00:10:04 K4ts has joined #emc 00:10:26 hello 00:10:36 hello 00:14:46 hi alex. Hi Anna 00:14:53 ih les_w 00:14:58 hi les 00:15:06 happy new year 00:15:09 I just had a seafood dinner I prepared 00:15:13 ty anna 00:15:15 wowwww 00:15:28 cod and scallops 00:15:34 good dinner 00:15:49 with lemon, paprica, butter, and orange peel 00:16:15 HI les_w :P 00:16:16 whit lemon? 00:16:24 hi jacky 00:16:28 yes lemon 00:16:29 whith 00:16:37 with* 00:16:39 hehe 00:16:45 with lemon? 00:16:50 fresh lemon juice 00:17:06 and grated orange peel 00:17:15 good :D 00:17:35 :-) 00:18:12 Holiday is at an end tommorow 00:18:18 back to some good work 00:18:19 les_w: wait .. K4ts want show to you a plate we eat yesterday.. 00:18:26 ah ok 00:19:34 http://digilander.libero.it/jackydgl0/capodanno2006/img003.jpeg.html 00:20:08 looking 00:20:28 we got about 20 different plates during the dinner.. :) 00:21:00 Jacky^: which one is you? 00:21:01 mmm looks very good 00:21:09 slurp 00:21:14 haha 00:21:24 alex_joni_: what? the fish ? :D 00:21:44 no, I mean in which picture can we see you 00:22:05 there are 2-3 photos in the album 00:22:09 alex_joni: http://www.Jacky.orgy/ 00:22:18 with the y? 00:22:21 alex_joni: hes the one in the red dress 00:22:34 Jymmm: haha 00:23:11 http://digilander.libero.it/jackydgl0/capodanno2006/img011.jpeg.html jacky and k4ts? 00:23:38 http://digilander.libero.it/jackydgl0/capodanno2006/img002.jpeg.html 00:23:43 here we are 00:24:10 les_w: if I know vel, acc and cycle time, how many cycles does it take to accelerate to vel? 00:24:11 ok 00:24:35 jacky and anna , you both look good! 00:24:37 Jacky^: you are the small one on the right, doing the kung-fu impression? 00:24:43 cradek: let me think a sec 00:24:48 heh :D 00:24:55 vel+=d(vel) (or accel) right? 00:24:58 your son? 00:25:01 les_w: pretty good 00:25:08 d(vel)/d(t) , or cycle_time 00:25:55 yeah just start taking multiple difference equations as an approximation of differentials 00:25:56 so you need to calc. d(vel) or accel in a cycle, and see how many times that is needed to get added before cur_vel makes it to desired_vel 00:26:35 alex_joni_: I think it should be a simple division 00:26:42 num_cycles = (desired_vel - current_vel) / accell 00:26:55 not sure how simple it is :D 00:27:06 you need cycle time in there somewhere 00:27:11 Well, there are typically multiple regimes in a trajectory cycle 00:27:13 accelll holds it 00:27:21 vel / accel / cycle time 00:27:37 that makes the units work out so it must be right? 00:27:38 servo or trajectory? 00:27:53 traj in TP 00:27:58 ok 00:28:07 servo is only afterwards on PID & such 00:28:07 I meant "starting at 0" but forgot to say it 00:28:10 hes anna's son, Emanuele 00:28:10 damn wifi .. lag :( 00:28:12 lost packets 00:29:20 well, there are many schemes. In general a new set of polynomial coefficients is generated at each waypoint. A set for each axis. 00:29:43 IRON MONKEY... [Chinese goes here] [5 second delay goes here] [english translation goes here] 00:29:51 les_w:think simple, don't complicate it too much 00:29:58 simple simple 00:30:04 ha hard for me I guess. 00:30:07 Finally, I done to fight with kde on my laptop ! 00:30:18 cradek: I feel funny about the / cycle_time 00:30:30 alex_joni_: me too, that's why I'm asking 00:30:39 got Ion3, save alot of space in a 800x600 monitor : http://digilander.libero.it/jackydgl0/schermata1.png 00:30:41 let's get an example 00:30:48 :P 00:30:50 desired speed = 10 00:30:55 accel = 100 00:31:03 cycle_time = 0.02 00:31:49 accel = units / s^2 00:31:51 right? 00:32:03 right :-P 00:32:05 so we need to convert that to units / cycle_time ^2 00:32:18 I just think simply this: a time parameterized path of ANY form can be represented by a set of polynomial coefficients. 00:32:33 so we know how much we can accel inside a single cycle_time 00:33:02 some sets will blend with prior or later ones with continuity in time , velocity, accel, jerk, etc. 00:33:23 les_w: think exact-stop trapezoidal 00:33:29 oh 00:33:30 ok 00:33:59 well, trapezoidal in velocity is quadratic in position 00:34:03 les_w: you're thinking of the answer to a question different from the one I asked 00:34:12 often called parabolic blending 00:34:19 no blending 00:34:22 if you are at zero speed, and you're starting motion 00:34:34 you have desired velocity, and max_accel 00:34:37 I want a trapezoidal profile of velocity 00:34:43 how much time will it take to reach that? 00:34:56 right 00:35:01 how many cycles 00:35:12 time, we convert to cycles afterwards :D 00:35:16 Well, blending in the sense that a parabola describes position vs time as it comes to a stop 00:35:17 maybe round up 00:35:28 LES stop blending 00:35:55 ok, in time, it's surely vel/acc 00:35:59 ha ok. I see exact stop as just a special case. 00:36:01 right 00:36:02 the answer is vel/acc seconds 00:36:16 now you need to convert that to cycles 00:36:20 right 00:36:23 round up 00:36:47 round(vel/acc/cycle_time,0) 00:36:55 round(vel/acc/cycle_time+0.5,0) 00:37:21 so it's ceil( (vel/acc) / cycle_time) 00:37:34 I'd say so 00:37:37 I simply think of time....derived from cannonical distance and feed 00:37:44 always changing 00:37:51 K4ts_ has joined #emc 00:38:35 The equations on my page cover exact stop or blended 00:38:44 http://lumbrgrip.com/ 00:38:52 les_w: where? 00:39:16 just a sec 00:39:16 Jymmm: was that a joke? 00:39:31 alex_joni no, it's a real product. 00:40:12 gravity hook for any lumber 00:40:52 great:( 00:41:00 lol 00:41:03 http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?Trapezoidal_Velocity_Profile_Trajectory_Planner 00:41:11 link to my page there 00:41:18 thanks to alex 00:42:24 oh fuck me... product $10, shipping $16. 00:42:31 me? what did I do :D 00:42:59 haha 00:43:16 les_w ? 00:43:34 yessssss? 00:43:48 les_w was wondering what cha was laffin at? 00:43:55 ih ih 00:44:14 anna laughibg too 00:44:21 bleh 00:44:29 sticky keys. 00:44:39 les_w Ah, I thought it was the hooks 00:45:05 Jymmm: you got good affair! 00:45:21 K4ts until I get caught =) 00:45:41 Still using proto keyboard. I have a crate of them. the design is flawed. Lasts about 2 weeks before the keys stick. 00:45:46 K4ts I had an uneventful New Years eve, it was nice. 00:46:08 Changing the plastic composition... 00:46:10 les_w but you said you were on the last one two months ago 00:46:27 Jymmm: he's on the refurbished ones now 00:46:28 :D 00:46:34 alex_joni ah, heh 00:46:36 naw still have a couple 00:46:39 sent more 00:46:46 couple hundred or couple thousand 00:46:47 no improvement though 00:47:33 vI think the ABS is too highly filled withg talc 00:47:52 keys quickly jam 00:48:16 too much gap in the key to keyboard linear bearing 00:48:30 but tooling is already made 00:48:42 like $100k 00:49:04 so trying to fix it with special snake snot added to th abs 00:49:16 rather than new molds 00:49:58 just trying to solve this issue for a friend 00:50:13 I don't make keyboards, but he does. 00:50:56 I make suggestions, he sends crates of keyboards. 00:51:55 my keyboard is 20 years old and like new 00:51:57 les_w: my dog dead the 24 december sigh 00:52:09 sorry anna 00:52:15 get another one 00:52:21 is $75 for servo amp brand new good? 00:52:35 Jymmm: what kind of amp? 00:52:53 alex_joni SER-VO =) 00:52:55 cradek, We used to make very high quality saturable reactor keyboards 00:53:01 but sold the division 00:53:11 "saturable reactor"? 00:53:23 motion science 00:53:35 what kind? DC or AC? what kind of power, how do you control it (analog voltage or bus?), does it use a tach or encoder? 00:54:01 brush type, 21 to 180 VDC, 24A peak, 12A cont 00:54:07 40KHx switching 00:54:08 yes Fenn. Magnet on key saturates a simple single turn transformer 00:54:10 KHz 00:54:21 what brand jymmm? 00:54:22 les_w: wild 00:54:34 Cleveland Motion Controls 00:54:43 Anna will you get another dog? 00:54:43 ahh, Burny 00:54:50 les_w: I don't know how these IBM keyboards work - there is a non-conductive "paddle" that "touches" two pcb pads 00:54:59 must be some kind of capacitive effect 00:55:07 i thought it was conductive 00:55:10 the cheap ones are membrane 00:55:12 no mechanical contacts to wear out 00:55:24 les_w: no, I'm talking about 20 year old buckling spring technology 00:55:34 Jymmm: CMC is also Burny, maker of servos for many plasma/oxy-fuel rigs. quite well thought of, but rather old tech usually 00:55:40 I made an awful lot of membrane switches in the past 00:55:41 no membranes, no rubber parts, no contacts to oxidize 00:55:48 hate em...but made em. 00:56:05 that stuff is shit 00:56:28 K4ts has quit 00:56:34 Jacky^ has quit 00:56:44 yeah we just used a stamped matrix of simple 1 turn transformers 00:57:05 les MDC50 --> http://motionscienceinc.com/downloads/servo_amplifiers/ds_mdc_mdcx.pdf 00:57:11 a cheap magnet could easily saturate them with no contact 00:57:34 K4ts_ has quit 00:57:35 robin_sz I have nfc. 00:57:36 did it consume a lot of power? 00:57:37 crummy transformers are easy to make. In this case it was a feature! 00:57:47 no very little 00:58:09 but they have the PS and stuff built in to them. 00:58:30 just from the connector. 00:58:35 that was enough. 00:58:45 but these were expensive 00:58:45 ok to make this harder now, we know how long it takes to do the accel, but how far does it travel? 00:59:06 cost on most pc keyboards used now is a buck or two. 00:59:06 Saw some (maybe) 100mm NSK rail and slide about 24" long for $50 (too pricey) 00:59:12 10mm 00:59:16 seems like you have to integrate or something 00:59:32 yes cradek 00:59:40 two integrations 00:59:44 Jymmm: yes I know ;) 00:59:54 les_w no comment? 01:00:16 good deal! 01:00:27 fenn: still around? 01:00:29 yea 01:00:37 I'm having problems understanding this translation from the continuous functions in the math world to these discrete time slice cycles 01:00:37 100mm rail? 01:00:40 wow 01:00:41 Jymmm: those amps put out a lot of power 01:00:47 les_w are these as useful as geckos? 01:01:00 need distance travelled with time for constant accel? 01:01:32 yes, that's one thing I want to know 01:01:34 I dont think I have ever even seen 100mm rail ... thats HUGE 01:01:35 fenn I know nothign of servo, except eventually will go to them, so... 01:02:09 Jacky^ has joined #emc 01:02:19 but that could be years from now 01:02:50 1/2 At^2 basically 01:02:50 uhm.. windows sharing system fault :/ 01:03:16 ajoni has joined #emc 01:03:36 wb alex 01:03:40 so a weight dropping at one g (9.81 m/s) goes 9.81/2 distance in one second 01:03:50 are servo amps "universal", or is there something I should be aware of? 01:03:53 alex_joni_ has quit 01:03:58 ajoni is now known as alex_joni_ 01:04:01 K4ts has joined #emc 01:04:06 oops 9.81 m/s^2 01:04:07 so you're integrating a*t dt 01:04:11 Jymmm: not at all universal 01:04:25 because our acceleration gives v = at 01:04:26 fenn: I wanted to ask you about a feature request you added to SF 01:04:33 well consider inegrating a constant 01:04:37 uhm.. lemme look 01:04:38 alex_joni I foudn these for $75 http://motionscienceinc.com/downloads/servo_amplifiers/ds_mdc_mdcx.pdf 01:04:45 alex_joni MDC50 01:04:51 * alex_joni_ looks 01:04:57 integrate a constant twice 01:06:49 so again I have vel and acc 01:06:52 hence 1/2 at^2 01:07:02 Jymmm: they seem pretty nice 01:07:09 but you need motors with tach feedback 01:07:10 I want to accel from 0 at acc in/sec2 up to vel in/sec 01:07:19 how many inches do I traverse? 01:07:27 alex_joni dont all servo motors need that? 01:07:31 nope 01:07:39 Jymmm: some take the feedback from the encoder 01:07:43 or maybe resolver 01:07:49 some only use encoders 01:07:55 solve the system 01:08:01 but tach for speed feedback is not bad (it's the older way..) 01:08:12 oh, ok. 01:08:22 at constant accel, velocity is accel*time 01:08:32 ive seen that drive on some of the larger Burny controllers, its OK, but nothing amazing ... 01:08:34 Jymmm: just need to make sure you get motors that are compatible 01:08:40 they sell matching motors too 01:08:40 and dont most today go brushLESS if they can? 01:08:42 and position is 1/2 accel* time ^2 01:08:57 Jymmm: whole different controller for brushless afaik 01:09:08 yep 01:09:14 brushless is AC 01:09:15 so, that is an initial final value problem. 01:09:19 thats a DC controller 01:09:21 alex_joni: Right, but most today go brushLESS, correct? 01:09:30 where initial and final value are both 0. 01:09:32 robin_sz ah. 01:09:36 Jymmm: different pricerange too 01:09:52 the Baldor range are good for small brushless 01:10:31 would these be good to hide in the closet for a year or so? 01:10:39 cradek shouldn't you be doing it the other way around? start with inches and find out how many seconds you need for a given accel, what max velocity is etc 01:10:40 robin_sz: there is brushless DC too, pricier than AC 01:10:40 the motors at least 01:10:45 or just wait and see what's available in a year? 01:11:04 alex_joni: brushless DC? how can that work? 01:11:15 robin_sz: it's not really dc, it's a trapezoidal waveform motor 01:11:23 robin_sz: it makes the control electronics simpler 01:11:23 ahh. 01:11:35 fenn: commutation within the motor? 01:11:41 exact stop is just a special case of the equations on my page where both initial and final velocities are zero 01:11:51 robin_sz: no, it's an alternating current trapezoidal waveform 01:12:04 plug'nchug 01:12:05 alex_joni_ has quit 01:12:07 oh, so just AC by another name 01:12:08 alex_joni_ has joined #emc 01:12:20 alex_joni: here, have this. 01:12:28 ok, after dinner I'll read your web page a few times until I understand everything 01:12:49 robin_sz: what? 01:12:57 alex_joni_: its what I call a "working internet connection" 01:13:00 ok chris. It' is pretty much right out of a very old college text of mine. 01:13:18 robin_sz: if it works in the mountains I'll get it 01:13:23 ...dated 1842 01:13:28 close. 01:13:32 1742? 01:13:38 haha 01:13:40 1492 01:13:41 1842 AD ? 01:13:42 les_w: I hope I'm just rusty, not dumb :-/ 01:14:35 robin_sz: not just AC by another name, it's not 3-phase, it's 1-phase 01:14:42 I am not a typical engineer chris. I just do math. Everything, even hammering a nail, I think of in math terms 01:14:42 so .. different 01:14:43 :D 01:14:44 if you want small servos, check out the small Baldor stuff, the drives are cheap enough and they work very nicely with their motors .. we have several in production now 01:15:06 alex_joni: 1 pahse, OK, so how do you make it go backwards? 01:15:43 you output it dephased backwards 01:15:54 robin_sz : No, I was just wondering if I should buy these now and store them, or just wait and see whats available later on. 01:16:14 robin_sz no motors are in the equation atm. 01:16:18 Jymmm: better wait 01:16:26 well, id guess "probably not" then 01:16:28 and get drives that suit the motors you choose/find 01:16:35 * robin_sz nods 01:16:42 or just use geckos 01:16:43 alex_joni yeah, just didnt know is all. 01:16:48 night hall 01:16:56 robin_sz $75/ea not too shabby. 01:17:05 good price ... 01:17:10 all 01:17:11 but not the right thing 01:17:11 robin_sz: http://www.maxonmotorusa.com/files/catalog/2003/pdf/03_Technik_kurzund_buendig_ec_26_27_e.pdf 01:17:15 good night anna 01:17:15 sorry 01:17:16 night K4ts 01:17:27 G'Night K4ts 01:17:55 Hmmmm but and sell on ebay maybe? 01:18:10 K4ts has quit 01:18:25 Jymmm: check ebay first for prices 01:18:33 doing that now =) 01:18:41 You know, I don't think of eact stop much in cnc motion. It is not practical in cutting, brcause chip load goes to zero. Burned work and dull tools rapidly. 01:19:26 alex_joni: look sliek a 3 phase motor to me ... 01:19:45 yup, but a bit different I guess :) 01:20:12 mmm ... has a self-commutating option with "block" current drive 01:21:01 i thought the difference was in the magnetic flux coupling (or whatever that's really called) 01:21:17 the gap between the rotor and stator laminations 01:21:36 shrug 01:21:37 probably 01:21:39 anyway ... 01:21:44 it's shaped so that when you spin the motor like a generator, you get a trapezoidal waveform 01:21:46 I think I have found a compressor :) 01:21:55 http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=7208012779 01:22:24 oh, I did that already 01:22:25 sorry 01:22:59 oh for nitrogen generation? 01:24:29 yep 01:24:34 For that to work, the market for ntrogen/argon as a byproduct of oxygen manufacture has to be saturated? 01:24:39 for the compression into the storage cylinders 01:25:17 shrug ... i just gte fed up of buying so much nitrogen ... 01:25:49 its not *that* expensive ... but its easy to use a 10 cylinder pack in a day 01:26:07 then just having run out is inconvenient 01:26:13 it's almost like buying air huh 01:26:32 onsite generation should cost 10% of buy in price 01:26:43 really 01:26:45 alex_joni: what did you want to ask about my feature request? i know it's a little wrong in some places.. 01:27:05 then someone is making obscene profits somewhere 01:27:10 les_w: well that swhat the notrogen PSA system sellers claim 01:27:22 hmm 01:27:26 its �100 a month just to rent a 10 pack 01:27:33 plus 75 a refil 01:27:42 plsu 20 delivery 01:28:00 I use so little I have not thought about it much 01:28:14 this will pay off in 2 months I hope ... 01:28:27 I think I can build it for �1000 or so 01:28:31 with ebay bits :) 01:28:35 We threw out all the LN2 each shift end at harris 01:28:41 heh 01:28:44 just poured it on the floor 01:28:52 100 liters or so 01:28:54 well, yeah, some places have too much 01:28:58 other not enough 01:29:12 that was from liquefaction? and distilation? 01:29:26 thompson apparatus? 01:29:39 werent you worried about suffocating? :) 01:29:42 it was from the thin film sorption pumps 01:30:06 alex_joni_ has quit 01:30:12 coo 01:30:24 Jymmm: looked at it more carefully, those are AC drives 01:30:29 jmkasunich has joined #emc 01:30:29 Once a worker poured it down my pants...so I poured it down her shirt 01:30:29 fenn: no, he was worried about freezing his toes off 01:30:31