr, contour
[16:19:33] <tom3p> StonedSlacker, image-to-gcode is primarily intended to translate chiaroscuro to Z position,
[16:19:44] <tom3p> so is mostly not suited to a pure xy machine. You'd be throwing away most of it was designed to do.
[16:19:44] <tom3p> It'd be better for you to look at 2D cad and preview in AXIS's Z view.
[16:20:50] <SWPLinux> pcw_home: do you have a project file for the 3x20 utils source?
[16:21:27] <SWPLinux> I think Lazarus may be able to import it if you do (which would save me the trouble of creating one :) )
[16:22:44] <tom3p> oh, the idea of waterline is interesting, didnt 'see' that, like Bali terraces... like (doh) contour maps
[16:24:52] <SWPLinux> heh
[16:31:55] <SWPLinux> pcw_home: it looks like there are two files missing, which may be part of the DOS compiler package (probably defined in some other unit name with freepascal): DataRec and General
[16:52:42] <StonedSlacker> tom3p: Everything runs fine with no Z axis. EMC has no idea that there is no Z axis, it just rolls with it. The reason I'm trying to translate Z->XY is so I can see it. I want to draw things on my etch-a-sketch from image files and since I have no Z axis I have to do something else. I'm thinking a small 45 tick everywhere a plunge is will do the trick. Simply exchanging Z-.02 with G91X.02Y-.02;X.02Y.02;G90;
[16:53:07] <StonedSlacker> Well, time to learn some python
[16:55:22] <tom3p> StonedSlacker, i'd use C for file filters (locate & exchange text), but go for python if its more familiar.
[16:55:37] <tom3p> best of luck
[16:57:56] <StonedSlacker> Neither are familiar. I know cnc programming better than I know the english language but I'm lost with C or python or any other language.
[17:03:42] <SWPLinux> StonedSlacker: you can do that substitution with two passes of sed (maybe one, but I don't know how)
[17:04:08] <StonedSlacker> Me either
[17:04:48] <StonedSlacker> I need to take a minute and clearly define what it is I need and I'm betting I can find someone with the skillz to do it
[17:07:52] <SWPLinux> one sec
[17:13:54] <StonedSlacker> I'm thinking where the code currently varies the Z depth for lighter/Darker areas I would like for it to varies the angle of it's 'tick', darker area=wider angle
[17:18:51] <pjm> SWPLinux - was there any further thoughts on the 7i43 non-reset on EMC shut down issue?
[17:26:27] <SWPLinux> pjm: we couldn't reproduce the problem with the 400k unit we had, hopefully we'll have access to a 200k one tomorrow or Wednesday
[17:26:44] <pcw_home> SWPLinux: freeby.mesanet.com/missing.zip = missing pascal includes Not sure if Freepascal/Lazarus has a way to poke at PCI
[17:27:23] <SWPLinux> inline assembly works with the addition of {$ASMMODE intel}
[17:27:39] <SWPLinux> so that might work for part of it
[17:27:48] <SWPLinux> I'll have to work on that later, in Ann Arbor
[17:28:38] <pcw_home> yabut the DOS stuff works by querying the BIOs
[17:30:28] <SWPLinux> ok, there may be a way, and there's always C if I can't figure out the pascal way :)
[17:31:23] <pcw_home> with the right PCI peek&poke utility you could re-write the subsystem ID by hand
[17:31:24] <pcw_home> (yes using the EEPROM hardware interface it would be pretty trivial to re-write in C)
[17:31:56] <pcw_home> (you need PCI9054 databook)
[17:32:05] <SWPLinux> I've got that at home I think
[17:33:20] <pcw_home> The batch file info is valuable as the info as to what needs to be
[17:33:22] <pcw_home> in the EEPROM requires a lot of flipping back-and forth in the databook
[17:33:56] <SWPLinux> right, I'll look later, time to check out now :)
[17:34:48] <pcw_home> bye
[18:14:45] <spastic_teapot> Can anyone recommend an Autodesk Inventor tutorial, please?
[18:27:05] <frallzor> odd the Alu suddenly started gooking up
[18:27:13] <frallzor> but then suddenly stopped another point
[18:27:18] <frallzor> no difference in cooling
[18:28:17] <Jymmm> try ln2
[18:41:15] <frallzor> Im pretty sure that is a bad idea
[18:41:42] <Jymmm> LN2 or explosives are NEVER a bad idea.
[18:43:04] <Jymmm> Unless substituted for LHe4
[18:47:42] <JT-Work> frallzor: some aluminum is of poor consistency and will have soft areas
[18:48:05] <JT-Work> or if it was exposed to some heat at some point that part will be gummy and soft too
[19:01:27] <fragalot> Rawr.
[19:10:39] <foxtrot> hi
[19:11:11] <micges> hi
[19:11:27] <foxtrot> so Im trying to connect my CNC's 3 axis drivers (3x Allegro 4983s from Pololu) to the parallel port
[19:11:37] <foxtrot> and im trying to figure out what pins I need
[19:11:42] <foxtrot> what is the probe pin?
[19:12:06] <foxtrot> and i dont have an optical encoder so I cant do the home position stuff i dont think
[19:12:24] <foxtrot> this is my first cnc mill build and i am justtrying to complete it
[19:13:14] <micges> probe pin is to use by tool length sensor
[19:13:34] <micges> do you have home/limit switches?
[19:13:34] <foxtrot> i dont have that
[19:13:43] <foxtrot> nope
[19:14:15] <foxtrot> im just trying to get it working for the first time
[19:14:36] <foxtrot> there are no inputs
[19:14:37] <alex|lappy> foxtrot, have you run the stepconf wizard?
[19:14:42] <foxtrot> i dont have any sensors
[19:14:51] <foxtrot> alex|lappy: yeah im in it now
[19:15:06] <alex|lappy> foxtrot, and do you have a breakout board of some type?
[19:16:46] <alex|lappy> hmm
[19:16:53] <alex|lappy> how do I find out how much my table weighs..
[19:17:11] <Valen> big scale?
[19:17:11] <foxtrot> alex|lappy: just the motor driver boards
[19:17:14] <alex|lappy> should I just take a tape measure and dead recon
[19:17:32] <foxtrot> im trying to figure out how to wire them to the parallel port
[19:17:50] <micges> foxtrot: so in stepconf configure only outputs to drivers
[19:17:58] <Valen> alex|lappy: depends on how accurate you want ot be
[19:17:59] <alex|lappy> Valen, well, i could do it when I lift it off, attach a dynometer to the lift
[19:18:09] <frallzor> is it pissible to have a countdown for milling?
[19:18:20] <Valen> or put a scale in the strap
[19:18:25] <frallzor> so when it should start normally it waits like 10 seconds
[19:18:34] <alex|lappy> Valen, I need to figure out if I can run my seros at like 1:2 or something to get a little more speed
[19:19:03] <micges> foxtrot: get parport pin schematics and connect step and dir to proper pins seen in stepconf
[19:19:18] <foxtrot> step and dir are all I need?
[19:19:29] <alex|lappy> foxtrot, yeah, set up at least a quick estop as well
[19:19:31] <Valen> alex|lappy: fair bit of maths involved in working it out that way
[19:19:46] <alex|lappy> foxtrot, hands are a good thing to have
[19:19:55] <ds2> lO2 works better then lN2 on ferrous materials ;)
[19:19:59] <micges> yes
[19:20:26] <alex|lappy> Valen, the maths is not a problem, I just need to get a good estimate of the weight of my table, knee and sadle
[19:20:53] <foxtrot> alex|lappy: how can i setup an estop
[19:20:59] <Valen> you cant work off your existing setup
[19:22:16] <alex|lappy> foxtrot, the easiest way is to take a momentary closed, normally open switch, and attach it to a pin (say pin 10) on one side and +5v from the other. When you close the switch, the pin on the paraport goes high and shuts down the system
[19:22:25] <alex|lappy> well, at least it stops emc from sending output
[19:22:38] <alex|lappy> if your motors are running away because of electrical noise or something you need a more complicated setup
[19:22:49] <foxtrot> oh smart
[19:22:50] <alex|lappy> but at least that gives you a big red button to push that will most likely stop things
[19:23:16] <alex|lappy> just don't trust your life with it ;)
[19:23:45] <foxtrot> is there like a specific way i have to wire parallel
[19:23:46] <alex|lappy> hmm, i wonder if there are any legal liabilities for the advice we give in this channel....
[19:23:47] <micges> I agree, some even simple estop must be on every machine
[19:24:05] <foxtrot> i cant just run step/dir/estop to a 25pin connector with no additional componennts, can I?
[19:24:15] <alex|lappy> yes
[19:24:18] <Valen> alex|lappy: can tell your american ;->
[19:24:20] <alex|lappy> it might not work :D
[19:24:31] <alex|lappy> Valen, tis true, but at least it does not stop me
[19:24:34] <micges> alex|lappy: GPL one ;)
[19:25:52] <alex|lappy> foxtrot, you might of course run into problems with how quickly you can send out pulses, etc etc
[19:26:11] <alex|lappy> foxtrot, there are a lot of commercial break out board type things that you can control with EMC
[19:26:11] <foxtrot> so i dont need any capacitors or anything
[19:26:22] <foxtrot> arduino?
[19:26:41] <alex|lappy> foxtrot, I don't know the details of your setup, but you should be able to plug the pins from your parallel port directly into step/dir on your drivers
[19:26:48] <alex|lappy> external power from a PSU into the drivers
[19:26:58] <foxtrot> yeah
[19:26:58] <foxtrot> ok
[19:27:00] <alex|lappy> and that is all
[19:27:02] <foxtrot> well what about ground?
[19:27:25] <Valen> our kill switch removes low voltage power to our servo amps
[19:27:26] <alex|lappy> there are pins on the paraport for ground (but don't connect your motor power ground to those otherwise you could kill the PC)
[19:28:55] <alex|lappy> foxtrot, you might find that you are going to want to buy some hardware for breaking out your parport
[19:29:07] <foxtrot> why
[19:29:08] <alex|lappy> foxtrot, for example, mine has optical isolated outputs for step/dir
[19:29:10] <foxtrot> like what
[19:29:18] <alex|lappy> so that noise can't get back to the PC from the drivers
[19:29:52] <alex|lappy> plus it has some safety features and a couple of other niceties like an analogue 0-10v out for controling a VFD, some relays on board for coolant and stuff
[19:31:01] <alex|lappy> oh yay, i have a maths class in 3.5 hours
[19:31:28] <alex|lappy> but first day is just going tobe the sylabus
[19:31:37] <alex|lappy> no actual class, I am almost sure, which makes me ;_;
[19:32:42] <alex|lappy> foxtrot, I have something like this, http://www.cnc4pc.com/Store/osc/product_info.php?cPath=33&products_id=181
[19:32:49] <alex|lappy> not necessary, but kinda nice
[19:33:56] <foxtrot> awesome
[19:34:15] <foxtrot> so you think i can not even worry about connecting a gnd to the par port?
[19:34:27] <foxtrot> simply step/dir/estop?
[19:34:38] <foxtrot> 7 total pins in a 25 pin connector
[19:34:54] <foxtrot> what about 'Enable'
[19:36:27] <alex|lappy> foxtrot, is there a ground pin on the driver?
[19:36:34] <foxtrot> yes
[19:36:36] <alex|lappy> ok
[19:36:43] <alex|lappy> yeah, i was wrong about not needing the ground pin then
[19:36:51] <foxtrot> i was going to connect it to the PSU that is also running the drivers
[19:36:52] <alex|lappy> the last pin on your parport (25 I think)
[19:36:53] <alex|lappy> is ground
[19:37:14] <alex|lappy> I think :D
[19:37:28] <alex|lappy> anyone else who is more of a guru then I have an opinion?
[19:37:51] <foxtrot> so i should do 5v from the PSU to the driver, then the GND from the driver to the pin25 on the parport?
[19:38:17] <foxtrot> this is just a modified ATX PSU, 5v for log, 12v for driving the motors
[19:38:59] <Valen> keep in mind for atx you will want to load the 5v rail to get the 12v in spec
[19:39:36] <alex|lappy> foxtrot, so your drivers have a +5v input and a ground for logic, step/dir, +12 and ground for the motor?
[19:40:54] <foxtrot> yeah
[19:40:56] <foxtrot> but i think the ground is going to be all the same
[19:41:02] <foxtrot> being an ATX PSU
[19:41:17] <foxtrot> just power supply black wire
[19:41:59] <alex|lappy> foxtrot, well, uh, plug it in and see what happens :D
[19:42:18] <foxtrot> im gonna be so sad if i smoke it
[19:42:23] <alex|lappy> worse is you kill the computer, but that is relatively unlikely
[19:42:24] <foxtrot> so so sad
[19:42:32] <alex|lappy> most likely is motor noise would pop up
[19:42:41] <alex|lappy> they will start moving all on their own, like MAGIC
[19:42:45] <alex|lappy> scary, finger crushing magic
[19:44:28] <celeron55> the parallel port has lots of ground pins: http://www.iuac.res.in/~elab/phoenix/parport_phoenix/hardware/pinout.gif
[19:44:43] <celeron55> the green ones
[19:45:01] <foxtrot> yeah 18-25 are GND
[19:45:08] <foxtrot> but what does that _mean_
[19:45:13] <foxtrot> do i need to supply them ground
[19:45:19] <foxtrot> or can i just leave them be
[19:45:22] <alex|lappy> I would use them for the logic ground
[19:45:40] <alex|lappy> foxtrot, you got a multimeter?
[19:45:50] <foxtrot> yeah
[19:45:55] <frallzor> hmm which generates more heat, deeper at the same speed or same depth but more feed?
[19:46:09] <alex|lappy> plug in your PSU, check voltage between 5v and the ground on your parport
[19:46:31] <foxtrot> what will that tell me
[19:46:39] <celeron55> why not the PSU ground and the parallel port ground?
[19:46:43] <alex|lappy> if it is a stable +5 then you can probably use the parport ground to return the _5 logic port
[19:47:06] <alex|lappy> fragalot, very complicated question actually
[19:47:11] <celeron55> if they are the same, it's definitely safe to connect them. they should be the sdame
[19:47:15] <celeron55> same*
[19:47:28] <celeron55> that is, the volt meter will read out 0V
[19:47:38] <fragalot> * fragalot glares at alex|lappy
[19:47:49] <alex|lappy> frallzor, in general I would say that more feed is more preferable
[19:47:59] <alex|lappy> it cuts larger chips, which act as a heat sink for the tool
[19:48:00] <alex|lappy> and the part
[19:48:08] <alex|lappy> larger chips have more thermal mass
[19:48:28] <frallzor> tried going from 500 too 1500mm/min in alu
[19:48:36] <frallzor> allmost ok but some clogging
[19:48:39] <frallzor> at 1mm per pass
[19:48:42] <alex|lappy> frallzor, http://www.cnccookbook.com/CCChipThinning.htm
[19:48:44] <alex|lappy> might be relevant
[19:48:49] <alex|lappy> gah
[19:49:12] <alex|lappy> fragalot, what kind of machine?
[19:49:19] <frallzor> a mechmate =)
[19:49:22] <alex|lappy> coz 1500mm is not al that fast
[19:49:38] <alex|lappy> ah, ok
[19:49:45] <bootnecklad> :S I think I did baaaaaaaddddd oin my maths test
[19:49:47] <frallzor> well I tried faster before and then It was cloggheaven
[19:49:53] <frallzor> spindle = 18krpm
[19:49:57] <frallzor> so Im limited
[19:49:59] <alex|lappy> on a router type machine I would def. go for a low DOC high speed
[19:50:08] <alex|lappy> fragalot, what size/type of endmill?
[19:50:12] <fragalot> alex|lappy: glaring.
[19:50:15] <fragalot> lots of glaring.
[19:50:17] <frallzor> 2-4.5mm
[19:50:26] <celeron55> frallzor: anyway, you *need* to have the parallel port ground connected to the thing to where the other pins are connected, too. if there is a voltage difference between the PSU ground and the parallel port ground, you might not be able to connect them and you'll have to use something to decouple them
[19:50:29] <frallzor> broke both 2 and 3 tools
[19:50:32] <alex|lappy> because i run 4k RPM@ 50 IPM all the time, although on a much heaftier machine
[19:51:04] <alex|lappy> ah, so little itty bitty cutters
[19:51:10] <frallzor> and 18k rpm :P
[19:51:13] <alex|lappy> 2 and 3 being 2 and 3 flutes?
[19:51:15] <bootnecklad> wobblybootie !!!
[19:51:20] <frallzor> 2 flutes
[19:51:33] <alex|lappy> frallzor, any chance one of the two flutes just broke on one corner?
[19:51:49] <frallzor> they still sit in the work piece =)
[19:51:57] <alex|lappy> lulz, k
[19:52:08] <alex|lappy> do you have coolant/air?
[19:52:15] <frallzor> coolant
[19:52:22] <alex|lappy> no air?
[19:52:32] <frallzor> just "flooding" the cuts
[19:52:43] <alex|lappy> a nice air blast can do a lot for chip clearing
[19:52:56] <fragalot> so can a decent jet of water
[19:53:36] <alex|lappy> fragalot, of course, but if you don't have a high pressure coolant pump, low pressure flood+air can get the job done
[19:53:50] <fragalot> true
[19:53:59] <frallzor> 500mm/min at 1mm per pass is real ok
[19:54:10] <frallzor> no issues there but a biiit slow =)
[19:54:44] <alex|lappy> frallzor, are you using cam software?
[19:54:53] <alex|lappy> and if so, are there options for high speed milling?
[19:55:10] <frallzor> well I can set speed myself If I want to
[19:55:11] <fragalot> high speed milling is insane
[19:55:14] <alex|lappy> if there is, the toolpaths that keep constant cutter engagement also tend to allow chipds to evacuate
[19:55:31] <alex|lappy> frallzor, not quite what I am talking about...
[19:55:32] <alex|lappy> one second
[19:55:32] <foxtrot> ok well i think i gotit
[19:55:36] <frallzor> well the paths cant be any different than these
[19:55:43] <frallzor> just cutting profiles =)
[19:55:49] <frallzor> from sheets
[19:55:54] <foxtrot> what should i put for my driver? and what about these step time/step space dir hold/setup?
[19:56:02] <frallzor> so speed and depth is all I got to play with
[19:56:16] <celeron55> ah, foxtrot: i meant you in the above "frallzor:" line
[19:56:34] <frallzor> but It seems I could use my spindle even lower than expected it seems
[19:56:34] <alex|lappy> frallzor, http://www.youtube.com/watch?v=JATjRSzXgjw
[19:56:37] <bootnecklad> http://fab.cba.mit.edu/classes/MIT/863.09/people/dcarr/final/final.html
[19:56:39] <bootnecklad> interesting
[19:56:44] <alex|lappy> those kinds of toolpaths tend to clear chips v. well
[19:56:45] <frallzor> I might try going lower in speed first
[19:57:48] <foxtrot> celeron55: no voltage difference, but definitely an amperage difference
[19:58:21] <foxtrot> im breadboarding thisup now
[19:58:36] <foxtrot> but now my next issue is running stepconf
[19:58:39] <celeron55> well, if there is no voltage, there is no current
[19:58:44] <foxtrot> it doesnt have any options for homemade type stuff
[20:03:27] <alex|lappy> foxtrot, urm
[20:03:28] <alex|lappy> ?
[20:03:33] <alex|lappy> didn't quite grok that
[20:03:39] <alex|lappy> home made type stuff?
[20:03:46] <foxtrot> for 'Driver type'
[20:03:53] <alex|lappy> oh
[20:03:55] <foxtrot> imusing Allegro 4983s
[20:04:00] <foxtrot> on bipolar steppers
[20:04:13] <alex|lappy> you just need step frequency and stuff, right?
[20:04:19] <alex|lappy> that should all be in the docs for your drivers
[20:04:27] <Endeavour_> Endeavour_ is now known as Endeavour
[20:04:34] <foxtrot> no its like 'step time' step space'
[20:05:22] <alex|lappy> foxtrot, yeah, all that info should be in your driver docs
[20:05:34] <alex|lappy> step time= 1/step frequency
[20:06:09] <foxtrot> ah
[20:08:23] <foxtrot> sigh i cant find it
[20:10:48] <foxtrot> nothing about direction setup or direction hold
[20:11:01] <foxtrot> does it matter if these settings are wrong
[20:11:05] <foxtrot> im just trying to test
[20:11:51] <celeron55> very probably nothing will get broken
[20:12:37] <alex|lappy> yes
[20:12:41] <alex|lappy> worse is it does not work
[20:12:54] <alex|lappy> it might look like noise if it is too fast
[20:13:53] <frallzor> how odd
[20:14:11] <frallzor> same feed, lowered rpm 18k>12k = worse
[20:14:58] <Endeavour> Hello
[20:15:02] <alex|lappy> fragalot, you increased the chip load
[20:15:08] <alex|lappy> which makes bigger chips
[20:15:12] <alex|lappy> which are harder to evacuate
[20:15:18] <foxtrot> so lower numbers will most definitely work tho
[20:15:23] <foxtrot> like its on 20000ns right now
[20:15:29] <foxtrot> and step time/space are 5000ns
[20:15:30] <alex|lappy> foxtrot, no, but give it a try :D
[20:15:48] <alex|lappy> bah
[20:15:53] <frallzor> I see
[20:15:58] <alex|lappy> frallzor, the chip thing was to you
[20:16:16] <fragalot> * fragalot glares at alex|lappy a bit harder
[20:16:21] <frallzor> hmm maybe increase RPM and try again
[20:16:25] <frallzor> smaller chips
[20:16:55] <frallzor> 24k RPM @1500mm/min might do the trick
[20:16:58] <alex|lappy> hey, you know, fuck all yall. foxtrot fragalot and frallzor all talking at the same time?
[20:17:11] <alex|lappy> how the hell am Is autocomlete supposed to know ;_;
[20:17:26] <alex|lappy> frallzor, I didn't know you had spindle speed to spare
[20:17:38] <alex|lappy> I was under the impression that you were maxed out
[20:17:39] <frallzor> 300Hz = nominal at 18k =)
[20:17:45] <frallzor> so Ive just used that
[20:17:54] <alex|lappy> push that shit faster :D
[20:18:16] <alex|lappy> and i assume you can't use a big tool for roughing, and a little tool for re machining?
[20:18:30] <frallzor> define that
[20:18:56] <frallzor> not really neede when just cutting profiles =)
[20:18:59] <frallzor> *d
[20:19:44] <alex|lappy> you use a large tool to clear stock, then a small tool for a finish pass or two to get into all the corners
[20:19:58] <alex|lappy> frallzor, well, the bigger endmill will clear chips a lot easier
[20:20:07] <alex|lappy> just because it has bigger flutes
[20:20:19] <frallzor> Ive ordered a batch of cheap tools like the ones Im using now
[20:20:23] <frallzor> 2-8mm
[20:21:02] <alex|lappy> if you can I would try to do as much of the work with the 8mm as I could, and then go back and finish the profile with the smaller one
[20:21:10] <frallzor> Ill try 24k rpm @ 1500mm/min tomorrow =)
[20:49:39] <SWPLinux_> SWPLinux_ is now known as SWPLinux
[20:53:49] <SWPLinux> StonedSlacker: image-to-gcode won't work as you like, but you could do something else that may be interesting
[20:54:28] <SWPLinux> the reason it won't work well is that it emits combined XZ moves (IIRC), so you wouldn't be able to separate the Z motion from the XY motion
[20:54:46] <SWPLinux> but what you could do is combine Z with X and Y in HAL
[20:55:36] <SWPLinux> you'd tie the motor position commands to the feedbacks, so emc doesn't see a following error
[20:55:50] <SWPLinux> and then add Z (scaled by some amount) to X and optionally Y
[20:56:12] <SWPLinux> by varying the scales, you change the offset angle
[20:56:48] <SWPLinux> so what you'd get is a horizontal raster that looks a little like you're looking at a 3D thing from the side
[20:56:50] <SWPLinux> sort of
[21:09:57] <Jymmm> Phuuuuuuuck.... I need to get/beg/buy/steal/make some super tiny tools!
[21:10:10] <alex|lappy> how super tiny?
[21:10:34] <alex|lappy> I have a teacher who has some .003" endmills lying around
[21:11:18] <Jymmm> Needs to be not conductive.
[21:11:44] <Jymmm> and in some cases, as thin as a razor blade
[21:11:50] <SWPLinux> ceramic might work
[21:12:19] <Jymmm> SWPLinux: Yeah, just don't know where to find, or how to machine.
[21:12:29] <SWPLinux> me either
[21:12:37] <alex|lappy> what shape of tool Jymmm?
[21:12:46] <Jymmm> various
[21:14:03] <Jymmm> I'm replacing the LCD in a camera that's only 1/2" thick. and would be much easier with tweezers, pry bars etc
[21:14:33] <SWPLinux> you could do it with the power off, and then you could use metal tools
[21:15:18] <Jymmm> SWPLinux: There is no power, but metal gouges the flex cables, connectors 30ga wires etc
[21:15:35] <SWPLinux> so would ceramic
[21:16:18] <Jymmm> SWPLinux: Better feel in the hand though
[21:17:03] <SWPLinux> get a pair of EROP7AASA (I think) tweezers from DigiKey, and use a good magnifier
[21:17:49] <Jymmm> 404
[21:17:51] <SWPLinux> ah, it's the EROP7SA
[21:18:02] <SWPLinux> http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=EROP7SA-ND
[21:18:11] <Jymmm> I have those
[21:18:23] <SWPLinux> ok, I suggest you use those then :)
[21:18:47] <Jymmm> point is too sharp, and the tip is too big
[21:20:02] <Jymmm> SWPLinux: These help, but are too big for some things
[21:20:04] <Jymmm> http://www.dealextreme.com/details.dx/sku.29544
[21:21:28] <SWPLinux> dental tools
[21:21:48] <atmega> ever notice that dealextreme also sells condoms... and pregnancy test things
[21:22:09] <SWPLinux> no. I had never noticed that
[21:22:13] <SWPLinux> thank you very much
[21:22:31] <atmega> you can find out if you are pregnant for like $0.89 with free shipping
[21:24:07] <atmega> or a super-value 50pack of condoms for $3 shipped
[21:24:25] <atmega> I'd suggest getting the pregnancy thing also if you do order those.
[21:24:34] <SWPLinux> coincidence? I don't think so.
[21:25:16] <alex|lappy> yeah, condoms don't seem to me to be the kind of thing you want to skimp on
[21:31:12] <alex|lappy> hmm, so If i put my power supply in a different enclosure then my control hardware, I can keep all the AC current out of the control hardware enclosure
[21:31:22] <alex|lappy> but at the expense of longer DC lines coming out of my power supply
[21:31:27] <alex|lappy> which migh attract noise themselves
[21:31:46] <alex|lappy> i wonder which is more better
[21:33:44] <Endeavour> Anyone here have any experience with Mach3?
[21:37:53] <mikegg> heh, Mach3 is for girl
[21:40:05] <Endeavour> mikegg: Eh?
[21:40:56] <celeron55> 8)
[21:41:04] <Jymmm> SWPLinux: You know of any raw material (polymer) that can take a sharp edge or be machined?
[21:41:16] <SWPLinux> depends on what you want to cut
[21:41:23] <Jymmm> SWPLinux: somewhat brittle would be good
[21:41:27] <SWPLinux> Delrin is very machinable
[21:42:10] <SWPLinux> phenolic would be pretty brittle and is somewhat machinable
[21:42:33] <SWPLinux> but at the small sizes you're talking about, it would probably just break
[21:42:42] <Jymmm> SWPLinux: Ok, too brittle, and a lil less toxic please =)
[21:42:49] <SWPLinux> delrim
[21:42:51] <SWPLinux> n
[21:44:50] <mikegg> reference to the movie Borat "everyone know Ipod mini is for girl"
[21:49:12] <alex|lappy> Jymmm, bakalite
[21:49:17] <alex|lappy> srsly, if you can find some old stuff
[21:49:37] <Endeavour> Anyone here have any hands-on experience with the A3977 by Allegro?
[21:49:47] <Jymmm> alex|lappy: Phenolic IS bakelite
[21:49:53] <alex|lappy> oh
[21:49:54] <alex|lappy> lulz
[21:49:58] <alex|lappy> sorry
[21:50:09] <Jymmm> and it;s toxic to fabricate
[21:50:19] <alex|lappy> meh, that is what a mask is for
[21:50:22] <alex|lappy> :D
[21:50:48] <alex|lappy> Jymmm, what about just plain old wood?
[21:51:03] <alex|lappy> too squishy?
[21:51:06] <SWPLinux> oak or maple
[21:51:34] <Jymmm> not dense enough and too stringy
[21:51:53] <Jymmm> orange stick maybe
[21:52:40] <alex|lappy> http://www.tedpella.com/plastwzr_html/plastic-tweezers.htm
[21:53:40] <Jymmm> Glass filled nylon would work, but is typically mold injected, not machiend
[21:57:56] <Jymmm> Ok, got all 6 flex pcb's reconnected to a 1"x1" PCB and it works. Watch the damn thing fuber AFTER I've installed the last screw
[21:58:47] <alex|lappy> Jymmm, the camera actually has a sensor that makes it only work if it detects there is an extra screw from the inside on your bench when you are done
[21:59:10] <alex|lappy> then, when you take it apart again to put that screw back, it sees the screw is in and stops working when you put everything back together
[21:59:28] <alex|lappy> at least that is how it feels to me sometimes.
[22:01:10] <Jymmm> That's why i *ALWAYS* test before closing up
[22:01:58] <alex|lappy> Jymmm, all of what i described happens regaurdless of the user's actions :D
[22:10:40] <Jymmm> alex|lappy: But I'm not your average bear BooBoo!
[22:13:54] <Jymmm> "Hey BooBoo, how about we go get a picnic basket!" "But Yogi, what about Mister Ranger?" "Phuk the Ranger Booboo!"
[22:15:10] <Jymmm> alex|lappy: thanks for the link btw
[23:20:05] <JT-Hardinge> got the coolant tank out... not as bad as I imagined at this point
[23:26:24] <JT-Hardinge> oh wow no wonder it is so heavy