#emc | Logs for 2009-10-15

[00:01:33] <MIke3800> arg - "Home switch active before start of latch move" - Homing seems like it isn't backing off the switch far enough before it wants to begin the latch search. - is there a simple way to increase the amount it backs off?
[00:03:28] <Valen> push really really hard?
[00:03:44] <MattyMatt> ah cool. steel expands just a bit more than iron. so, a frame made with steel tube could be heated up and filled with cast iron, and then the steel would make a tight skin. steel reinforced iron.
[00:04:06] <Valen> untill it got hot
[00:04:25] <Valen> though i spose in service it shouldn't get up to near the casting temp
[00:05:51] <MattyMatt> nowhere near. flaw in the idea is that the steel would anneal, so you'd end up with cast iron with a soft skin.
[00:06:07] <Valen> heat treat it again
[00:06:17] <MattyMatt> the right shape tho, one piece casting :)
[00:06:18] <Valen> hows the melting point compare?
[00:06:29] <MattyMatt> work harden it with a hammer
[00:06:34] <Valen> lol
[00:07:43] <MattyMatt> or hydraulic press. it'd give you a chance to straighten it
[00:08:00] <Valen> I'll take re-heat treating for $5 larry ;->
[00:08:03] <MattyMatt> squash each square member between polished plates
[00:08:54] <MattyMatt> how would you quench a solid iron CNC? or heat it up to cherry red even?
[00:09:09] <Valen> In a furnace
[00:09:10] <MattyMatt> * MattyMatt only knows 2000 yo heat tratments
[00:09:28] <Valen> If your doing it propperly you need to follow a temperature profile
[00:09:52] <Valen> "cherry red" is great for "close enough" but you can do loads better
[00:10:03] <MattyMatt> I'll use the furnace they use for armour tempering tank shells
[00:10:14] <MattyMatt> 200C for 2 weeks
[00:10:33] <Valen> if thats whats right for the steel your using
[00:11:04] <MattyMatt> I know mild and high carbon
[00:11:41] <MattyMatt> I won't be making jet blades this year
[00:13:11] <MattyMatt> I watched some OU progs on metalurgy, but all I really remember is triple points and how cool their induction furnace was
[00:22:26] <MattyMatt> wow. there's a washing machine that uses a stepper
[00:22:54] <MattyMatt> only in america :) I bet there's no fischer & tykel on MY tip
[00:24:32] <Valen> is it a stepper or a brushless DC
[00:25:36] <MattyMatt> www,instructables.com/id/SDFFVXAFBVHE5LN/
[00:26:22] <MattyMatt> I'd call that a brushless DC with so many rotor coils it might as well be called a stepper
[00:28:20] <MattyMatt> http://www.instructables.com/id/SDFFVXAFBVHE5LN/
[00:28:37] <MattyMatt> watch out for the comma
[00:31:26] <MattyMatt> linear steppers look easy to make if you've got a laser or if you make 4 stamp dies
[00:35:11] <MattyMatt> stack 4 shapes with spacers, T shaped with the vertical stroke in one of 4 positions, repeat as long as you want it, then wind 4 simple coils around the 4 rows of tabs, like a guitar pickup
[00:39:13] <MattyMatt> the slider needs more thought
[00:42:00] <MattyMatt> permanent magents would work nicely
[00:46:30] <MattyMatt> how much is an hour on a laser-equipped torchmate worth? in $
[00:48:44] <MattyMatt> transformer kit, E pieces, 3 pole experiment, mmmmmm
[01:16:12] <AchiestDragon> hmm an interseting mod been looking at the seig x2
[01:16:57] <AchiestDragon> seems that most of the conversion kits for it are like almost as much if not more s the mill itself ,,
[01:17:20] <AchiestDragon> then seen this http://www.cnczone.com/forums/showthread.php?t=51449&highlight=cnc+conversion makes me wonder
[01:19:05] <AchiestDragon> just the way i thought would be good to do the z axis ,, and seems for him works without problems ,, not only that seems he has managed to do a cheap well perfoming conversion to it without spending more than 3times the cost of the machine in parts
[01:23:09] <tom3p> what? run the 'Z' off the 'micro feed knob? the one he made the universal link drive for ?
[01:23:57] <AchiestDragon> the links there on the machine basicaly its a pully where the knob is and a plate to mount the stepper on
[01:24:16] <AchiestDragon> ultra simple by looks
[01:25:41] <MattyMatt> ah so the hilical gearox is fitted already for microfeed? that bit looked expensive
[01:25:42] <tom3p> the orig knob was way behind that plate, on the black rectangle, he's extended it to the pulley plate he's added. whats the 4th amp for, i see just 3 motors.
[01:26:38] <AchiestDragon> he added a 4th rotary acis on the bed later if you follow the thread
[01:26:57] <tom3p> the big spider isn't automated, ah a rotary, nice
[01:28:21] <AchiestDragon> makes you wonder if all the people saing this is better than that have actualy looked at it in proportion to the cost of the machine , or are just basing there facts on what there told and not trying to improve whats there like he has
[01:29:46] <MattyMatt> yeah, I'm still gonna try a stepper driver made with uln 2003 :) probably with transistor power stages tho
[01:30:21] <AchiestDragon> as i keep trying to tell a friend that is looking to convert a simlar sized machine for little cash who has been told to avoid steppers and must have servos ,,, that for his needs and costs it would be like fitting a super charger to a petol lawnmower because your told that it cuts grass better
[01:30:27] <tom3p> the microfeed to stepper is pretty common trick. an easy entry point to automate a quill.
[01:32:22] <MattyMatt> those belts and pulleys arent cheap, but it looks a lot better than motors hanging out
[01:33:01] <AchiestDragon> he replied to the last comment well it cuts faster ,,, but fails to realise that the max feed rate is only going to be what the spindle motor will actualy take off
[01:35:19] <AchiestDragon> true ,, given the budjet i have then i am struggling to afford the machine atm
[01:35:38] <AchiestDragon> the bridgport i was looking at is too big unfortunatly
[01:37:01] <AchiestDragon> but have steppers and controlers and some couplers that i could do a quick mod to a machine like that (with steppers the ugly sticking out style ) using the tools i have already to make
[01:37:48] <AchiestDragon> then spend some time and cnc out proper plates and maybe pullies as looking to make it 5 axis in the end
[01:38:00] <tom3p> are bridgeports common in the UK? its a good machine and spending some budget on housing it may pay off.
[01:38:01] <AchiestDragon> like his 4th axis method
[01:38:41] <AchiestDragon> theres a few keep appearing on ebay
[01:39:28] <tom3p> in the US, shops had rows and rows of them, in the 70's to 90's, lotsa guys cranking handles making molds and dies.
[01:39:47] <tom3p> not cnc at all, usually dro's
[01:39:58] <AchiestDragon> unfortunatly if it was posible for me to get sutable workshop space then it would not pose a problem ,, but the costs are going to be too high
[01:41:07] <AchiestDragon> £12.50 a week for a single car garrage , a 4 year wating list to get one , and if you do theres no mains power so another £2500 for a supply for electric
[01:41:44] <tom3p> leave it whee it is, and bargain work for space and power?
[01:42:34] <AchiestDragon> most get sold eather because the place is closing down or the current owners need the room for something new
[01:42:36] <tom3p> its a shame to see good machines goto scrap
[01:42:49] <AchiestDragon> yea
[01:43:42] <AchiestDragon> unfortunatly with the resession quite a few places are going under , and with no new startups a there is sort of a glut of machines
[01:44:46] <dmess> ive seen 500 pre 1995 cnc machines in 1 auction lately...
[01:44:47] <AchiestDragon> most of the places that have gone bust so far have only closed doors they not started to try to sell on the machines yet , the recivers are wating till the econamy picks up a bit so they standa chance of finding buyers
[01:44:59] <AchiestDragon> then there will be a glut of them on the market
[01:45:00] <MattyMatt> but there's a new industrial revolution happening
[01:45:25] <MattyMatt> recession = bugger banking, let's make stuff
[01:45:26] <AchiestDragon> seen bridgeports not sold at £50
[01:45:27] <dmess> on DIFFERENT stuff all together... large is in
[01:46:39] <dmess> Toshiba has a 2 yr waiting list for 2M table vertical boring mills
[01:46:52] <Valen> servos don't need to be expensive
[01:46:57] <AchiestDragon> but at that time did not have the 6foot square workshop i have now eather
[01:47:39] <Valen> we used 500W scooter motors in ours
[01:47:48] <dmess> i put my harding in my sons bedroom... he's ok with it.. ;)
[01:48:06] <dmess> Hardinge SP
[01:48:27] <MattyMatt> dig down. build cellars. brave new world etc
[01:48:45] <AchiestDragon> as to the whole stepper /servo debate , for me atm i have steppers drivers and psu so would cost £0 , also have some 300W 36v scooter motors , and drivers but would cost for encoders
[01:49:12] <AchiestDragon> lol dig down in a flat the guy downstairs would not be too pleased
[01:49:17] <AchiestDragon> :)
[01:49:26] <dmess> it is in the basement.. i pitty the fool that tries to steal my lathe... the base is 750 lbs
[01:49:29] <MattyMatt> did you see that circuit for making an encoder from a small stepper?
[01:51:02] <AchiestDragon> what i would do is make the conversion , then do a proper job of the conversion mounting using the mill , then at a future point look at servos ,, but if steppers work fine on it then would be no need to upgrade
[01:51:44] <toastydeath> i crashed a lathe at work the other day
[01:51:46] <toastydeath> pretty bad
[01:51:49] <AchiestDragon> as far as i am aware the only disadvantage of steppers is if you have it incorrectly matched you can drop steps
[01:51:56] <toastydeath> the turret's now 1/8th of an inch below centerline
[01:52:06] <tom3p> toastydeath: sorry
[01:52:15] <AchiestDragon> ouch not good
[01:52:20] <BJT_Shop> ouch
[01:52:30] <toastydeath> we have no idea what moved to get it that low
[01:53:18] <tom3p> or raised the spindle ?
[01:53:41] <toastydeath> the casting would have cracked, and the front end of the casting is visible and has no cracks
[01:54:07] <toastydeath> an indicator shows a whole bunch of misalignment on the turret pockets, so that's what I'm going with
[01:54:13] <AchiestDragon> scairy one for me on a lathe was broke a parting tool cutting of a 2" dia steel bar ,,, the tool shatted a bit flew past my head shaving a bit of hair on its way , then i was covered in glass as it shatted the window light in the workshop 25 foot above on its way out
[01:54:41] <toastydeath> i thought i tore the X axis ballscrew out, though
[01:55:06] <toastydeath> when it hit, it moved 3 inches toward me, and the machine's hard limit is way before that
[01:55:59] <toastydeath> it's only got .5 or so of travel past centerline
[01:56:02] <AchiestDragon> that sounds like you broke the casting someware
[01:56:19] <toastydeath> nah, everything is working fine, the turret is just rotated clockwise a few degrees
[01:56:39] <toastydeath> I finished running the particular job because it had no drills
[01:56:48] <dmess> got to have punched a thrust bearing to move that much in 1 hop
[01:57:05] <toastydeath> dmess: the servo shut down and let the ballscrew rotate freely
[01:57:25] <toastydeath> the whole machine did jump, though
[01:57:28] <toastydeath> freaked me out something fierce
[01:57:46] <dmess> it'll happen
[01:58:20] <dmess> they are beasts of their own once you hit cycle start
[01:59:52] <AchiestDragon> Valen: 500w scooter motors but why when the spindle power of the machine is only 300w ,, if machining aluminum that would give you the speed but only to show you that the spindle motor is not up to takeing that much off the job
[02:01:10] <AchiestDragon> end result may be fater feed travel but because of the spindle power you would need to do more passes to remove the same amount of material ,,, maybe worth having the speed on final surfacing but not a lot more
[02:03:03] <dmess> ever heard of a HIGH FEED CUTTER... 3-4mm per TOOTH... yeay baby...
[02:03:19] <AchiestDragon> the other advantage closed loop well it may prevent the stepper problem of dropping steps if you driving it too hard for steppers of the right power
[02:07:45] <MIke3800> logger_emc bookmark
[02:08:21] <AchiestDragon> atm though £400 or so for a mill and using steppers i have for a quick and dirty stepper conversion is affordable ,,, by the way some people talk it seems that the attitude of " must have servos & must have ballscrews" would add more than double the cost of the machine to it plus the impressioin people are giving is basicaly if you dont then it would be totaly unusable
[02:09:25] <AchiestDragon> the latter i know is not true ,, just that it needs a bit more care in its use to use it right ,, and so what if its a bit slower
[02:12:04] <MIke3800> Usability is relative : You could use an Axe to cut down a forest, Others, such as myself would insist on a chainsaw.
[02:12:41] <tom3p> a lot of those bridgeport cncs were steppers, early agie wedms, japax wedms were steppers. they were good machines.
[02:12:43] <MattyMatt> I would insist on a JCB with 6 legs and a logg stripper
[02:12:54] <MattyMatt> is this a whole forest?
[02:13:20] <tom3p> new machines are not stepper, thats a manufacturing decision driven by money
[02:13:48] <AchiestDragon> still perfer my analogy of its like fitting a supercharger to a petrol lawnmower as you get told it cuts grass faster ,, but if you still have to push it your not going to get it done any quicker
[02:14:14] <MattyMatt> feedback is ultimately more reliable, although less failsafe than a stepper
[02:15:38] <MattyMatt> and once you have the feedback sensor solid, you can swap motive force. hydraulic etc
[02:16:42] <AchiestDragon> i could build optical encoders np ,, but from experiance on reparing a0 ploters the most comman problem is dust build up on the encoder sensors that cause the macine to fail , and effectivly can cause stepper like skip if the dust settles on the encoder disk
[02:17:16] <MattyMatt> yeah, image recognition on a vernier would solve a lot of that :)
[02:17:26] <MattyMatt> lots of redundancy
[02:17:46] <AchiestDragon> unless you spend the £60 or so on sealed encoders but steppers of the power need are only half that
[02:17:51] <tom3p> pour a bottle of alcholo thru the encoder, wait a day, good and clean then. 99cents for 90% alcho litre.
[02:18:21] <MattyMatt> wait a day? no chance
[02:18:24] <tom3p> old heidenhain trick
[02:18:25] <AchiestDragon> ok if the encoder disk is one that does not desolve with alcohol
[02:19:08] <tom3p> dunno any non glass wheels, bei, heid, sony.. all glass
[02:19:22] <AchiestDragon> used to strip the ploter encoder case off and vac out and use a fine soft brush to clean the sensor lense
[02:20:16] <bjt-plasma> new plasma computer is live and in the shop... now to hook it up to the machine
[02:20:56] <MattyMatt> running a a baby stepper in reverse wouldn't be affected by non-magnetic contamination
[02:21:26] <MattyMatt> http://www.webx.dk/oz2cpu/20m/encoder.htm
[02:21:31] <tom3p> why cant we tune into bjt's shop and watch? they told me we'd be doing that by now. i read it in popular mechanics.
[02:22:02] <bjt-plasma> I just need a web cam :)
[02:22:33] <AchiestDragon> for a machine the size of the seig x2 would say that 24v electric motor drills would make good suitable servos
[02:23:22] <bjt-plasma> say goodnight Gracie
[02:23:30] <tom3p> gnite gracie
[02:23:44] <bjt-plasma> goodnight all
[02:24:15] <tom3p> and its goodnite from him
[02:25:02] <MattyMatt> and them. they have to go and pick up their new sawmill table by 2pm
[02:25:24] <MattyMatt> 12" blade scares me
[02:25:34] <AchiestDragon> sounds like you will be having fun playing with that
[02:26:19] <MattyMatt> task one, make long pushing sticks, and buy a cheap yellow helmet
[02:26:20] <AchiestDragon> hopes you got some bits of wood to try it out on ,,, otherwise think your mom may object to find you been pracitcing on the furniture
[02:26:45] <tom3p> i installed 6 machines and the customer had made his own Z&W axis. he used hall effect limits and home. they did a lot of drilling on them after mounting. they homed different places every day. i pulled handfulls of drill chips off the trigger magnets on the hall home sensors :)
[02:26:47] <AchiestDragon> was this chair always only 4" high
[02:27:42] <MattyMatt> I've got 2 plywood desks to practice on
[02:28:38] <MattyMatt> and a lod of warped timber to attempt to straighten
[02:29:31] <AchiestDragon> got a full 8*4 sheet of ply that i overbought for the trailer sat here that i still have not decided what to do with ,, only 4.5mm thick but hardwood exteriror ply with a good surface both sides
[02:30:55] <MattyMatt> make concrete former and iron it into a part for your dehavelland mosquito
[02:32:24] <MattyMatt> that wikipedia page was fascinating. I knew it was made of wood but it was a monocoque shell of veneer ply
[02:33:09] <MattyMatt> literally ironed together by canadian ladies with flatirons
[02:34:00] <MattyMatt> one day, I'll have the land
[02:34:10] <AchiestDragon> still have about 10kg of aluminium "foil" on a roll 12" wide about the thickness used on beer cans
[02:34:37] <MattyMatt> nailed over ply, that's a sound roof
[02:35:00] <AchiestDragon> yea thats what i used the other 5k of it for
[02:35:07] <AchiestDragon> 5kg
[02:36:01] <AchiestDragon> and sealed with that nice aluminium foil tape that they use on heating systems
[02:36:33] <MattyMatt> well which is easier to transport? an 8x4 or a 50kg table?
[02:36:54] <MattyMatt> you can use my saw :)
[02:37:18] <MattyMatt> 6mm cuts fine with a stanley knife
[02:37:49] <AchiestDragon> folded a paper airoplane from a bit of it , flys quite well ,,, its would be light enough to make a model acraft from , if you fold it to make small i beems and box section its strong als o
[02:38:38] <MattyMatt> has anyone tried truss rods in a wooden cnc frame? they could compensate for wood warping
[02:39:49] <AchiestDragon> seen a few wooden desgins that look good , but usualy the good ones need a good acurate cnc to cut out the shapes needed
[02:40:00] <MattyMatt> yeah :)
[02:40:25] <AchiestDragon> so its a problem if you dnot have a cnc in the first place
[02:41:08] <MattyMatt> my plan is still to make a woodcarver, then try to cast pieces for mk2
[02:41:28] <MattyMatt> and mk2 may be a lathe instead
[02:43:09] <MattyMatt> gingery style
[02:43:15] <AchiestDragon> may be better making it from steel from the scrap yard then for the first ,, at least once you have built the mk2 you should be able to weigh in the first and get back what you payed for the metal in the first place
[02:43:33] <AchiestDragon> more if the value of metal has gone up in the mean time
[02:44:10] <MattyMatt> if it carves good wood, I'll sell it as a complete machine
[02:44:21] <AchiestDragon> theres that also
[02:44:40] <AchiestDragon> what size you making it
[02:44:56] <tom3p> look at fenns' concrete base with re-bar, cheap and very solid
[02:45:01] <MattyMatt> 540 x 375 x 200 is still the plan
[02:45:20] <MattyMatt> concrete a no go. upstairs construction
[02:45:48] <tom3p> like ferro cement, who was the italian concrete boat builder?
[02:46:14] <MattyMatt> I want the high z to try additive stuff
[02:46:38] <MattyMatt> and lost foam patterns
[02:46:58] <AchiestDragon> the steel one i made does 600mm by 300mm by about 250mm ,, takes a bosh 500w router ,, weighs about 85kg , excluding the stand ,, the steel cost about £45 and another £60 for the stainless steel
[02:47:17] <dmess> 150x150mm tubes .... all joined and filled... SOUNDS like a time warp again..
[02:47:56] <MattyMatt> yeah steel is looking best plan now, except for my drawings and woodpile
[02:48:27] <AchiestDragon> it needs moding realy to take proper leadscrews , and takes std nema 23 steppers ,, if you want it then willing to part with it for what it cost me for the stainless bar
[02:48:35] <MattyMatt> and the difficult of drilling 22mm bearing housings with a 1/4" drill
[02:49:53] <MattyMatt> is this your coffee table? :)
[02:50:16] <dmess> interpolating a bearing fit is a chore on many
[02:50:17] <dmess> REAL cnc... your hole will not be really round
[02:50:18] <MattyMatt> that's a kind offer
[02:51:20] <dmess> with the right tool and speeds and feeds you could bore it..
[02:52:34] <dmess> then its round and in as good a position as you can put it
[02:52:40] <MattyMatt> is there such a thing as a boring bar that goes in a B&D?
[02:53:30] <tom3p> thinking scrapyard, look into http://opensourcemachine.org/ a homebuilt machine from engine block
[02:53:51] <tom3p> a way to get soild without much money
[02:54:01] <tom3p> hah solid
[02:54:40] <dmess> i can fit any tool into any machine. they call me "hack&slash"..on the fly and just a mess
[02:56:01] <MattyMatt> that's how a car mechanic would do it
[02:57:17] <dmess> my new boring mill can cut a 90 degree inside corner with a boring bar.. min corner rad is tool nose +.005"
[02:58:44] <toastyde1th> dmess: do you use contouring heads
[02:59:07] <toastyde1th> (or have you)
[02:59:38] <MattyMatt> one off a mini lathe, threaded to fit the B&D body, would do but it's only 2 speeds, and 140W
[03:00:51] <MattyMatt> and I suppose I'd need the grind the threads, so I'd need a good machine to do that
[03:02:48] <MattyMatt> so either they exist, or I'm shit outa luck for now. That Meddings with the dead motor went for £65 and I gave up on a newer 3 phase one so the other guy got it for £22
[03:05:05] <Valen> so use smaller spindle motors?
[03:05:40] <Valen> sorry smaller servo motors
[03:06:44] <MattyMatt> who me?
[03:07:39] <Valen> AchiestDragon
[03:08:00] <MattyMatt> ah well that would be me doing the modding
[03:08:27] <MattyMatt> because that's an attractive offer. shame I've blown my wad on tools
[03:09:13] <MattyMatt> but more giros loom, provided I can fake 4 weeks of job seeking activity on monday
[03:10:00] <MattyMatt> giro = welfare check
[03:10:09] <AchiestDragon> Valen: as was often told if a jobs worth doing its worth doing well ,, but when prototyping its better to thow something togeth and try it , work out the bugs and then start again buld something that works well and looks the part
[03:11:07] <MattyMatt> I'm hoping for a lego approach. I'll make bolt on legs, independant table etc
[03:11:15] <AchiestDragon> rather to spend months making the protype look the part only to find out it dosent do a good job
[03:12:46] <dmess> have yo got a way to keep track of the co-ordinate system??
[03:13:58] <MattyMatt> only cartesian steps in the controller
[03:14:29] <dmess> but what about programming??
[03:14:35] <MattyMatt> you mean a baseline for alignment?
[03:14:52] <MattyMatt> emc to begin with
[03:15:59] <dmess> we have bore/boss end alighnment... 1 edge alignment
[03:16:54] <MattyMatt> I thought the zero line on the verniers would be my baseline
[03:17:37] <dmess> well i've done it on more than 1 control... its all in the math and a few variables
[03:18:25] <AchiestDragon> up to now been manulay jogging the machien to one corner of the work that i then setting that as 0,0,0 then that both alligns the machine and work up assuming the clamp/vice is set square ,, can manage .2mm that way better if i take some time about it
[03:18:28] <MattyMatt> I can do quaternions and stuff. I do 3d games so trigonometry don't scare me
[03:18:33] <dmess> i resolve to 9 decimal places with the probe on my new Toshiba
[03:19:18] <dmess> so this should be easy...
[03:20:13] <MattyMatt> mmm new plan. make shell of nice ply frame, fill with concrete
[03:20:47] <MattyMatt> momma never knows until it falls through to the kitchen, or she chucks me out
[03:20:52] <AchiestDragon> hopes floor can take weight
[03:21:22] <AchiestDragon> and once cast in concrete going to be a job to move it
[03:21:23] <Valen> I was wondering if there is some way to account for a rotational offset in the job when you load it
[03:21:54] <cradek> devel version of emc2 has coordinate system rotation
[03:22:02] <dmess> we have macro that sets all our fixture offsets/(different orientations) of the b-axis on horizontal full rotary
[03:22:56] <AchiestDragon> would be kool that for some of the less acurate home builds where theres no phisical fix for axis mis allignment
[03:23:47] <MattyMatt> that's more than rotation, but yeah still an affine transformation, same code
[03:23:48] <cradek> I can't picture a kind of misalignment that can be fixed with a rotation
[03:24:17] <MattyMatt> each axis would be a vector
[03:24:20] <AchiestDragon> seen way to meny build scrapped because they forgot to include allignment and got it out
[03:25:13] <MattyMatt> it can all be corrected in software, but you'll get moire patterns
[03:25:38] <cradek> for some value of 'all'
[03:25:38] <AchiestDragon> would be more usefull to more people than trying to do backlash compensation in software
[03:29:43] <MattyMatt> right bed for me 9.5 hrs until I start incurring storage charges
[03:29:56] <AchiestDragon> laters
[03:30:03] <MattyMatt> goodnight
[03:30:08] <AchiestDragon> bed for me also i think
[03:35:09] <tom3p> nite
[03:36:42] <tom3p> cradek: coord sys rotation? does that go as far as tilting the Z ( tilting the xy plane and its norm ?)
[03:45:11] <Optic_> Optic_ is now known as Optic
[03:59:50] <cradek> tom3p: it is just rotation around Z
[04:14:01] <tom3p> thx
[04:15:32] <tom3p> i imagined its sorta like kins to tilt the whole system, where a motion in Z' is mode by moving X and Y and Z
[04:17:56] <AchiestDragon> well i guess that some error may be too hard Z is giong to be a non started if say z is one degree out then you cant realy drill down z giving an offset correcton for the angle as would mill it as a elongated hole
[04:18:42] <AchiestDragon> but if the xy is off suare then you could compencate the position in software
[04:18:53] <AchiestDragon> suare square
[04:19:40] <AchiestDragon> started / starter
[04:20:01] <AchiestDragon> too late for typing for me
[04:20:03] <toastyde1th> depends on how flat you need it to be
[04:20:45] <toastyde1th> i have had relatively poor success with software correction of anything but cosmetic surfaces for mills with .0001" resolution
[04:20:48] <AchiestDragon> depends on how far out the machine is
[04:21:15] <toastyde1th> i have seen pictures of software-corrected surfaces on mills with .00001" resolution, looked a lot better
[04:21:28] <toastyde1th> but there's a pretty obvious optical step
[04:22:04] <toastyde1th> often it winds up not moving at all until a couple tenths are registered, then it jumps like, .001"
[04:22:11] <toastyde1th> stiction is a huge problem
[04:22:31] <AchiestDragon> if the machine is out by say 2mm over 300mm then the step would be a better option in most cases than the job beeing 2mm out at the ends
[04:23:07] <toastyde1th> i guess, what machine is 2mm out in just 300mm
[04:23:27] <toastyde1th> personally I'd try to correct the machine
[04:24:20] <AchiestDragon> but by the same token for hobby builds the correction would at least make there machine usable even though rough rather than scrap ,, but if they built it and did not allow for adjustment then there often stuck
[04:24:29] <toastyde1th> sure
[04:24:45] <toastyde1th> also i guess to be clear I wasn't saying DON'T include correction
[04:24:50] <AchiestDragon> and have to say i do aggree that it will never be as good as a properly alligned machine
[04:25:08] <toastyde1th> just listing some caveats
[04:25:44] <toastyde1th> correction on my linear way lathes works pretty well, no stiction and high resolution
[04:26:03] <toastyde1th> but the very accurate, way based lathes, not so much
[04:26:17] <toastyde1th> but they're accurate to begin with so whatever
[04:29:05] <AchiestDragon> it would save a lot of first time home builders form having to scrap there machine when they may not even have antibacklash on it eather so the amount of error you are talking about would be irrelivent anyway
[04:29:48] <tom3p> i wanted to tilt the plane so a simle program of Z-1 actually moved at some angle. we do this on sink edms all the time, the tool (carbon) is machined at that angle, and the motion matches it. the machine is fine and really square. its just simplified programming.
[04:29:59] <toastyde1th> tom3p: that is cool
[04:30:31] <tom3p> toastyde1th: look at a heidenhain 4xx or better, its built in
[04:31:16] <AchiestDragon> with most of the stuff i have seen produced on home made routers like are like flat cutout parts from balsa acurate to +-0.5mm is good enough but its more inportant that the xy is square , if not then there parts dont fit
[04:31:40] <toastyde1th> achiestdragon: i never said to not include it
[04:31:45] <toastyde1th> or that it didn't have uses
[04:32:07] <AchiestDragon> just giving reasons for inclusion
[04:32:09] <AchiestDragon> :)
[04:34:45] <toastyde1th> that's numberwang
[04:35:38] <AchiestDragon> theres a lot of nice designs for say wood made cnc routers ,, some realy good ones , unfortunlatly to build one well you need a cnc to make them on ,, and most people building there first machine are doing so becasue they dont have a cnc
[04:36:45] <AchiestDragon> its not an easy task to make a cnc with only basic hand tools and maybe an electric hand drill
[04:37:14] <AchiestDragon> but thats the only way for some people
[04:38:27] <toastyde1th> GIT OUT YER SCRAPER
[04:39:48] <tom3p> BRING OUT YOUR DEAD (cnc's) cnc's for the people! yay!
[04:39:55] <AchiestDragon> i did not even have a pillar drill when i built my first cnc ,, hacksaw , mig welder , electric drill and spanners and a file
[04:40:22] <tom3p> you'll have to explain pillar drill to us yanks ;)
[04:40:38] <AchiestDragon> drill on a stand
[04:40:59] <AchiestDragon> electric drill as in hand held one
[04:41:02] <ds3> isn't 0.080 over 12" a bit much for a metal working tool?
[04:41:21] <tom3p> yeah i know but uncommon term in US
[04:42:36] <AchiestDragon> ds3 yea ,,, but if you dont alow for adjustment and you weld it slightly wrong or dont allow for welding distrotions then you can end up with that
[04:44:04] <toastyde1th> drill press?
[04:44:06] <AchiestDragon> i did spot the fact that i would not be able to get the allignment so i accounted for 10mm just over 1/4" of build inacuracy in the adjustments
[04:44:09] <tom3p> .007/inch is better than carpentry, as good as cabinetry, not very good for metal work. you'd flunk the file a cube test in metal working 101.
[04:44:29] <toastyde1th> i wonder if you can make a straightedge with wood and some glue
[04:44:33] <toastyde1th> for scraping
[04:44:40] <toastyde1th> if you're making a wood machine.
[04:44:49] <tom3p> a pillar drill is a bit diff from what we call a drill press, but it is a drill, guided better
[04:44:56] <toastyde1th> oh.
[04:46:54] <toastyde1th> so like, say you took some .75"x4"x24" mdf panels, and glued them together so you got yourself a nice chunk of wood
[04:47:02] <toastyde1th> then started to build up a glue surface on one edge
[04:47:12] <toastyde1th> i bet you could scrape that and make yourself a straightedge
[04:48:23] <AchiestDragon> was looking at someones cnc build on cnczone couple of days back ,,, he had got a big section of i beem welded 2 bits of angle iron to it for the bearing runners and started to build up quite a nice frame
[04:49:27] <AchiestDragon> then sudenly the updates on the posts stopped ,, bet he found out the hard way that he should of made those angle iron sections he used for the rails adjustable
[04:49:28] <tom3p> weird site, its a library of sounds , and had a pillar drill http://www.freesound.org/samplesViewSingle.php?id=58497
[04:49:41] <tom3p> well thats enuf, gnite all
[04:50:14] <toastyde1th> baiu
[04:50:16] <toastyde1th> oops
[04:50:51] <AchiestDragon> http://cgi.ebay.co.uk/PILLAR-DRILL_W0QQitemZ180418767408QQcmdZViewItemQQptZUK_BOI_Building_Materials_Supplies_Carpentry_Woodwork_ET?hash=item2a01cbea30&_trksid=p3286.c0.m14
[04:51:36] <AchiestDragon> not shoure if thats what you would call a drill press ,, but its called a pillar drill in the uk
[04:53:45] <AchiestDragon> well almost 6am here ,, about the right time to get a mc'd's breakfast for supper , but thinks i will give that a miss today ,,, nn
[04:54:22] <toastyde1th> yeah, that's a drill press
[04:55:15] <tomp> tomp is now known as tom3p
[04:56:32] <tom3p> last note: this guy cnc'd a smll drill press, i was thinking jig drill when i heard pillar drill, and this guy also has made some servo motors http://www.flyelectric.ukgateway.net/motors.htm
[04:56:45] <tom3p> gnite^2
[05:07:41] <NTU> Hi. I am wondering about what DRM / Mesa / DDX code conflicts with the RTAI kernel patches since the current DRM code is now "in-kernel." I have dug through the documentation and the wiki but I cannot find the source to the diffs nor patches themselves of the RTAI kernel patch source. My english isn't too great so please let me know if I need to clear anything uo. ;) Thanks
[05:09:51] <NTU> I found the b-zip tars for RTAI but I dont see any source. Makefiles and configure files only
[05:15:30] <NTU> Oh I found it, base > arch > i386 :) The scheduler code and pre-emitable (low latency) patches shouldn't effect anything related to DDX rendering, Mesa, and drm code, right?
[05:46:08] <NTU> NTU is now known as Neo_The_User
[06:23:14] <micges_work> hi all
[06:23:56] <Neo_The_User> hi
[06:27:42] <alex_joni> hi
[06:28:22] <jymm> lo
[07:17:02] <MarkusBec_away> MarkusBec_away is now known as MarkusBec
[08:08:13] <pjm> good morning
[08:09:24] <archivist> Model Engineer ex tomorrow
[08:21:25] <pjm> damn, i am working ;-(
[08:21:49] <pjm> i did spend an hour last night getting to grips with my 7i43 which is now running fine with EMC2
[08:22:04] <pjm> just need to get all the pins linked and the board integrated into the mill
[08:22:41] <archivist> my mill is in bits being moved home
[08:23:53] <Valen> good chance to clean it all up?
[08:28:05] <archivist> a bit
[08:28:29] <archivist> the move from work is a huge shift though
[08:29:23] <Valen> yeah
[08:29:31] <Valen> moving mills is never fun
[08:29:46] <archivist> the PC part of the mill is running 24/7 compiling and testing
[11:56:35] <Valen> The Ohnosecond, defined as the amount of time between clicking yes on a delete prompt and the realisation that you picked the wrong RAID set.
[11:58:02] <archivist> followed by the resignation letter writing hour
[11:58:24] <Valen> that comes after the backupsearch period
[11:59:45] <archivist> there is no backupsearch, raid is always too large to backup
[12:00:09] <Valen> thats what the stationwagon full of tapes is for
[12:00:13] <Valen> but yeah, i hear that ;->
[12:00:36] <archivist> the number that ask in #mysql after they have hit the delete is fun
[12:01:17] <Valen> you would probably be more screwed in a DB than on a file system
[12:01:44] <archivist> about the same depends on tools and brains
[12:02:22] <Valen> I feel there would be less tools for a db than a fs?
[12:02:59] <archivist> probably, depends on db
[12:03:15] <archivist> and db retore may include fs restore
[12:04:09] <MarkusBec> MarkusBec is now known as MarkusBec_away
[12:04:11] <Valen> I always dump the db before i do anything dodgy ;->
[12:04:23] <Valen> and if I cant do that, snapshotting the VM is made of win
[13:05:55] <archivist> robh, I had a word with the arc eurotrade owner about a meeting place at the show
[14:39:24] <AchiestDragon> ok seems i have a good idea on how to impliment the controls now
[14:40:30] <AchiestDragon> basicaly need to provide the estop and machine monitor controls like this does http://www.cnc4pc.com/Store/osc/product_info.php?products_id=181&osCsid=42bdf840623427575a71afda652e46bb
[14:41:56] <AchiestDragon> fpga control in a simlar way to the mesa 7I43/H
[14:43:25] <archivist> bit of thinking and you can do all that for less
[14:45:34] <AchiestDragon> yea since using a fpga should be able to build drives onboard also
[14:46:15] <AchiestDragon> 2 hbridges could drive a stepper or used as a single to drive a servo
[14:46:55] <AchiestDragon> ok so the drivers are going to be limited to 15A at 50v depending on the size of the heatsink
[14:47:46] <AchiestDragon> but would suit 4.5A steppers or servos used on most hobby machines without a problem
[14:48:03] <tomp> inside the fpga?
[14:48:47] <AchiestDragon> no extrenal fet final drivers but the other driver functions in the fpga
[14:49:18] <AchiestDragon> no / no ,
[14:49:58] <tomp> ok, h bridge control inside fpga, there are some open source cores you might use
[14:50:08] <tomp> tomp is now known as tom3p
[14:51:15] <AchiestDragon> yea ,, hbridge control is simple ,,, will take me a bit longer to code a core for microstepping but full stepper control is not a problem eather
[14:51:26] <AchiestDragon> as in coding the vhdl for me
[14:52:34] <AchiestDragon> the main issue i have been trying to find a way to make it cost effective for a hobby constructor and home buildable
[14:53:08] <AchiestDragon> 3 axis with a second addon board to make it 6 is the goal
[14:53:08] <archivist> current control is the hard part
[14:54:58] <AchiestDragon> the 40 pin versions of this module http://www.enterpoint.co.uk/component_replacements/craignell.html 100k gate ones are £35 each
[14:55:59] <AchiestDragon> using one would give 3 axis , another linked would be on the expantion to 6,, should allow the pcb needed to use them to be double sided so could be milled on a cnc router
[14:57:08] <AchiestDragon> and choce of external components can then be eather std dip/ though hole or the at least easy to manage large surface mount that a hobby user would not have a problem assembling
[14:57:46] <skunkworks_> http://cnczone.com/forums/showthread.php?t=81125&highlight=vhdl
[14:57:55] <skunkworks_> he goes thru a micro-stepping example
[14:58:02] <skunkworks_> (mariss)
[14:58:46] <AchiestDragon> wont be usb but parrallel ,, once i get it going will have to work with emc devs to sort out a communications protocol
[14:59:55] <archivist> AchiestDragon, you write the other side as well :)
[15:00:05] <skunkworks_> This is an example of my H-bridge in cpld - original circuit http://cnczone.com/forums/showpost.php?p=596446&postcount=67
[15:00:11] <skunkworks_> http://cnczone.com/forums/showpost.php?p=596857&postcount=72 mariss conversion
[15:00:53] <AchiestDragon> will make it a gpl hardware /softcore design so would allow any willing compaines to manufacture it or make compatable units
[15:04:05] <AchiestDragon> archivist: will probablay see what emc2 can support and code the vhld for it but rather than say sending to a usb device you may need to send data packets to a p port
[15:07:09] <AchiestDragon> usb from other experiance with it is not sutable ,,, ethernet would be other than cost and complexaty of the interface
[15:09:11] <AchiestDragon> parrallel or serial works ok but parallel control would need to be muxed in some way to allow single port use for 6 axis control
[15:10:07] <AchiestDragon> its not a practilacl solution to fit 2 parallel ports to some pcs
[15:10:53] <AchiestDragon> the only option some offer is usb to parrallel
[15:12:48] <archivist> usb2 does have a bidirectional synchronous mode, just needs locking down to stop new device handshake time
[15:12:50] <AchiestDragon> the gate count in the fpga modules wont stretch to adding a usb core
[15:14:51] <AchiestDragon> but it would stretch to adding some command buffer to resync the data timing , although there would be a lag between the command the maching is doing and the commands that the pc is sending
[15:16:00] <AchiestDragon> but the controler would have a bit more "inteligance " than the usual p port method
[15:17:35] <AchiestDragon> anyway atm the hardware needs to be designed ,, the coding and opperational requirements can be sorted out later and adapted where needed
[15:20:19] <AchiestDragon> i have no porblem doing the hardware or vhdl for it but my c++ would not be up to coding a driver for the pc side
[15:20:47] <archivist> its mostly C as far as I know
[15:21:05] <AchiestDragon> not done any c in 18 years
[15:26:35] <NTU> NTU is now known as Neo_The_User
[15:35:12] <Neo_The_User> Does anybody know the reason why the closed source nvidia drivers conflict with the real-time kernel in 8.04? The reason I ask is because I wonder what else as far as graphics go, wouldn't work. Thank you so much in advance! :)
[15:36:37] <archivist> define conflict, it causes latency
[15:37:05] <tlab> AchiestDragon, http://www.fpga4fun.com/CNC.html
[15:39:04] <Neo_The_User> Oh I could see that. Thanks. Lightweight 3d code wouldn't have the same problems would it?
[15:39:21] <Neo_The_User> like erm.... mesa + openchrome
[15:40:30] <archivist> they dont care about delaying the user as updating the display is important to them, its the other way round in a cnc the display should be delayed
[15:40:51] <archivist> not a weight question
[15:42:23] <celeron55_> hmm, do the open source drivers make less delay then?
[15:42:33] <Neo_The_User> I would assume so
[15:42:46] <AchiestDragon> would be the other way arround wouldent it ,, the commands sent form emc to the machine would get buffered ,, so the mahcine is a few instructions behind the commands that emc are displaying
[15:43:17] <archivist> AchiestDragon, the control is IN the pc
[15:43:28] <SWPadnos> AchiestDragon, there is no buffering between EMC2 and the machine, it's got to be realtime both outgoing and for feedback
[15:43:53] <AchiestDragon> was going to handle the feedback in the fpga
[15:44:14] <SWPadnos> celeron55_, the open source drivers, combined with the Mesa openGL library, don't exhibit significant delays like the closed source drivers sometimes do
[15:44:36] <Neo_The_User> thank you SWPadnos you answered my question too :)
[15:44:45] <SWPadnos> though there have been a couple of reports of latencies getting better with the nVidia drivers rather than the nv or vesa driver
[15:44:48] <SWPadnos> so YMMV
[15:44:57] <SWPadnos> Neo_The_User, any time ;)
[15:45:15] <SWPadnos> AchiestDragon, you can do that, but it depends on what feedback you're talking about
[15:45:25] <tlab> AchiestDragon, you wanting to make a standalone fpga cnc?
[15:45:32] <celeron55_> i wouldn't trust rely upoin closed source code that isn't meant to give short latencies anyway
[15:45:37] <SWPadnos> and it gets a lot more complex if you want to support threading or other spindle-synchronized motion
[15:45:42] <celeron55_> -trust
[15:46:26] <celeron55_> on the other hand, some people are using windows and mach3 or similar 8)
[15:46:52] <AchiestDragon> could alamost go as far as issueing the fpga with a stripped subset of g code goto/feed speed command but think with 100k gates going to be a bit pushed to do much more
[15:48:14] <AchiestDragon> curves are going to involve a stream of comands to it but the communications rate should not be a problem
[15:50:16] <Neo_The_User> SWPadnos: I can have multithreaded mesa going. Seems fast as far as latency goes
[15:50:53] <Neo_The_User> heh I used to have that on my Playstation 3. Worked great with compiz :P
[15:50:59] <SWPadnos> heh
[15:56:23] <AchiestDragon> ie given that 1 "step" = 0.1mm and a starting pos on 3 axis of 0,0,0 you could have a command structure like "move 10,100,0 over 10 seconds ,, converting that to 10/10 to one step per second on the x 100/10 to 10 steps a second on y and avoding the div by zero on z with phase locked timing in hardware is not a major issue
[15:58:25] <AchiestDragon> while the fpga is performing that command the next one gets buffered to it , and once compleated the first it would start on the next
[16:00:06] <AchiestDragon> actualy doing it that way could be achived just by a file preporcessor/converter other than you need user intrerface and some feedback
[16:01:24] <AchiestDragon> as all the actual machine timing would be phase locked and handled by the fpga removing the need for real time timing constrants in the pc itself
[16:03:37] <tom3p> anyone use modo/mindi for backup/recovery?
[16:03:48] <tom3p> mondo/mindi
[16:04:07] <Neo_The_User> I use rsync...
[16:08:44] <cradek> amanda
[16:20:05] <jepler> dump/restore
[16:20:30] <archivist> cp (just the important files)
[16:25:57] <celeron55_> i use rdiff-backup
[16:26:24] <celeron55_> and cp too 8)
[16:26:59] <tom3p> varied favorites!, will try one now
[16:29:35] <MattyMatt> 14" blade on this saw. 9A motor
[16:30:32] <MattyMatt> I think the blade might be too big. the height adjuster leaves more than an inch of blade showing at lowest
[17:06:01] <MattyMatt> interesting extrusion used for the guide. it's L shaped, with a T slot on the upper side of the bottom stroke, and the outside of the upright. window frame?
[17:18:24] <AchiestDragon> SWPadnos: well if steppers then its open loop so feedback would be with internal step count , for servos its closed loop so the fpga would count the optical encoder "steps " for its position
[17:19:16] <SWPadnos> AchiestDragon, every few months, someone suggests making an FPGA to replace the EMC2 motion controller with realtime hardware
[17:19:37] <archivist> we need a wiki page
[17:19:47] <SWPadnos> in all cases so far, there has not been a solution that gives the flexibility and features of EMC2 in an FPGA
[17:19:49] <AchiestDragon> whats the reasons for them not compleating one ?
[17:19:57] <SWPadnos> there probably is a wiki page
[17:20:29] <SWPadnos> there are several devices that have some features, like step generation and the like, and they're on the market already not wth EMC2 drivers thouhg)
[17:20:31] <SWPadnos> though
[17:20:45] <SWPadnos> smoothstepper is one, the Geckodrive G-Rex is another
[17:20:46] <AchiestDragon> i can see a few issues that cause common flaws ,, but also know some workarounds
[17:20:54] <SWPadnos> there's a (probably vaporware) "CNC Brain"
[17:21:03] <SWPadnos> etc. etc
[17:21:05] <archivist> inertia, and the needed redesign of emc to work that way
[17:21:31] <SWPadnos> so far, all of them fall flat when you add jogwheels, threading, rigid tapping, slaved axes, and servo PID to the mix
[17:22:03] <SWPadnos> there is another one that more or less does everything, called the ncpod
[17:22:28] <SWPadnos> that one uses a version of the emc2 interpreter/motion planner to spit out positions to a USB device
[17:22:41] <SWPadnos> I don't remember the problems with that one
[17:23:05] <SWPadnos> in all cases so far, the people designing the device say "look, step generation, we'll have threading soon"
[17:23:19] <SWPadnos> and soon never arrives, and PID/PWM servo control are never even on their raday
[17:23:21] <SWPadnos> radar
[17:24:05] <SWPadnos> if you can get it to work, more power to you, I'm just pointing out that you're not the first to have the idea, and it's not as simple as it seems at first
[17:24:34] <archivist> it would never fit in a cheap fpga imo
[17:25:11] <SWPadnos> not a $2 one, but maybe a $15-20 one
[17:26:16] <archivist> I do see merit in a mitsubishi? idea where they communicate servo to servo so they can lock the trajectory
[17:26:56] <AchiestDragon> i dont look at the problem as a pid/pwm problem ,, its better solved as a pwm/pll problem
[17:26:57] <archivist> one of the cheap stepper pic designs has a similar linkage
[17:27:24] <AchiestDragon> the timing generation is pll using pwm for the rate and drive power
[17:27:36] <archivist> AchiestDragon, its the linking of ALL drives that we are talking about, not individual
[17:28:14] <archivist> if X is slowed by load, others must follow
[17:28:50] <AchiestDragon> yea the linking of all drives is the need the issuing of a move to comand with a time for that command is needed
[17:29:16] <archivist> you still not seeing the reaction and ALL following
[17:30:02] <archivist> playtime...
[17:30:05] <AchiestDragon> ie you can give 3 diferent motors diferent values and the time it should take to get to that point ,, the fpga would do the math to produce the actual rate of each individual axis
[17:31:07] <archivist> x feedback affects y,z,....
[17:31:21] <SWPadnos> and when the user moves the feed override slider, what happens?
[17:31:29] <AchiestDragon> you can use tricks to compencate when one motor slows its posible to decrese the feed on the others by clock speed control thats phase locked to the actual motion of the machine
[17:32:03] <SWPadnos> having rapid (nearly instantaneous) response to changing trajectory planning conditions is a requirement, which eliminates the possibility of buffering or sending long moves to the FPGA
[17:32:39] <SWPadnos> then there's probing, or other motion that needs to be stopped (or modified) by external events
[17:32:55] <AchiestDragon> you derive a pll referance clock ,, if the feed rate slows dew to load then the logic slows the referance clock that in turn slows the feed of all axis movments
[17:33:48] <AchiestDragon> making the effective result a closed loop system locked on all axises
[17:33:50] <SWPadnos> you're talking about the mechanics of generating steps, which is a trivial sub-project of this non-trivial project
[17:34:56] <SWPadnos> if you severely restrict the feature set that the user would have available, then it is possible to make an FPGA-based motion controller, and make it compatible with EMC2
[17:35:19] <AchiestDragon> i usualy find it easyer to do a state machine in logic than to program a micro to do the same
[17:35:33] <SWPadnos> but that goes against one of the interesting and valuable features of EMC2, which is that you get all the features we have, regardless of your hardware
[17:35:58] <MattyMatt> a cpu would be more useful
[17:36:10] <MattyMatt> pic even, they have plenty of IO ports
[17:36:15] <SWPadnos> changing hardware may increase performance or speeds, but even a lowly parallel port can do rigid tapping, threading, probing, and classic ladder
[17:36:31] <SWPadnos> and servo PID
[17:36:38] <L84Supper> anyone know if the latency tests on the VIA c7 cn700 boards were done with the onboard graphics or PCI card gpu?
[17:36:42] <SWPadnos> and jog wheels, and feed override, and anything else
[17:37:04] <MattyMatt> hard end stop switches
[17:37:52] <SWPadnos> MattyMatt, that's a hardware problem ;)
[17:38:52] <MattyMatt> not if the cpu is at the servo end of the cable. it's "hard" inasmuch as the controller PC can't crash it
[17:39:23] <AchiestDragon> cpu one with a std instruction set like a avr or simlar would not be a problem ,,, but a full instruction set is not needed for the task , and the application is better suted to a state machine design as its then parrallel and can get better timing resouluton in hardware than using software loops
[17:39:46] <AchiestDragon> other than a avr core would probablay use up more gate space
[17:41:18] <MattyMatt> yeah polling every limit switch inside a timing loop could bring even a 50Mhz machine to its knees
[17:41:44] <MattyMatt> although they'd all be bits in one word, so maybe not
[17:42:24] <MattyMatt> something to try with PIC. this eprom programmer I got does them too
[17:42:32] <L84Supper> getting random jitter on the latency test out to 300000, no regular interval, 10-15 min apart to 9 hrs
[17:43:45] <MattyMatt> someone in your street puts the kettle on every 10-15 mins?
[17:44:27] <L84Supper> maybe it's that switch on the wall that doesn't seem to do anything at work :)
[17:45:20] <AchiestDragon> kettle ,, central heating system , or even celphone causing emi ,, you need to add filtering to the input power , and maybe concider screening the cables and equipment housings
[17:47:25] <archivist_attic> logger_emc: bookmark
[17:47:25] <archivist_attic> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-10-15.txt
[17:49:10] <AchiestDragon> such interferance can be eather a/. relativly local ie within 200 to 300 yards ,, or a more remote power supply glitches ,, ie a 30kw compressor 200 miles away kicking in every 10 to 15 mins
[17:49:30] <L84Supper> kernel log mentions RTAPI: ERROR: Unexpected realtime delay on task 1
[17:51:34] <L84Supper> if it's just EMC then it's no problem, I was concerned about SMI or onboard GPU acting up
[17:52:15] <L84Supper> I'm currently testing it without a case or shielding of any kind
[17:52:21] <SWPadnos> it's unlikely that it's EMC2 or RTAI
[17:52:30] <AchiestDragon> SWPadnos: although at one end doing a almost total fpga design is posible i was thining of implimenting it as more of a semicoprocessor type system were it sends a command to keep the machine running , and one for the next step in a buffer to pipleine it but while also providing feedback to emc on its progress and alowing for overides
[17:52:49] <SWPadnos> it could be hard disk, network, or video drivers
[17:53:01] <SWPadnos> sure
[17:53:14] <SWPadnos> you're talking about the Mesa or Pico Systems products then
[17:53:26] <SWPadnos> take a look at the Mesa VHDL, it's in the EMC2 repository
[17:53:59] <AchiestDragon> will do but was actualy looking at a fresh hadware design
[17:55:02] <SWPadnos> sure, it's definitely a fun project to hack on
[17:55:50] <SWPadnos> it's just not too necessary IMO - the Mesa hardware is $89 for the FPGA board (with a 400k gate FPGA), and another $70 or something for some isolated I/O
[17:55:53] <archivist_attic> I want the alleged impossible, usb2
[17:56:00] <MattyMatt> I saw the cost of VHDL modules. $1700 for a multiplier
[17:56:06] <SWPadnos> the average person wouldn't be able to put one together for that little
[17:56:23] <SWPadnos> opencores
[17:57:12] <AchiestDragon> one board with drivers to support 3 axis stepper or servos and an expantion to provide another 3 axis driver support ,,, ok so limited to 15A at 50v steper/servo power but that would fit most small /mid hobby mill conversions
[17:57:24] <MattyMatt> mp3 players have usb2, z80 and dsp
[17:57:34] <MattyMatt> and cost $20
[17:57:46] <SWPadnos> and are made in million quantities
[17:58:45] <AchiestDragon> the fpga module is £35 for one with a 100k gate chip , in a 40 pin dip module the rest of the pcb will be eather the larger type of smd res caps etc and dip type chips that would allow for home construction
[17:58:45] <MattyMatt> they even have a dinky display
[17:59:32] <SWPadnos> 35 UKP is almost the cost of the entire 7i43 board, with its 400k gate chip
[18:00:35] <L84Supper> I'm working on a 400k fpga + ARM9 board now about the size of the 7143
[18:01:21] <MattyMatt> there's a 150k board that plugs into a GBA, but it's $199
[18:02:14] <L84Supper> 400k is large enough for 32bit soft micro and stepper controllers
[18:02:18] <AchiestDragon> the 7i43 is $161 with the 100k device according to there website ,,, $161 is about £70gbp plus 15% vat and a £40 customs processing fee if by ups
[18:02:42] <SWPadnos> what
[18:02:47] <AchiestDragon> makes it more like £160 gbp
[18:02:58] <L84Supper> 7i43 is $99 USD
[18:03:02] <AchiestDragon> 4I34-1
[18:03:02] <AchiestDragon> Anything I/O 64 bit 100k gate 161
[18:03:14] <SWPadnos> it's $79, 89, or 99 depending on the options
[18:03:25] <archivist_attic> 99 1 off
[18:04:52] <AchiestDragon> $161 http://www.mesanet.com/ there price page looking at it now the 200k gate one is $184
[18:05:28] <SWPadnos> why are you looking at the 4i34 when I was talking about the 7i43?
[18:05:30] <AchiestDragon> thats the 1 to 4 off price
[18:05:39] <archivist_attic> look at the right model number AchiestDragon
[18:06:10] <AchiestDragon> there moddel numbers are confusing
[18:06:29] <SWPadnos> yeah, and it would also be nice if they were in alphabetical/numerical order :)
[18:07:02] <AchiestDragon> yea ,, like one digit out and you end up with a washing machine
[18:07:09] <SWPadnos> heh, yep
[18:07:16] <archivist_attic> or on the page with the item
[18:10:34] <AchiestDragon> still $89 is £54 +15% plus the £40 customs processing that ups charge works out at more like £100
[18:11:23] <archivist_attic> I wonder if there is mileage in a brit stocking some
[18:11:37] <AchiestDragon> its the custom charge that ups charge you for processign the vat before delivery thats the killer ,, not so bad if your ordering 200 of them as its still £40
[18:12:19] <archivist_attic> I didnt have to pay ups anything last time I imported something
[18:12:22] <AchiestDragon> but it does put a damper on individual orders from hobby users
[18:13:06] <AchiestDragon> if theres more than £5 vat content there will be a charge ,, basicaly make it payable on any goods over £45
[18:13:54] <AchiestDragon> if theres £6 to pay in vat or £60,0000 ups will demand £40 for processing it to be payed before they will relese the items for delivery
[18:14:37] <AchiestDragon> some other carriers have diferent rates ,, but usualy the carriere used is at the discression of the supplyer
[18:15:14] <AchiestDragon> been stung by ups twice for that
[18:15:30] <tomp> tomp is now known as tom3p
[18:15:48] <AchiestDragon> i dont object to the vat but its extortion on small amounts
[18:16:01] <AchiestDragon> the proccessing fee that is
[18:16:50] <L84Supper> we had to pay $275 for customs clearance on a box of sheet metal sent from China since they just sent it Cathay pacific vs DHL
[18:16:50] <AchiestDragon> local supplyers that inport usualy dont resell for much less than what it would cost you to inport one with the fee
[18:17:22] <L84Supper> talk about gouging by customs brokers
[18:18:16] <AchiestDragon> the efika a nice littel ppc based board is $99 actualy would be good value at that but by the time you get one inported its cost you more like £150
[18:19:02] <AchiestDragon> making it expencive compared to say a mini itx system
[18:19:36] <AchiestDragon> http://www.enterpoint.co.uk/component_replacements/craignell.html
[18:20:06] <AchiestDragon> there made in the uk the cr40-100 is £35gpb
[18:20:43] <L84Supper> I was about to order some of those
[18:20:47] <AchiestDragon> its a 40 pin dip module with a 100k gate fpga and 5v tolarant leve shifters
[18:20:59] <L84Supper> any experience with them?
[18:21:57] <AchiestDragon> there good boards , got some of there bigger pga modules here ,,, only down side is there software support leaves you haivng to create your own generic ucf file
[18:23:12] <AchiestDragon> would advise that you get a program adaptor and a programming cable from them also ,, works with the usual xilinx ise webpack (free download from xilinx)
[18:23:21] <L84Supper> http://apple.clickandbuild.com/cnb/shop/ftdichip?productID=80&op=catalogue-product_info-null&prodCategoryID=5
[18:24:08] <pjm> AchiestDragon i just got a 7i43 plus a couple of the IO cards, very pleased with it so far, I'm just at the start of converting my machine from parport to proper IO. Finally I'll be able to read my spindle encoder when the machine is running flat out!
[18:24:13] <L84Supper> similar to the craignell but with FTDI usb controller
[18:24:14] <AchiestDragon> avoid the linux version of the webpack ,, its too buggy to use , the windows version is fine other than needing windows , could say that makes it worse ,
[18:25:42] <AchiestDragon> well since theres emc support for the 7i43 think the way to go that would be beter for all is for me to do the hardware for the one i am planning that would appeal to the eu guys that dont want to have to pay the extra because of customs
[18:26:25] <AchiestDragon> when it comes to the vhdl is develop compatable code that works with eather
[18:26:31] <pjm> i didnt have anything at all to pay with my shipment
[18:26:39] <pjm> and i'm in rip-off britain too
[18:26:54] <archivist_attic> you cheated :)
[18:27:02] <pjm> well no, i proxied it
[18:27:06] <pjm> its not cheating!
[18:27:22] <L84Supper> is the VAT collected there hit or miss?
[18:28:14] <archivist_attic> duty and vat is at import
[18:28:17] <pjm> good question, i try to avoid extra taxation where ever possible
[18:28:20] <AchiestDragon> determined by whats on the customs declaration ,, unless they open the package as part of there checks then you can be in serius trouble
[18:28:33] <pjm> not if its 2nd user stuff
[18:30:13] <AchiestDragon> because of those in china selling you a £20,000 machine , charging you £1 for it and £19,999 shipping then they now charg you vat if its a free gift over worth over the £5 vat amount
[18:32:21] <tom3p> 1) mark it parts 2) mark it used (never buy whole machines, and have it tested, then both are true )
[18:32:46] <AchiestDragon> usualy you would get away with it , but they do random spot checks and if its not what it says on the customs delclaration then you would probablay spend less time banged up if you had robed a bank for the cash for it
[18:35:33] <AchiestDragon> i only order parts direct if its worth paying the charges ,, but in a lot of cases it makes them too expencive to be worth wile ,, on ebay you usualy find a uk supplyer selling the same part for the same sort of price (given vat ontop in with that )
[18:37:00] <AchiestDragon> faster delivery also 1 to 3 days from anyware in uk ,, 5 to 12 days from china , and for some reason 12 to 26 days from the states is usual dilivery times ,, with the exeption of ny where it can be 3 to 4 days
[18:37:55] <AchiestDragon> the other 3 to 4 day one to here is hongkong
[18:39:36] <AchiestDragon> customs processing usualy adds another day to it as they send you a letter to tell you what to pay , you have to phone them when and sort out payment then only then will they load it for delivery
[18:39:49] <AchiestDragon> pita
[18:41:34] <AchiestDragon> L84Supper: yea that does not look too bad a board ether
[18:43:41] <AchiestDragon> pjm should hope so the p port is nothing like as fast as what you can achive on fpga i/o pins ,, upto 165mhz is posible
[18:44:44] <AchiestDragon> where you would be struggling to manage 1mhz on a printerport
[18:46:16] <AchiestDragon> the problem with the ftti chips is there basicaly a usb to pio,, fine at lower speeds but once you want a constant rate real time stream there limit is about 4mhz , above that and you start to get problems
[18:46:34] <AchiestDragon> dew to the fact that usb is a packet based protocol
[18:49:39] <archivist_attic> so is ethernet, but people are happy to use that
[18:50:09] <archivist_attic> if packet rate is fast enough, there is no problem
[18:50:12] <AchiestDragon> at least with ethernet you are working with a set speed packet
[18:50:27] <archivist_attic> usb2 has a fixed rate mode
[18:51:17] <archivist_attic> its a turgid read but may I commend the spec as bedtime reading
[18:51:32] <AchiestDragon> fixed rate ,,, maybe if the usb on the machine is actualy 4 seperate ports and not as is often a single port with a hub on the board
[18:52:10] <AchiestDragon> ie a usb2 device may opperate at a few speeds so you set it to a fixed one , fine for coms with that
[18:52:14] <archivist_attic> one has to specify use mode for cnc
[18:52:22] <L84Supper> these discussions always go around in the same circles
[18:52:42] <archivist_attic> and few bother to read the spec
[18:53:26] <AchiestDragon> but if on the same machine you have a usb mouse then you find that the time it takes the mouse packet thats using a slow usb 1 packet on the same hub then it interupts the avalable time that is avalbale on the bandwith
[18:53:51] <AchiestDragon> unless you happen to have a usb 2 mouse that will also support that fixed rate mode
[18:54:37] <celeron55_> if that's the problem, why not just connect nothing else than the cnc controller to the usb ports
[18:55:41] <archivist_attic> that what ive being saying use must be controlled to avoid pitfalls
[18:55:42] <AchiestDragon> yea great , unfortunatly that may or may not be posible ,, and more of the problem is then you find that most of the time you end up doing tec support for those that have failed to understand it
[18:56:55] <celeron55_> sometimes there are things connected to usb integrated on the mainboard, too 8)
[18:59:02] <archivist_attic> be the same as now with integrated video, some wont work
[19:03:07] <L84Supper> just whip up an open PCI or PCIe core :)
[19:04:15] <L84Supper> or Hypertransport , then you can just plug it into a dual socket mainboard
[19:05:06] <cradek> "October 5-10, 2009, marked the return of the EMO show to Milan, Italy, after a six-year absence. Since the last EMO
[19:05:09] <cradek> in Italy, the city of Milan constructed a new fairground"
[19:05:23] <cradek> this reminds me of the Wisconsin Tourism Federation logo
[20:48:48] <rob__> archivist, you around
[20:56:36] <archivist> yes
[20:57:55] <archivist> rob__, I will get there early and probably have to be kicked out at closing time
[20:58:17] <rob__> i should say i aim to get there around 11 or so
[20:58:40] <rob__> u thinking of meeting at the eurotrade, saw ur msg earlyer
[20:59:36] <archivist> yes I was going to say around 11 ish there is a cnc stand they are having this time, near there
[21:01:05] <archivist> we can laff at mach3 then :)
[21:01:30] <rob__> lol
[21:01:37] <archivist> had to come back to work, left the ticket here
[21:01:47] <rob__> erm yea if we say 11:30 to be safe as im sure ill be there by then def
[21:02:03] <archivist> ok /me is the hairy git
[21:02:06] <pjm> damn i will miss out on laughing at Mach3 ;-( have work tommorow
[21:02:18] <archivist> be ill!
[21:02:28] <pjm> hahh
[21:02:42] <archivist> I dont have a job I dont need to be ill
[21:02:49] <pjm> very good
[21:03:24] <cradek> will you be relaying the festivities over irc so us foreigners can hear what's going on?
[21:03:26] <rob__> thats ok next year we have mach2010 real machine show
[21:04:45] <archivist> can we get a freebee stand at mach2010
[21:04:56] <rob__> i hope so as i dont think we can afford one
[21:05:29] <archivist> would be interesting to be at the pro show with EMC
[21:05:33] <rob__> i wunder if hardinge will let us share, as iv nearly done one of there machines, we know one of the guys at handringe well
[21:06:11] <archivist> hmm emergency cant see ticket!
[21:06:12] <rob__> one year at mach there was centroid
[21:07:13] <archivist> * archivist calms down found ticket
[21:08:01] <archivist> cradek, will take pics anyway and will add another directory to my online archive
[21:08:26] <cradek> cool.
[21:08:55] <alex_joni> heh: http://www.linuxcnc.org/component/option,com_kunena/Itemid,20/func,view/catid,20/id,976/lang,en/#980
[21:10:34] <rob__> brb
[21:17:19] <archivist> http://www.linuxcnc.org/component/option,com_kunena/Itemid,20/func,view/catid,20/id,976/lang,en/#980
[21:24:13] <archivist> alex_joni, loads for me
[21:40:05] <tom3p> archivist: doesnt run as i expect, never moves down y
[21:41:49] <archivist> buggy g code no movement in the loop
[21:44:02] <tom3p> g0 y#29; dec #29...
[21:44:37] <cradek> the complaint is that it doesn't load
[21:45:21] <archivist> Im single stepping and adding debugs, it is a strange one
[21:45:38] <cradek> I'd expect it to move from 4 to 3
[21:45:48] <cradek> or maybe 3.2
[21:46:15] <archivist> I had to hack values to fit machine
[21:46:28] <archivist> but it loaded as was
[21:46:51] <tom3p> yes, loads fine
[21:46:53] <cradek> oyeah it goes from 4 to 3.2
[21:46:56] <cradek> -o
[21:46:59] <cradek> works fine
[21:47:01] <tom3p> only
[21:47:20] <tom3p> never again, how can we watch values of o words
[21:47:36] <archivist> (debug, x29)
[21:47:39] <cradek> (debug, variable 1234 is #1234)
[21:48:43] <archivist> am testing on 2.2.8 (because that box is on)
[21:48:54] <cradek> 2.3.3 here
[21:51:40] <archivist> hmm he never sets any standard settings like inch etc
[21:51:42] <tom3p> EMC2 - 2.4.0~pre here ( tahts waht the VERSION file says in the rip dir )
[21:51:53] <tom3p> i added g20 g90 f80
[21:52:46] <cradek> that gcode doesn't care what units you have
[21:53:01] <cradek> and g91 would change the behavior but it would still load
[21:53:18] <cradek> I bet he was not being careful and just posted the wrong code
[21:53:31] <cradek> and the first one he posted just has a bug (loop never exits)
[21:55:21] <tom3p> ran fine in non-rip, advances in y
[21:56:38] <tom3p> how do i find the version of the installed emc? ( the rips have VERSION files )
[22:01:40] <rob__> archivist, im off now, see you tomorow
[22:01:56] <archivist> see ya
[22:10:24] <micges> tom3p: python -> import emc -> print emc.version
[22:18:28] <tom3p> micges thx, its 2.2.5 ( why is it so NOT available! )
[22:19:53] <micges> tom3p: if you type emc in console it will also print version(just remembered;))
[22:21:56] <tom3p> argh! i'm running from a console and didnt see that, thx
[22:23:55] <tom3p> the (debug, y is #29) works also, didnt expect a dialog to popup ( where's the 'watch' window?) and (print, y is #29) runs the whole gamut at load and at run, even if you're single stepping.
[22:23:58] <tom3p> ok, but quite unexpected behaviour.
[22:25:07] <L84Supper> interesting VIA Epia mini-itx C7-D 1.5GHz vs C7 1GHz both with cn700 chipset, superIO, the C7 has far less jitter than the desktop C7-D cpu
[22:25:44] <micges> tom3p: readahead behaviour
[22:28:32] <tom3p> readtillend behaviour
[22:28:59] <micges> on short programs yes
[22:29:09] <micges> readahead is 1000 lines
[22:29:52] <tom3p> ah, thx i really like finity. didnt know it had a limit
[22:33:07] <micges> on 2.4 pre it also can be redefined in ini:
[22:33:07] <micges> [TASK] INTERP_MAX_LEN = 10
[22:33:56] <micges> (maybe usefull)
[22:33:57] <tom3p> i chgd his program to run from 4 to 2 inches in .001" increments, so it 'read ahead' only th e6 or so lines till end of programs, but the preview immediately showed the 2000 iterations ( the complete path )
[22:34:28] <tom3p> thx again
[22:34:51] <tom3p> ( that result was in 2.2.5 )
[22:36:52] <SWPadnos> the preview runs in its entirety before the code gets executed
[22:37:21] <SWPadnos> also, the interpreter runs significantly ahead of the motion controller, so you'll get a lot of debug output before the machine moves much
[22:43:08] <tom3p> after i run the rip version ( . scripts/emc-environment, then emc ) how do i run the installed version from command line?
[22:43:09] <tom3p>  ( it keeps running the rip version 2.4 pre installed of the installed 2.2.5 )
[22:43:53] <tom3p> where's the 'fergedaboudit' command?
[22:44:42] <micges> open new console and run it without emc-env sth
[22:45:30] <SWPadnos> and/or unset all EMC* variables (I think)
[22:45:31] <tom3p> yep! new console fixes that
[22:45:48] <SWPadnos> but you probably have to munge the path and stuff too, so a new console is the easier way
[22:46:18] <tom3p> heck even a new tab in the existing terminal works
[22:49:30] <SWPadnos> yep
[22:49:44] <SWPadnos> makes it easy to swap between several development versions
[23:15:21] <Guest865> Guest865 is now known as skunkworks
[23:50:41] <MarkusBec> MarkusBec is now known as MarkusBec_away