#emc | Logs for 2009-04-28

[02:16:52] <Valen> howdy do yall
[02:16:55] <Valen> or is that all yall
[02:16:59] <Valen> i can never tell
[02:19:18] <eric_unterhause1> you'uns
[02:20:20] <Valen> gotcha
[02:20:39] <Valen> I was wondering, what would be involved in getting EMC to run steppers like servos?
[02:21:02] <Valen> basically I am thinking to wrap the existing stepper module in a "virtual servo"
[02:21:30] <Valen> so you can drive it like a DC motor and get position information from an encoder rather than leaving it open loop
[03:52:40] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[04:07:53] <Valen> * Valen zaps the room Frankenstine style MUWHAHAH!
[04:36:29] <tomp> tomp is now known as tomp3
[04:58:33] <eric_unterhause1> cutting with the lathe is fun, too bad I have almost no idea what I'm doing
[04:58:57] <Valen> removing all the bits that dont look like the end product
[04:59:20] <eric_unterhause1> that's my goal anyway
[04:59:46] <Valen> getting some 1um linear scales to use on our mill today
[05:00:00] <Valen> as the encoder
[05:00:12] <Valen> should have enough resolution that the servo loop will run nicley
[05:00:14] <eric_unterhause1> I have some scales for my lathe
[05:00:27] <Valen> and got a price of $360 for a set of ballscrews
[05:00:30] <Valen> thats Au
[05:00:52] <Valen> that we are going to manually anti-backlash with some valve springs or belvile washers or something
[05:00:59] <eric_unterhause1> that doesn't sound too bad
[05:02:57] <Valen> so hopefully we can have a "0ish" backlash system
[05:48:14] <MrSunshine> gaah feels like using PWM to limit he current wont work to well, say i have a motor with a very high inductance, a small PWM pulse wont do much on that, on the other hand a motor with very small inductance a small pwm pulse will do alot right?
[05:48:38] <Valen> yah
[05:48:42] <Valen> fry the controller
[05:48:54] <MrSunshine> ?
[05:50:30] <Valen> well not nesicarily fry but it will pull lots more current
[05:50:49] <Valen> basically the PWM frequency you run needs to be somewhat matched to the inductance of the motor
[05:51:00] <Valen> ie low inductance high PWM freq and vice versa
[05:51:10] <Valen> the inductance smooths out the current flowing
[05:51:11] <MrSunshine> ye, thats what i figured, so a fixed pwm current wont work :/
[05:51:20] <Valen> it will for a given motor
[05:51:37] <Valen> most motors are pretty much the same
[05:52:01] <Valen> well enough that 3khz+ PWM will give a decent smoothing
[05:52:14] <MrSunshine> first thing i was thinking was to use timers, one timer that checks ADC until a threashold value, disables the enable pin, fires another timer, that continiously checks until the current is low enough and turns it on again .. or if its all manageable in one timer
[05:52:45] <Valen> what are you really trying to do?
[05:52:54] <MrSunshine> running a stepper with an AVR
[05:52:56] <Valen> the goal of the whole project
[05:53:06] <MrSunshine> Valen, learning how it all works and make a microstepping controller =)
[05:53:13] <Valen> just PWM it at 20Khz or so it'll be fine
[05:53:27] <Valen> you can use the hardware PWM modules then (i presume the AVR's have them)
[05:53:56] <MrSunshine> ye, 5 or 6 output pins
[05:54:10] <Valen> easy peasy then ;->
[05:54:24] <Valen> the higher the frequency the better lol
[05:54:35] <MrSunshine> but then i have to take into account reference voltage ... for max current the motor can run at, and the current limiting per step :)
[05:54:48] <Valen> until you melt the silicon, but then most steppers are pretty low power anyway
[05:54:51] <MrSunshine> Valen, noticed the motor screams at to low frequencies :)
[05:54:59] <Valen> magnetostriction
[05:55:04] <Valen> cool word no ;->
[05:55:34] <Valen> what are you using for current sense?
[05:56:15] <MrSunshine> its a current sense resistor and i put that throught the ADC of the AVR
[05:56:35] <Valen> if its for current limiting thats typically regarded as too slow
[05:56:58] <MrSunshine> Valen, ye i figured that also but the AVR people insist it shouldnt be a problem
[05:57:17] <Valen> what you would usually do there is have that output go to a comparitor (possibly via an amp) and then into an and gate in line with your AVR
[05:57:22] <MrSunshine> so ive been looking at doing hw chopping with opamps/comparators
[05:57:37] <Valen> its also handy that it'll keep it protected if the AVR faults
[05:57:58] <MrSunshine> ye loads of problems to solve =)
[05:58:21] <Valen> Personally driving steppers i wouldn't worry about cycle by cycle current limiting
[05:58:35] <Valen> if its for a personal project i wouldn't bother with current limit at all
[05:58:50] <MrSunshine> Valen, wel for microstepping i have no choice :)
[05:58:59] <Valen> just map the current Vs voltage for your motors
[05:59:21] <MrSunshine> either its high torque full stepping or its high/low torque halfstepping :)
[05:59:27] <Valen> run it open loop
[05:59:56] <Valen> you can still micro step, you just measure the voltage vs current for your motors "offline" and put that in as a table into the AVR
[06:00:05] <MrSunshine> you mean never sense the voltage just adjust the curves of the PWM according to the max current of the steppers?
[06:00:27] <MrSunshine> so if i got motors that has max 1A i just hardcode the curve for that you mean? :)
[06:00:35] <Valen> yeah basically
[06:00:52] <Valen> at low speeds (when the micro stepping matters) it should work pretty well
[06:01:07] <Valen> otherwise you need to factor in the back EMF from the motor spinning
[06:01:23] <Valen> comes down to what you like more, hardware or software ;->
[06:01:29] <MrSunshine> alot of stuff to take into account :)
[06:01:32] <MrSunshine> software i know :)
[06:01:36] <MrSunshine> as ive programmed for many years :)
[06:01:41] <Valen> might be easiest to do that then
[06:01:55] <Valen> like i said, when your doing the microstepping its not going to matter to much
[06:02:14] <Valen> and factoring in back EMF is pretty easy
[06:02:32] <Valen> stick it in a drill and attach a scope
[06:02:39] <Valen> see what the volts per RPM you get is
[06:02:59] <Valen> you can then subtract that from the measured voltage to get the "true" voltage
[06:03:27] <Valen> its not as "good" as doing it all the right way but hey its pretty easy ;->
[06:03:43] <MrSunshine> does the allegro chip etc do that kind of thing ? ..factor in that ? :)
[06:03:57] <MrSunshine> chips
[06:04:05] <Valen> whats the allegro chip? the current sensor?
[06:04:15] <MrSunshine> nah the stepper chips
[06:04:19] <Valen> dunno
[06:04:22] <MrSunshine> that is a full translator and h-bridge and everything
[06:04:47] <Valen> if they do the microstepping, they probably do proper current sensing
[06:04:57] <MrSunshine> but i should be able to calculate the curves fromt he reference voltage also right ? :)
[06:05:07] <MrSunshine> ye, they have a 3 bit or so ADC
[06:05:14] <Valen> they almost certainly do that because they wont know the motor your running
[06:06:11] <MrSunshine> using software i could hook up a small memory and make a program to set up the chip to the motors running on it from the PC :)
[06:06:29] <MrSunshine> so it loads the settings each time it starts and runs with them :)
[06:07:15] <Valen> I wish that hostmot2 knew how to drive "brushless DC servo motors" directly
[06:07:32] <Valen> cos that is a stepper motor driver right there ;->
[06:09:50] <MrSunshine> so i could use the REF voltage to a comparator to cut the power to the coils if exessive current flows throught the coils for safety and do the rest in software right? :)
[06:10:27] <MrSunshine> hadcoded values in code for current and PWM duty cycles, just pushing that to the controller for each step and voila :)
[06:10:34] <MrSunshine> sounds ALOT simpler =)
[06:11:36] <MrSunshine> Valen, you know how a chopper works also ? .. does it turn down the voltage after it reaches its destination current or does it not ?
[06:11:50] <MrSunshine> is the same say 24V applied each time i pulse the coils ?
[06:33:00] <Valen> sorry was repairing a speed controller for somebody
[06:33:08] <Valen> not going to make post today dang
[06:34:42] <Valen> i dunno what exactly a "chopper" is
[06:34:56] <Valen> i think thats what is refered to as this whole thing lol
[06:35:35] <Valen> Another way you can do it if you want a more analog approach is a kind of "servo" system
[06:36:04] <Valen> basically if you are putting a current sense resistor in it at all and want current limiting
[06:36:41] <Valen> you can get the AVR (through some small filtering caps) to output an analog voltage
[06:37:19] <Valen> then you run that into one input of your comparator and the other input via an amp from the sense resistor
[06:38:23] <Valen> so you are effectively using the comparator to do the PWM for you
[06:38:39] <Valen> (the output from the comp goes into the FET driver)
[06:40:12] <Valen> MrSunshine: you need a D FlipFlop and a clock signal as the easy way to get some hysteristis (cant spell that damn) into the system so it doesn't float
[06:41:03] <Valen> Otherwise you put the output of the amp into a difference amp and run the fet in analogue mode
[06:42:05] <Valen> old school style that one ;-> with a modern twist using the micro to generate the analogue voltage rather than a counter of some kind
[08:39:05] <MrSunshine> Valen, thats what i was thinking for a hw only chopper .. chopper is the current limiting .. cutting the current at a limit and then turning it on again when its fallen far enough
[08:39:24] <MrSunshine> im out doing yard work so wont be very active :)
[08:39:33] <MrSunshine> moved into a house where theres TONS of stuff to do :)
[08:56:14] <archivist> pjm__, useful lump near you 260399234970
[09:10:05] <Valen> MrSunshine: I reckon its probably easiest to just do open loop control with a decent map of current Vs Voltage (for a software guy) with a RPM correction. Every step after that is basically leading you closer to an analogue controller,
[09:10:54] <Valen> At the power levels your talking about a straightup analogue amp based controller should be pretty simple to throw together just using some decent FET's and a heatsink
[09:16:55] <piasdom_> g'mornin all
[09:17:01] <Valen> zup
[09:23:16] <Valen> I should prolly blog this mill conversion i guess
[11:50:27] <micges> good morning
[12:36:42] <als> a thought on stepper performance! If you could give more than one max_acceleration parameter (high for the first part of a rapid move when the step motor makes its best torque than drop to a lower acceleration as the torque drops off)
[12:39:41] <micges> cvs server seems down
[12:42:54] <SWPadnos> there was a hardware problem last night. I'm not sure what the status is now (though I confirm that it does seem to be down)
[12:43:48] <als> my machine preforms better at a lower max_velocity ,as when it reaches the top of its velocity at higher ipm settings it can no longer keep up do to torque loss and in order use these setting my max_acceleration must be so low
[12:45:24] <SWPadnos> yeah. I think DeskCNC uses a torque curve for that reason
[12:45:54] <als> feature request?
[12:46:17] <SWPadnos> there's a more general problem in that accel limits can be dependent on position as well - consider a rotational joint on a robot
[12:46:40] <skunkworks> SWPadnos: home?
[12:46:50] <SWPadnos> you can file a feature request if you like
[12:46:56] <SWPadnos> skunkworks, yes, finally
[12:47:02] <skunkworks> :)
[12:47:10] <SWPadnos> got stuck in Miami for an extra day (horror! :) )
[12:47:25] <skunkworks> watch out for the swine flu... ;)
[12:47:33] <SWPadnos> Miami, not Mexico
[12:47:43] <SWPadnos> though I have no idea how many people from Mexico were on the cruise
[12:48:19] <SWPadnos> or Spain - that seems to be the other hotspot
[12:49:31] <SWPadnos> this is cool: http://maps.google.com/maps/ms?ie=UTF8&hl=en&t=p&msa=0&msid=106484775090296685271.0004681a37b713f6b5950&ll=32.639375,-110.390625&spn=15.738151,25.488281&z=5
[12:50:10] <skunkworks> uh - couple in kansas
[12:50:36] <SWPadnos> luckily, they're "perfectly willing to isolate themselves" ;)
[12:51:43] <SWPadnos> "Hey honey, we're supposed to stay home all week. can you think of anything to do?"
[12:52:12] <skunkworks> there goes another baby boom
[12:54:01] <SWPadnos> heh
[12:54:17] <SWPadnos> ba-da-bang, ba-da-boom :)
[12:54:58] <SWPadnos> man, you guys were yakking a lot while I was gone. there are more emails on teh EMC users list than CCED and Geckodrive combined
[12:59:17] <Valen> how possible/hard is it do you think to "wrap" the stepper drive such that it appears like a servo to EMC
[12:59:25] <Valen> i am thinking in software rather than hardware
[12:59:47] <SWPadnos> simple. use stepgen in velocity mode, and stick a PID on it
[13:00:22] <Valen> so it can be done in HAL without changes to EMC propper?
[13:00:23] <SWPadnos> that's what the Pico Systems USC does - the step generator hardware is just a rate generator, and you use PID just like on a servo
[13:00:51] <SWPadnos> yes. "EMC proper" only outputs positions, it's up to other things (PID or stepgen) to make the motors get there
[13:00:59] <Valen> I'm just looking around and it seems much easier to get steppers than servos in a decent power/torque range
[13:01:13] <SWPadnos> why not just use stepgen then?
[13:01:28] <Valen> I had my first read through of the manual last night lol, i'm going up from the electronics/ballscrew level now
[13:01:41] <SWPadnos> if you're asking about using steppers with encoder feedback, that's a totally different problem
[13:01:49] <Valen> When i had looked previously everybody seemed to say you couldn't use steppers with encoders
[13:01:57] <SWPadnos> see above ;)
[13:02:02] <Valen> thats what i am talking about ;->
[13:02:37] <SWPadnos> well, it might be possible, but you'd have to write a HAL module to do it
[13:02:40] <Valen> then what ezacterly is the PID getting its information from?
[13:02:50] <cradek> they may have said you don't want to -- different from saying you can't
[13:02:58] <SWPadnos> and you should also have specialized stepper drives
[13:03:01] <cradek> or, your question was slightly different :-)
[13:03:12] <SWPadnos> which don't exist AFAIK
[13:03:16] <Valen> it seemed answered pretty unequivacly ;->
[13:03:42] <SWPadnos> here's the difference between steppers and servos, you can decide for yourself:
[13:04:11] <SWPadnos> if a stepper is out of position, it has already stalled and the part is ruined
[13:04:26] <Valen> I wanted to use the stepper output and such from EMC so that the hardware was off the shelf, and i wanted to use encoder inputs from my shiny new 1um scales for position
[13:04:27] <SWPadnos> if a servo starts lagging, you can ask it for more torque to catch up
[13:04:33] <cradek> what is the actual goal you're working toward, that led you to putting encoders on your steppers? maybe we haven't talked about the right question yet.
[13:04:51] <SWPadnos> yes, everyone wants to do that, and it's not a solved problem
[13:04:55] <cradek> so you're not even talking about encoders, you're talking about scales?
[13:05:11] <Valen> as far as EMC is concerned theres no difference
[13:05:18] <Valen> provided the backlash is "minimal"
[13:05:23] <cradek> as far as control loop, there's a huge difference
[13:05:31] <Valen> how?
[13:05:33] <SWPadnos> no, that's not true - there is a huge difference
[13:05:55] <Valen> its a quadrature input that is theoretically based on position
[13:06:00] <SWPadnos> PID looks at the difference between command and feedback, and outputs a control value to make the two match
[13:06:29] <Valen> and how are the pulses from an encoder different to those from the scale?
[13:06:51] <cradek> one gives the motor position, one gives the axis position
[13:06:53] <SWPadnos> steppers have no "reserve torque", and can therefor never put out any "extra effort" like servos can
[13:07:06] <SWPadnos> hold on, there are at least two questions being answered here
[13:07:11] <cradek> at least
[13:07:20] <cradek> I'll go have breakfast - that'll help
[13:07:21] <SWPadnos> one is the difference between loosely coupled feedback (scales vs. encoders)
[13:07:23] <SWPadnos> heh
[13:07:23] <Valen> they have a torque limit the same as a servo, in both cases you cant excede it
[13:07:28] <SWPadnos> hold on
[13:07:37] <anonimasu> um, no
[13:07:39] <SWPadnos> ok, problem 1: loosely coupled feedback
[13:07:47] <Valen> yes that is a given
[13:07:56] <Valen> that problem we are planning on solving mechanically
[13:08:05] <SWPadnos> ok, so zero backlash
[13:08:14] <Valen> yes, otherwise the whole system is screwed
[13:08:16] <SWPadnos> not "almost zero", but zero
[13:08:20] <SWPadnos> ok, problem 1 solved ;)
[13:08:28] <SWPadnos> problem 2: reserve torque
[13:08:32] <Valen> it should be within a count or 2
[13:08:43] <Valen> given that there is no such thing as "zero"
[13:08:46] <SWPadnos> a servo has two ratings, continuous and peak torque
[13:09:00] <SWPadnos> you design for continuous torque, not peak - else your motors melt
[13:09:13] <Valen> yup, and steppers have the same properties if not the same ratings
[13:09:23] <SWPadnos> therefore a servo has extra torque available sometimes, for limited amounts of time
[13:09:27] <SWPadnos> no, they don't
[13:09:32] <anonimasu> no they are completely different
[13:09:39] <Valen> doesn't matter
[13:09:39] <anonimasu> steppers have a speed/torque limit
[13:09:40] <SWPadnos> steppers are rated at their peak (holding) torque
[13:09:51] <Valen> point is both have a maximum torque that you cant excede
[13:10:07] <SWPadnos> no, that is *NOT* the point
[13:10:20] <Valen> so you are saying you can excede the max torque on a servo?
[13:10:34] <SWPadnos> you are correct, in that you will demagnetize either if you exceed some limit, but that is irrelevant here
[13:10:54] <SWPadnos> you can exceed the *design* torque on a servo, yes
[13:11:01] <Valen> its just a different design limitation, you just have to spec the system to not excede the torque the stepper can output in both acceleration and velocity
[13:11:05] <SWPadnos> because you design to the continuous rating, not the peak
[13:11:12] <Valen> same as a servo, just with different numbers
[13:11:21] <SWPadnos> well, you're not letting me get to the important part
[13:11:38] <Valen> that a steppers peak torque is the same as its continious torque?
[13:11:51] <archivist> no its not
[13:11:52] <SWPadnos> no, that's important, but not the main point
[13:11:53] <Valen> or that when the servo lags you can stuff more amps in to get more torque out
[13:12:07] <Valen> which you cannot do if you are already at max torque
[13:12:18] <archivist> the stepper cogs in that situation
[13:12:24] <SWPadnos> ok, so you've caught everything so far, excellent
[13:12:27] <Valen> ergo both have limitations that you have to take into account
[13:12:33] <Valen> i knew all this stuff to start with
[13:12:52] <SWPadnos> and one limitation of a stepper is that it is already dead in the water once you can detect a position mismatch
[13:13:12] <Valen> not if you are using the stepper like you would a brushless DC motor
[13:13:22] <Valen> which is essentially what is happening
[13:13:33] <SWPadnos> sure, like I said - you need a specialized drive to do this, and I'm not sure any exist yet
[13:13:59] <SWPadnos> and you won't be able to use software to make that happen, the timing is too tight
[13:14:01] <Valen> if you can get EMC to put out stepper pulses, but to treat it like it would an encoder then it will work
[13:14:34] <Valen> you still have to run open loop on the stepper output so you dont "slip"
[13:14:51] <anonimasu> steppers dont slip unless you are cutting too hard
[13:15:08] <Valen> which is usually taken care of by the driver in a brushless system but provided you keep on the inside of the torque curve that's not a problem
[13:15:11] <anonimasu> and then you are out of position already, and applying more speed/torque to catch up makes them stall even worse
[13:15:28] <Valen> if you are using linear scales you can correct for a dropped step
[13:15:37] <Valen> or encoders for that matter
[13:15:51] <SWPadnos> sure, you can correct for it in software, but physically, your motor will not lose one step
[13:16:09] <SWPadnos> it will stall, and lose many steps, and not restart until the speed has been reduced
[13:16:10] <Valen> it'll loose some multiple of poles and magnets
[13:16:45] <Valen> or until the torque is reduced to the point it works but in any case you have exceded the design of the system and you have your self to blame
[13:16:47] <SWPadnos> for a positioning system, that's fine - it will eventually get where it's supposed to go
[13:17:13] <SWPadnos> but for a contouring system, the contour will already be screwed up, because presumably the other axes don't all stall
[13:17:16] <Valen> not really relavent, if you exceed the limits of the machine it wont work, thats pretty common
[13:17:29] <SWPadnos> yeah, nobody is arguing that
[13:17:36] <Valen> its not a "magic" thing that makes everything work
[13:17:52] <Valen> I'm hoping to get ~5um accuracy without paying $4000 a screw
[13:18:01] <SWPadnos> the thing is that servo/PID assumes that you can exceed the design specs for short periods
[13:18:05] <Valen> and be able to use off the shelf drives
[13:18:17] <Valen> you cant exceed the specs though
[13:18:18] <Valen> ever
[13:18:31] <SWPadnos> there are two specs, continuous and maximum ...
[13:18:35] <Valen> if you exceed the max torque on a servo it just wont push any harder
[13:18:45] <SWPadnos> and you can exceel the continuous spec for short periods, but never the maximum spec
[13:18:49] <Valen> and the same thing happens your curve is screwed
[13:19:00] <SWPadnos> yeah, or it'll melt and/or demagnetize
[13:19:09] <Valen> yes, so you just design your system not to exceed the specs same as any other
[13:19:25] <Valen> in a servo system you leave some margin over what the max torque on the motor is
[13:19:32] <Valen> you do the same in a stepper
[13:19:44] <Valen> in a stepper the max and cont torques are ~=
[13:19:51] <SWPadnos> sure, so you design to 50 or 80% of the rated torque
[13:19:55] <Valen> in a servo there is a bit more range
[13:20:00] <SWPadnos> yeah, like 5x
[13:20:34] <Valen> like i said, its still not really important, if you don't ask for more torque than the stepper can give it wont drop steps
[13:21:02] <SWPadnos> in that case, you don't need scales
[13:21:11] <SWPadnos> since you will never have an "out of position" event
[13:21:14] <Valen> you do if you have cheap crappy ballscrews
[13:21:29] <Valen> like say ones you can get for $300 from china
[13:21:39] <SWPadnos> scales don't magically change the number of steps per inch
[13:21:56] <Valen> no but they are accurate
[13:22:01] <Valen> they don't vary over their length
[13:22:09] <Valen> or with temperature (much)
[13:22:15] <SWPadnos> if you sufficiently limit the speeds and accels, then you can probably get by just by using the method I first mentioned
[13:22:27] <SWPadnos> which is to use a velocity mode stepgen and PID
[13:22:44] <anonimasu> I'm pretty sure your machine will grow 3µm because of heat..
[13:23:01] <Valen> you said that EMC wont accept scales for steppers
[13:23:11] <SWPadnos> there, you're making the assumption that the stepper will always move how you tell it, and the scale becomes your true feedback
[13:23:23] <Valen> thats the goal
[13:23:31] <SWPadnos> I said that using scales to correct steppers is not a solved problem
[13:23:44] <Valen> why not?
[13:24:09] <Valen> if you can drive the stepper the same as you would a servo it would work no?
[13:24:09] <SWPadnos> but, if you make it so the steppers will never ever ever need correction, by designing the machine to 50% of the stepper specs, for example, then you can do positioning based on scales
[13:24:26] <Valen> that was what i was looking at doing
[13:24:41] <Valen> It seems i can get steppers and their controllers on the second hand market
[13:24:48] <Valen> that doesn't seem true for servos
[13:25:07] <SWPadnos> "encoders on steppers" is usually brought up to describe a system where you correct for lost steps, not for a system where the number of steps per unit changes
[13:25:18] <SWPadnos> you can, but it depends on the sizes
[13:25:26] <Valen> no lost steps = crappy and not going to work
[13:25:44] <SWPadnos> DC servos are reasonably cheap, brushless and AC are quite a bit more
[13:25:58] <Valen> I see DC servos for $450 or so an axis US
[13:26:03] <Valen> for decent ones
[13:26:06] <anonimasu> hm.. that's pretty cheap
[13:26:08] <SWPadnos> how big?
[13:26:12] <Valen> 250W
[13:26:21] <anonimasu> you shop in the wrong place
[13:26:22] <Valen> http://www.magmotor.com/brushed/brushed.html
[13:26:27] <Valen> the bottom one
[13:26:34] <SWPadnos> do you have to buy new?
[13:26:39] <Valen> $330 gets a C40-300
[13:26:45] <SWPadnos> or have a reliable source?
[13:27:03] <Valen> no, but i am yet to find 2nd hand actualy available
[13:27:06] <SWPadnos> (ie, can something like surplus center or ebay work, or do you want to make more pmachines later)
[13:27:31] <Valen> ebay would be fine but there doesn't seem to be anything comparable to those motors on it
[13:27:59] <Valen> and the scrap yards i know of here don't seem to know about these "motars?"
[13:29:01] <Valen> great sp mpw tjere are ass;pads ;p;
[13:29:09] <skunkworks> http://kelinginc.net/DCBrushlessMotor.html
[13:29:12] <Valen> hmm off by one
[13:29:20] <Valen> "great so now there are assloads"
[13:29:26] <skunkworks> http://kelinginc.net/ServoMotor.html
[13:30:20] <Valen> noice, although they eem about half the spec of the mag motors though?
[13:30:52] <SWPadnos> the $149 one is 226 oz-in cont.
[13:30:55] <SWPadnos> vs. 250
[13:31:13] <SWPadnos> though it's only 3200 RPM vs. 4000
[13:31:15] <Valen> ahh i was comparing the peak, i hadn't got to the cont ratings yet
[13:31:27] <SWPadnos> never ever ever ever look at peak! ;)
[13:31:44] <Valen> its a servo system, peak gives an indication of acceleration
[13:32:02] <SWPadnos> true, but you always design to continuous specs, and the peak is a nice reserve
[13:32:24] <Valen> the other thing is those need 40A to hit peak which is 2x the mesa controller i was going to use can put out
[13:32:46] <Valen> yes but wouldn't you rather have 2x the "margin" ?
[13:33:07] <SWPadnos> probably
[13:33:11] <Valen> ;->
[13:33:16] <SWPadnos> but it also depends on the driver I'd be using
[13:33:29] <SWPadnos> you're right, the 40A spec is annoying
[13:33:31] <Valen> hmmm can the gecko's do 40A at ~90 volts?
[13:33:34] <skunkworks> no
[13:33:40] <Valen> didnt think so
[13:33:52] <Valen> the mags have a 23A peak torque winding
[13:33:52] <SWPadnos> 80V, 20A
[13:34:09] <Valen> same as the mesa, except the mesa is good for 170v or so
[13:34:37] <Valen> on that winding i run out of volts before i hit top speed, looking at ~2500 RPM on the mag
[13:34:56] <Valen> which is ~500 IPM traverse speed
[13:35:13] <skunkworks> what is your 2X peak going to be used for? accelleration?
[13:35:19] <Valen> thats about all
[13:35:26] <Valen> and not that much of it more than likley
[13:37:27] <Valen> ouch they know how to charge for the AC stuff ;->
[13:40:56] <Valen> yaknow if you wanted to get "funky" with using steppers as BLDC's without too much added hardware, you could use the scale signal as a "rotational position" and drive the steppers based on that
[13:41:18] <Valen> although it'd work better with a rotary encoder on the back of the shaft
[13:41:53] <Valen> so SWPadnos you reckons the stepper as "servoish" thing might work?
[13:42:19] <skunkworks> so what happens when you are contouring a part where you are hitting the 2x current continuously?
[13:42:27] <SWPadnos> I don't know how well it will work - a lot of that depends on how crippled you make the design (Relative to the stepper torque ratings)
[13:42:48] <Valen> steppers seem to have torque to spare really
[13:43:05] <SWPadnos> they seem to at zero speed, because that's where they're rated
[13:43:06] <Valen> their cont torque (in ~ the same package) seems to be ~= the peak torque on the servo
[13:43:11] <Valen> true
[13:43:26] <SWPadnos> yep, and that's because they lose torque at higher speeds, whereas servos don't
[13:43:26] <Valen> although the faster your going the less torque your typically going to need
[13:43:43] <skunkworks> thier torque drops off quite fast at higher rpm
[13:44:00] <SWPadnos> that's not true
[13:44:04] <SWPadnos> you still need to decelerate
[13:44:23] <Valen> a servo actually does the same, as the speed goes up your controller (typically) runs out of voltage, so you cant hit peak torque at near full speed
[13:44:38] <SWPadnos> servo torque is independent of voltage
[13:44:45] <Valen> no it isnt
[13:44:51] <SWPadnos> it most certainly is
[13:44:52] <Valen> it is dependant on amperage
[13:44:56] <SWPadnos> exactly
[13:45:15] <Valen> and when you spin the motor you need to counter its backEMF to get the same amperage
[13:45:40] <SWPadnos> so yes, at speeds where your drive can't overcome back EMF and still provide sufficient current, you have a problem
[13:46:46] <Valen> on the mag motor system i was putting togther you hit the controller limit at ~1200RPM as i recall
[13:47:00] <Valen> you run out of volts at that point
[13:47:42] <Valen> assuming i can get a 160 odd volt controller ;->
[13:47:47] <Valen> powersupply rather
[13:50:27] <Valen> anybody know a source for that or will i need to wind my own?
[13:50:37] <Valen> I'd really rather not wind a transformer lol
[13:50:44] <SWPadnos> phone
[13:51:00] <Valen> in that case i will have a shower ;->
[14:13:09] <Valen> mmm cleanliness
[14:14:47] <Valen> so on to another matter, where would i start for the "using scales as a DRO through the parallel port" project
[14:15:04] <Valen> I have seen the thing in the wiki, its just a little deep for a first step lol
[14:16:13] <archivist> going for max acceleration AND using the parport does not make too much sense
[14:16:23] <Valen> they are 2 seperate projects
[14:16:28] <Valen> the DRO is on a lathe
[14:16:49] <Valen> it might get converted to CnC at some stage but its a low priority
[14:17:11] <Valen> at this stage its a "feet wet" type thing
[14:24:57] <Valen> so any idea where to start? ;->
[14:25:10] <cradek> read the documentation
[14:25:12] <cradek> then hook it up
[14:25:17] <Valen> lol
[14:26:19] <cradek> there are lots of examples of hooking encoders to parallel ports in the sample configurations. people do this for simple servo machines and MPGs.
[14:26:52] <JymmmEMC> SWPadnos: Telegraph
[14:27:21] <Valen> I shall look more into the sample configurations
[14:28:06] <Valen> It was more the pyVCP side of things ;->
[14:28:11] <Valen> I shall muddle through ;->
[14:28:41] <cradek> you can use the regular DRO in any of the guis - you can defer or permanently avoid messing with pyvcp
[14:29:46] <Valen> the gui's and EMC are ok with not having any outputs?
[14:30:04] <skunkworks> Just don't bring it out of estop ;)
[14:30:15] <Valen> oooh shiny ;->
[14:30:29] <Valen> that sounds like the place to start then
[14:30:37] <cradek> yeah it will track the position fine as-is.
[14:30:47] <Valen> more "conventional"
[14:31:03] <cradek> if you use AXIS you'll have a nice etch-a-sketch
[14:31:32] <Valen> so emc wont whine about not having any outputs?
[14:31:58] <SWPadnos> emc still outputs whatever it wants. if those signals never get routed to the real world, who cares?
[14:32:16] <Valen> i mean it needs a dummy output configured, i cant just put the encoders into HAL?
[14:32:41] <JymmmEMC> http://www.opera.com/
[14:32:43] <Valen> fire up axis and let them play togther all without any intention of an output
[14:32:52] <JymmmEMC> that's funny
[14:33:50] <Valen> they should have used <blink> some
[14:36:06] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=79408
[14:38:15] <Valen> so SWPadnos, think it'll be happy without any outputs defined?
[14:38:31] <SWPadnos> yes, EMC doesn't care if anything is "listening"
[14:38:46] <SWPadnos> you just need to make sure that the inputs are satisfied, if you want to get out of e-stop
[14:38:47] <Valen> funky I shall put it on my ToDo list for tomorrow
[14:39:09] <Valen> should just be A+B+R/Z for the 2 encoders yes?
[14:39:52] <SWPadnos> yep, and Z is unnecessary unless you want to get more comples and test homing or something
[14:40:03] <SWPadnos> complex
[14:40:53] <Valen> I figure if its there, might as well wire it up
[14:40:54] <BJT-Work> yea it is cleaning up at 0.075" under size http://imagebin.ca/view/5N9foE.html
[14:41:24] <cradek> if you want to be able to set offsets and switch between them, or zero axes, you will have to be out of estop. you can give emc very large following error limits to keep it from giving following errors.
[14:42:27] <cradek> BJT-Work: oops, looks like you were not centered very well
[14:42:40] <Valen> if it has no outputs defined would it give a following error? it has nothing to follow?
[14:42:59] <cradek> that's why you would need to give large following error limits
[14:43:07] <BJT-Work> the top was worn down about 0.060"
[14:43:24] <BJT-Work> the rest was still round
[14:43:31] <cradek> ah
[14:43:35] <cradek> are you going to have to make a sleeve?
[14:43:57] <Valen> BJT-Work what is it?
[14:43:59] <BJT-Work> I'm just using a different size bearing
[14:44:20] <BJT-Work> center of front axle on my tractor
[14:45:00] <BJT-Work> I can't get the OEM bearings so I'm using mcmaster carr bearings for $6 each
[14:45:55] <Valen> heh OEM would probably be $50 if you could get them
[14:46:31] <Valen> http://www.sparkfun.com/commerce/tutorial_info.php?tutorials_id=47&page=1
[14:46:45] <BJT-Work> that is for real... I needed a screw I thought till they told me the price $30 for a 2" 1/4-20 screw
[14:46:48] <Valen> oh sorry wrong window
[14:58:55] <A4ndy_> what is the best GUI/IDE compiler on Linux? ...like Delphi on Win
[14:59:06] <A4ndy_> RAD?
[14:59:34] <Valen> ummm you need a language to compile
[14:59:43] <cradek> I don't think anyone here uses one - nobody ever has an answer for this question when it comes up.
[14:59:57] <Valen> i use SPE for python
[15:00:01] <A4ndy_> that contain alot of component to create like simple button, label, prety boxs..etc
[15:00:03] <Valen> the autocomplete is handy
[15:00:20] <cradek> you mean a gui builder?
[15:01:03] <A4ndy_> i need to create very-prety boxes that exist on today linux distribution...
[15:01:11] <A4ndy_> gyu builder yup
[15:01:21] <A4ndy_> GUI yes
[15:01:27] <Valen> wxpython has a guibuilder integrated into SPE
[15:01:33] <Valen> wxglade i think
[15:02:46] <Valen> not that awesome though really
[15:02:51] <Valen> i do it all by hand
[15:03:11] <A4ndy_> are there any component also... like we could make hallo-world in boxes, button, cycleing labelin a minute
[15:03:12] <A4ndy_> ?
[15:03:57] <A4ndy_> all awesome, but i need a way to program
[15:04:40] <cradek> maybe there is a different channel where people would have better answers
[15:05:17] <A4ndy_> it is free to talk all the better right? we do democratic
[15:05:19] <SWPadnos> hmmm. I wonder if they'll deliver my luggage to the restaurant at the address they have, or if they'll actually look at the luggage tags and bring it here
[15:05:55] <A4ndy_> i look for gambas... but they never update since 2007
[15:06:27] <SWPadnos> A4ndy_, Qt Designer is quite nice
[15:06:29] <cradek> A4ndy_: I didn't mean you weren't welcome to talk here, I meant only what I said: that there may be better answers elsewhere
[15:06:39] <SWPadnos> though I haven't used it in quite a long time
[15:07:08] <the_goat> the_goat is now known as Herr_Gemutlich
[15:07:24] <SWPadnos> glade has nice looking widgets also, and seems easy to use (though I don't know if they ever implemented real undo)
[15:07:37] <A4ndy_> i also look on dev-c or code::block ... but i wonder how emc2 gui build off
[15:07:41] <SWPadnos> unfortunately, Kylix is no more (it was Borland's delphi product for Linux)
[15:07:57] <SWPadnos> the GUIs are not built with a designer like that
[15:08:01] <SWPadnos> (in EMC)
[15:08:13] <SWPadnos> except for stepconf, which was done at least in part with glade
[15:08:14] <A4ndy_> how and why not?
[15:08:34] <SWPadnos> (1) EMC has been around for longer than RAD GUI builders
[15:08:42] <SWPadnos> (2) nobody has bothered
[15:08:59] <Herr_Gemutlich> Herr_Gemutlich is now known as the_goat
[15:09:09] <A4ndy_> is it incompatible isue on getting failing RTAI?
[15:09:20] <SWPadnos> GUI and RTAI have nothing to do with each other
[15:10:00] <A4ndy_> so they (GUIprogramming/builder + RTAI could walk together..)
[15:10:11] <SWPadnos> someone said that they had successfully imported the EMC source tree into code::blocks, but I didn't have success in the 5 minutes I devoted to my attempt
[15:10:29] <SWPadnos> RTAI works whether there's a GUI or not
[15:10:52] <SWPadnos> if you want to make a GUI builder for EMC2, there's a lot more to be done than picking a RAD tool and running with it
[15:11:12] <SWPadnos> RAD tools work because someone did a lot of work making components you can easily use
[15:11:25] <A4ndy_> yes. EMC is a longtime roads... but you guys build it better and better, grew
[15:11:39] <SWPadnos> that work is not easy, especially when the components have to talk to something other than other components (like the real world)
[15:12:16] <SWPadnos> if you do find something like Borland C+ Builder for Linux, let me know - I'd love to try it
[15:13:01] <A4ndy_> .... Dev C++ ?
[15:13:14] <SWPadnos> what?
[15:13:19] <A4ndy_> nop?
[15:13:26] <SWPadnos> I would be interested for non-EMC reasons
[15:13:42] <SWPadnos> I use BC++ Builder on Windows, and it has been quite useful
[15:14:05] <SWPadnos> of course, it doesn't seem to exist any more, even on Windows
[15:14:49] <A4ndy_> windows.. i go TC, Pascal, DElphi, even TAsm...
[15:15:02] <A4ndy_> true
[15:15:29] <A4ndy_> but gcc still exist from my dos era...
[15:15:34] <SWPadnos> oh sure. I've been using Borland products since the days of Turbo Pascal on the Apple 2
[15:15:42] <A4ndy_> wow
[15:15:48] <A4ndy_> :)
[15:15:53] <SWPadnos> complete with Z80 accelerator card
[15:16:14] <A4ndy_> i do Zilog80 too u know
[15:16:42] <SWPadnos> heh
[15:16:44] <A4ndy_> from tiny winny memmorys good old days
[15:17:08] <SWPadnos> we used to have fun with the debug monitor in the Exidy Sorcerer
[15:17:28] <A4ndy_> cool
[15:17:45] <SWPadnos> and of course changing the command names in TRS-DOS on the TRS-80
[15:17:59] <crice> Over on the propeller forum, they are building a CP/M machine
[15:18:11] <A4ndy_> wow
[15:18:19] <Valen> I reckons it'd be a better use of somebodys time to actually split the RTAI/low level stuff more from the display stuff
[15:18:22] <SWPadnos> nuevo retro :)
[15:18:34] <Valen> so you can run EMC headless without X
[15:18:44] <SWPadnos> Valen, look at the code before saying that - there is no RT in the UI code
[15:18:47] <Valen> but still have all the nice interfaces
[15:18:49] <SWPadnos> and vice versa
[15:19:06] <A4ndy_> :)
[15:19:08] <SWPadnos> you can run headless/networked in 3 or 4 ways
[15:19:13] <Valen> yeah that i knows, but the interfaces between the RT section and the UI section is fairly closley coupled as far as i am aware?
[15:19:20] <A4ndy_> all amaze me
[15:19:20] <SWPadnos> nope
[15:19:35] <Valen> whats the interface?
[15:19:43] <SWPadnos> AXIS has some extra HAL coupling, but it can be disabled with something like AXIS_NO_HAL in the ini file
[15:20:18] <SWPadnos> you can use NML, a text interface over ssh, a graphical interface over remote X, or some VNC-type desktop sharing
[15:20:42] <Valen> X and remote desktop defeat the purpose somewhat
[15:20:52] <SWPadnos> somewhat
[15:20:54] <Valen> can you run the nice frontends over ssh?
[15:21:06] <SWPadnos> using remote X, yes
[15:21:13] <SWPadnos> ssh -X
[15:21:22] <Valen> lol still defeating the purpose
[15:21:46] <Valen> i mean to have an interface between the nice front end and the backend over the network
[15:21:46] <SWPadnos> do you know of another way to run a graphical X app over ssh that isn't X forwarding?
[15:21:52] <SWPadnos> yes, that's NML
[15:22:02] <SWPadnos> which doesn't run on ssh, it's a separate network protocol
[15:22:02] <Valen> havent heard that one
[15:22:13] <SWPadnos> that's how all the parts of EMC communicate with each other
[15:22:23] <Valen> ahh that sounds like the ticket then
[15:22:43] <SWPadnos> EMC was designed from the ground up to have separate IO, motion, and UI processes, separated by shared memory, serial, or networked communications
[15:23:05] <Valen> A+ on design in that case
[15:23:19] <SWPadnos> yep. the NIST boys put a lot of thought into it
[15:23:38] <Valen> I had just never seen it mentioned much around the place
[15:23:51] <A4ndy_> wow
[15:23:58] <SWPadnos> most people don't use it, so there isn't a lot of discussion
[15:24:41] <SWPadnos> in fact, you can have several UIs running, on one or more computers, simultaneously
[15:24:44] <Valen> Be a nice way of getting around the "propriatry driver latency" issues while still having the beefy OpenGL performance people like in front ends ;->
[15:24:55] <SWPadnos> yep, can be
[15:25:18] <SWPadnos> oh, I completely forgot about the program "emcrsh", which is a remote shell with EMC-specific commands
[15:25:28] <SWPadnos> like telnet for EMC
[15:25:30] <A4ndy_> i never seen something grew from begining like this since turbo-pascal7 legally extict
[15:25:42] <A4ndy_> so do dos
[15:26:04] <SWPadnos> well, time for me to do some pating work. see you
[15:26:14] <A4ndy_> :)
[15:26:20] <Valen> thanks for your help SWPadnos
[15:26:25] <A4ndy_> thanks SWPadnos
[15:26:26] <Valen> hope i can return the favour some day
[15:26:29] <Valen> ;->
[15:26:42] <Valen> time for dinner here, tis 1:30 in the AM
[15:26:58] <A4ndy_> linux never someone left behind
[15:27:47] <A4ndy_> eh, linux is allways there 4 us
[15:28:50] <A4ndy_> since WinXP would extict in time...
[15:29:19] <A4ndy_> to the Vista era ... like Dos to Win98
[15:29:30] <A4ndy_> all over again from 0
[15:29:55] <piasdom_> do ya'll know of a C++ irc channel ?
[15:30:03] <A4ndy_> longhorn stuff?
[15:31:01] <A4ndy_> nop sorry, i beginner here
[15:31:15] <A4ndy_> go search the web
[15:31:21] <A4ndy_> for sure
[15:32:18] <Valen> theres a #c on here
[15:32:23] <Valen> they mught be able to point
[15:33:12] <piasdom_> A4ndy: thanks, i looked on the web but nothing thought someone here might know
[15:33:46] <piasdom_> Valen: that should be about the same topic thanks
[16:03:42] <jepler_> jepler_ is now known as jepler
[16:04:25] <tomp> tomp is now known as tomp3
[16:06:37] <JymmmEMC> Alright! a use for my 32 serial port over ethernet controller... http://www.youtube.com/watch?v=N15X5OeT2oM&NR=1
[16:06:56] <tomp3> BJT-Work: i was just looking at a 'new' belt drive system.
[16:07:05] <tomp3> it used a belt that could fold onto itself ( its teeth and space was symmetric ).
[16:07:13] <tomp3> by using this belt, pretty much like you used in your machine,
[16:07:29] <tomp3> they claimed to get rid of 'all' stretch and had upto 4G accell.
[16:07:30] <tomp3> the link is on another box, will get it to you.
[16:08:19] <tomp3> the motor pinion engaged the belt and had 2 'idlers', one each side, but the 'free length of belt was just that, from ideler to ider
[16:08:43] <tomp3> the belt meshed into itself for the rest of the lentgth of travel
[16:10:52] <tomp3> JymmmEMC: rgb led ? run by a propeller chip
[16:14:37] <jepler> that's nice, but it should be wireless instead of usb or rs232. http://emergent.unpy.net/01233419381 (I need to improve the light diffusion on mine, though -- his is much better in that respect)
[16:16:03] <tomp3> looks like a mood ring
[16:16:08] <jepler> yeah basically
[16:16:12] <jepler> but it's for my computer's mood
[16:16:31] <jepler> but it's for me too, in that if it's red when I get up, my day will be shitty
[16:16:41] <tomp3> ;)
[16:19:51] <tomp3> BJT-Work: http://www.bell-everman.com/servobelt.html
[16:20:51] <tomp3> jepler i like the idea of the python control, its a very analogish intuitive input device ( the color wheel widget )
[16:21:16] <JymmmEMC> jepler: Hmmm, I never tried difusing a sphere before. I'm thinking an inner sphere with reflective surface.
[16:21:29] <crice> If you like that one, you should try this LED mood lighting. http://www.parallax.com/tabid/606/Default.aspx. Take a look at the you-tube
[16:23:22] <JymmmEMC> jepler: I think this is cool, Time to make a cnc hot wire cutter think =) Already have all the parts; slides, hotwire, etc, just need to assemble them.
[16:24:25] <JymmmEMC> http://www.sparkfun.com/commerce/tutorial_info.php?tutorials_id=47&page=1
[16:28:21] <LawrenceG> http://www.bell-everman.com/serbvideo.html fast belt drive!
[16:28:47] <LawrenceG> tomp3, I like it!
[16:30:18] <MrSunshine_> smart to do it like that :)
[16:30:43] <tomp3> like the old cowboy joke... wanna see it again ?
[16:32:35] <anonimasu> thoose servobelt drives are nice..
[16:37:04] <MrSunshine_> shoulndt be to hard to create oneself either? :)
[16:37:29] <MrSunshine_> only problem is that the motor has to follow the carrier thingie :)
[16:38:07] <anonimasu> I dont see a problem with that
[16:38:19] <anonimasu> only thing that would be a issue is belt stretch :)
[16:38:37] <anonimasu> and it seems that that design isnt bad about that
[16:38:51] <anonimasu> ^_^
[16:45:26] <cradek> people should not make videos of the first run: http://www.youtube.com/watch?v=seLQ5Bo_AIc
[16:49:33] <JymmmEMC> Eh, learn from someone else's mistakes
[16:51:53] <archivist> I also noticed the pallet take a good lean as it hits :)
[16:52:13] <archivist> machine needs a service afterwards too
[16:53:42] <cradek> JymmmEMC: true.
[17:01:03] <JymmmEMC> DOH! http://www.youtube.com/watch?v=uWOAAyN3btc&NR=1
[17:42:56] <Valen> lol dont you just hate the gasp of air in that probe one?
[17:44:30] <archivist> * archivist has no sound
[17:44:56] <Valen> think of the sound you would make after you have just stabbed yourself in the leg with a screwdriver
[17:45:01] <Valen> the one before you start screaming
[17:45:18] <Valen> the noise you make as you get ready to get on with some serious screaming
[17:46:26] <Valen> http://www.youtube.com/watch?v=yln_IGDuOCo&NR=1
[17:46:28] <Valen> EMC 3?
[17:47:17] <skinnypup_> lol
[17:47:35] <Valen> so looks like it grinds his head on the floor when it starts off
[17:49:52] <skinnypup_> sure does, looks like fun though
[17:56:48] <pcw> MrSunshine: You can do micro-stepping with a UC that has PWM and current feedback via a ADC
[17:56:50] <pcw> trick is use a PI loop to control PWM (and indirectly) current. To avoid having the PI loop track the
[17:56:51] <pcw> sine wave currents, use a park transform to de-rotate the current vector, then re-rotate
[17:56:53] <pcw> with inverse park to drive the PWM. (then the PI loop just works on "DC" values), The I term will compensate
[17:56:54] <pcw> for Back EMF as the motor spins faster.
[17:57:03] <jepler> hi pcw
[17:57:14] <pcw> Hi Jepler
[17:58:01] <jepler> pcw: Give me a second to formulate a question for you
[17:58:48] <jepler> pcw: I'd like to talk to a hostmot2 board with an embedded system that has a low voltage i/o (1.8v) and a spi bus
[17:59:21] <jepler> pcw: can the cpld on the parport board be adapted to this, and any advice on voltage translation?
[17:59:27] <MrSunshine_> cradek, ough :)
[18:01:48] <jepler> this is setting and reading hostmot registers with spi, not like the spi I/O boards
[18:02:15] <pcw> If you dont need to configure via SPI, the CPLD could stay the way it is,
[18:02:17] <pcw> just use some of the parallel port pins with level translators for the SPI interface
[18:02:51] <jepler> pcw: ah, change the hostmot2 to talk spi instead of epp?
[18:02:59] <pcw> Yes
[18:03:39] <geo01005> is this for ethernet or something?
[18:04:03] <pcw> Just the top level wrapper
[18:04:18] <pcw> (of hostmot2)
[18:04:20] <MrSunshine_> * MrSunshine_ thinks that people would do quite awsome stuff with AVRs if they just stuff them with loads more ram and flash :)
[18:04:45] <MrSunshine_> doh
[18:04:47] <MrSunshine_> wrong channel :P
[18:05:17] <jepler> geo01005: I'm trying to find out the feasibility of hooking a beagleboard to a mesa board. no ethernet involved.
[18:05:35] <jepler> pcw: and if I want to configure the fpga?
[18:06:25] <jepler> that'd involve changing the CPLD?
[18:06:44] <pcw> Configure via onboard EEPROM
[18:06:46] <pcw> If you want to change, re-write EEPROM
[18:06:48] <pcw> (Dont make a config that fails or you have to go back to USB or EPP)
[18:07:39] <geo01005> jepler: Cool. Emc on the beagle board?
[18:07:56] <geo01005> or something else...
[18:09:22] <jepler> geo01005: don't count those chickens yet, they're not even laid let alone hatched
[18:09:36] <jepler> geo01005: first someone has to port rtai to beagleboard
[18:09:42] <pcw> Thats how we re-write EEPROM now
[18:09:44] <pcw> (bootstrap FPGA config contains simple SPI master for flash EEPROM interface)
[18:09:46] <pcw> FPGA has 'reconfig thyself' pin so after EEPROM has been wriitten you can reload
[18:09:47] <pcw> FPGA with new config
[18:10:30] <jepler> pcw: any direction restrictions I should observe when choosing the parport pins for SPI?
[18:11:49] <geo01005> jepler: I didn't expect that you had EMC running on a beagleboard. I understand that is not a trivial thing to make happen.
[18:12:28] <pcw> The data bits go directly to the FPGA (through 100 Ohm resistors) I'd use those
[18:12:30] <pcw> (You would need to disconnect the USB interface if using SPI since the data pins are shared)
[18:13:40] <jepler> pcw: thanks, that pretty much answers the questions I have right now
[18:13:54] <motioncontrol> good evening.i want control with emc only one circular axis , not x,y and z axis.Is possible ?
[18:13:54] <pcw> (I mean just dont communicate over USB at the same time as you would have a conflict)
[18:16:19] <jepler> pcw: It's OK to do USB powered, and even USB enumeration, just don't transmit?
[18:20:27] <pcw> Yes, USB data bus (shared with EPP) will never be driven unless characters are sent to 7I43
[18:20:29] <pcw> (and FT245 shows FIFO data ready)
[18:21:31] <jepler> that means that if I had spi + usb connection I could send firmware over usb then switch to spi
[18:22:10] <jepler> thanks again for the answers -- if this project gets off the ground, I'm sure you can expect more
[18:22:45] <pcw> Yes but the hand-off would need some care
[18:22:47] <pcw> (Also once configured with your SPI interface, you can just tie the USB read strobe high)
[18:27:53] <pcw> Work can be avoided no longer,,,
[18:27:55] <pcw> Bye all
[18:32:02] <gene> humm, got an m3; g4p0.2 for spndle spin up that is not working. no error, but no effect???
[18:34:52] <cradek> gene: set S to something nonzero
[18:35:02] <cradek> S1000 M3
[18:36:49] <gene> it is, 500. driving 1/2 forstner bit, don't want chips all over the place, or overheat the bit and its above those 2 lines in the loop
[18:37:07] <cradek> so the spindle does not come on?
[18:37:17] <cradek> or it doesn't delay?
[18:37:34] <gene> comes on ok, but no delay in the following z- motion
[18:37:41] <cradek> can you pastebin your gcode
[18:37:56] <gene> so it hits the wood at about 1/2 speed.\
[18:38:08] <gene> probably
[18:43:28] <gene> wtf, I've sent it twice, to pastebin.ca, but konquerer is showingh me a link???
[18:43:40] <gene> wtf, I've sent it twice, to pastebin.ca, but konquerer is NOT showing me a link???
[18:44:08] <gene> if you an find it, it left here as cartridge-block.ngc
[18:45:34] <gene> humm, file->properties says it is only 1/2 second longer with 144 of then total.
[18:46:31] <cradek> hm, I don't see it at pastebin.
[18:46:49] <gene> pastebin.ca?
[18:47:00] <cradek> just paste the line before, including m3, and after here?
[18:47:08] <cradek> yeah, looking at pastebin.ca
[18:47:09] <gene> or is there a better place?
[18:47:39] <jepler> pastebin.com is one alternative
[18:47:39] <cradek> not sure - I think there are a lot of pastebins
[18:47:59] <gene> o100 sub #1 #2 #3 #4 #5 #6 #7 #8 #9
[18:48:03] <gene> s#9 (spindle speed)
[18:48:07] <gene> o101 while [#6 le #7] (#6=#32 in call, #7=#37 in call)
[18:48:11] <gene> g1f#4z.1 (raise for move clearance)
[18:48:15] <gene> g1f#4x#2y#3 (arg1 is #36 in call,arg2, arg3 is #30)
[18:48:19] <gene> g1f#4z0.0 (back to work surface)
[18:48:20] <gene> m3 (start spindle)
[18:48:20] <gene> g4p0.2 (slight pause for spindle spin up)
[18:48:20] <gene> (first cut is now 2x deeper)
[18:48:20] <gene> (g1f#1z[#5/5]) (#4 is arg4=#35, #5 is arg5=#38 in call)
[18:49:04] <cradek> looks like it should wait 0.2 seconds. (that doesn't seem very long)
[18:49:17] <gene> the real next line goes 2x deeper for the first cut.
[18:49:30] <jepler> if you change it to g4p2 you'll be sure to know whether it's pausing or not
[18:49:52] <cradek> yeah I don't see anything wrong
[18:50:16] <gene> last block for the day, I'm out of well seasoned 5% 2x6 stock
[18:50:42] <jepler> try testing without material on the table
[18:51:38] <gene> I'll try that for the first run of the next one. Yeah, duh! And my z screw is still rumbling on the downstroke, darnit.
[18:52:24] <gene> The screw is stationary, the nut is driven in this design.
[18:53:21] <gene> And it seems like it gets worse as the day warms up
[18:53:55] <archivist> change to a different grease
[18:54:10] <gene> stp atm
[18:54:40] <pjm__> archivist yeah that item u pointed out this morning is in my village, btw 190303679774 might be the other 1/2 of it?
[18:55:51] <archivist> pjm__, I wonder where the rest is/was
[18:56:22] <gene> I have a g-gun full of no name lithium chassis grease, would that be better than stp?
[18:56:47] <archivist> depends on bearing design
[18:57:19] <archivist> and reason for noise
[18:59:39] <archivist> may be stiction between screw and nut and resilience in the belt at a resonant speed
[18:59:54] <gene> two large axial thrust (end ways) designed sealed ball bearings turn the nuts, screw is 1/2" 10 tpi acme, nuts are bronze, all from McMastterCarr
[19:00:33] <archivist> not from ball bearings
[19:00:42] <motioncontrol> good evening.i want control with emc only one circular axis , not x,y and z axis.Is possible ?
[19:01:01] <gene> pix, sort of, at <http://gene.homelinux.net:85/gene/emc>
[19:01:39] <gene> and the belt is tight, very tight. had to to reduce the rotational backlash.
[19:01:41] <archivist> I saw the pics yesterday
[19:01:57] <gene> messsy ain't it? :)
[19:02:08] <archivist> er no comment
[19:02:17] <gene> chuckle...
[19:04:42] <jepler> motioncontrol: yes, it works fine
[19:04:44] <gene> I just gave the screw a shot of (non-lithium, it really is no name crap) and the rumble is changing pitch, but only slightly. It will take hours for it to work down to the lower nut.
[19:05:20] <motioncontrol> thenk jepler i declare only in inifile axis_0 = circular ?
[19:05:26] <jepler> motioncontrol: no, not at all
[19:05:34] <jepler> http://linuxcnc.org/docs/html/config_ini_config.html#sub:[TRAJ]-section
[19:05:46] <jepler> "his has no effect on the mapping from G-code axis names (X- Y- Z-) to joint numbers--for “trivial kinematics”, X is always joint 0, A is always joint 4, and U is always joint 7, and so on"
[19:05:50] <jepler> s/his/this/
[19:06:34] <jepler> http://emergent.unpy.net/files/sandbox/oneaxis.ini http://emergent.unpy.net/files/sandbox/core_sim1.hal
[19:07:48] <gene> must be cleanup time on that screw, the grease turned black instantly from the black coating on the screw I assume.
[19:09:14] <motioncontrol> ok thanks jepler i want realize one 4 axis for my maschine, indipendent at my maschine.
[19:11:56] <pjm__> gene great machine fotos!
[19:12:04] <roh> motioncontrol how do you plan to sync these?
[19:13:48] <gene> thanks, but I should haver taken some of the z axis disassembled, one pix=1000 words you know..
[19:13:52] <motioncontrol> whith digital io
[19:14:16] <gene> Next time I have it apart, I will do that.
[19:15:07] <archivist> gene, turning black can mean grinding and wear
[19:15:27] <motioncontrol> on my maschine i have fagor cnc, i write in plc one m fuction for start emc partprogram and read in position emc signal for write in plc fagor start program
[19:16:06] <gene> That screw is coated with some blackening agent, probably wearing away. When this last block is done, I'll run it up and clean the screw
[19:20:11] <BJT-Work> dang Glyptal Red 1201 is expensive at $55 for a 12.
[19:20:20] <BJT-Work> 75 oz aerosol
[19:32:07] <A4ndy_> in emc2 using servo3go SERVO, is a "circle movement" job would translate to a "multiple line movement"?
[19:32:58] <jepler> the output of emc's motion controller is time-sampled position, not movement primitives such as "line" or "circle".
[19:33:00] <A4ndy_> or they go curve circle in 1 single task?
[19:34:19] <A4ndy_> in emc2 source, which part of program in .c that would do time-sampled task...?
[19:34:57] <A4ndy_> are they really do "curve" ...not circle really?
[19:35:01] <jepler> directory.map is an overview of the emc directory structure.
[19:35:17] <jepler> to answer your specific question, src/emc/kinematics or src/emc/motion
[19:35:35] <A4ndy_> :)
[19:35:40] <A4ndy_> thx
[19:36:47] <A4ndy_> is it using p id or fuzzy logc?
[19:37:06] <A4ndy_> or both?
[19:37:39] <SWPadnos> neither
[19:38:20] <SWPadnos> the motion controller outputs a series of positions over time. it's up to other components (such as PID or step generators in HAL) to move to the new positions
[19:38:20] <A4ndy_> can u explain ? why not using thats such control?
[19:38:51] <A4ndy_> onboard ofcourse.
[19:39:23] <jepler> in some configurations, the commanded position from the motion controller is used as an input to pid
[19:39:27] <SWPadnos> so there is PID in HAL if you use servos or servo-like systems, but there is a clear separation of motion planning (control) vs. actuator control
[19:39:29] <A4ndy_> board of servo2go
[19:39:31] <jepler> this is different from using pid to find positions
[19:39:58] <jepler> and fuzzy logic has nothing to do with any of it
[19:40:24] <SWPadnos> the STG board is used as a DAC and encoder feedback device, and generally will use PID to get feedback to match command positions
[19:40:34] <SWPadnos> if you want to write a fuzzy logic controller instead of PID, you can do that
[19:41:01] <SWPadnos> but all it has to do is take the command position and feedback position as inputs, and output some sort of control value to make them match
[19:41:01] <A4ndy_> and what the different between kinematic and motion?
[19:41:04] <SWPadnos> (like PID)
[19:41:14] <A4ndy_> no, it do fine
[19:41:25] <A4ndy_> do perfect already
[19:41:35] <SWPadnos> kinematics is the conversion between cartesian coordinates (XYZ, ABC rotations) to whatever joints your machine actually has
[19:41:44] <SWPadnos> a joint is a single freedom of motion
[19:41:54] <SWPadnos> like a table slide or a rotator
[19:42:18] <A4ndy_> i see
[19:42:27] <jepler> if you're asking "why is THIS file in src/emc/motion and THAT file in src/emc/kinematics", there may not be a good reason except that we are very resistant to renaming source files.
[19:42:31] <SWPadnos> so you can use robots like Puma or SCARA, or hexapods, or whatever other construction you like, while still programming in cartesian coordinates
[19:43:39] <A4ndy_> i was interrest in what a file work of
[19:45:30] <A4ndy_> how pci could do realtime(in servo2go) ...since it a stream kind of
[19:45:32] <A4ndy_> ..
[19:45:47] <A4ndy_> to do that math
[19:46:19] <A4ndy_> is servo2go has kind of southbridge?
[19:46:34] <A4ndy_> fpga?
[19:46:51] <jepler> I don't know much about servo2go
[19:46:58] <jepler> I know more about mesa
[19:47:16] <A4ndy_> how about mesa? they work alot the same
[19:47:17] <jepler> in the case of mesa, a very short sequence of outb() instructions will cause one of the DAC outputs to change to a new output value
[19:47:23] <A4ndy_> realtime i mean
[19:47:25] <SWPadnos> I thought servo2go was more or less a DAC and encoder interface
[19:47:40] <jepler> another short sequence of inb() will read the number of accumulated encoder counts
[19:47:43] <SWPadnos> plus some generic I/O
[19:48:03] <archivist> seems a shop to me http://www.servo2go.com/category.php?cat=10018
[19:48:06] <A4ndy_> thats is fast
[19:48:40] <A4ndy_> so encoder reading would go directly to PC clock/clock ??
[19:48:54] <A4ndy_> (clock per clock of it)
[19:49:18] <SWPadnos> the card would have a counter that counts up or down, and the PC would read the count
[19:49:21] <A4ndy_> rather than buffer it up on mesa
[19:49:24] <alex_joni> cool: http://www.tuxradar.com/content/road-jaunty-look-back-ubuntus-history
[19:49:33] <SWPadnos> there is no buffering, it's a counter
[19:49:46] <A4ndy_> sure.. buffered on mesa, yes i see
[19:49:55] <SWPadnos> if there are 100 counts since the last time the PC read the register, then the number will be 100 higher than before
[19:49:58] <A4ndy_> :> Very interresting
[19:50:49] <A4ndy_> and who decide the servo going too fast or not? PC / mesa?
[19:50:56] <SWPadnos> the PC
[19:51:30] <gene> I had to add an m5 at the top of the syb, for some reason the spindle was running before it was told to on the first hole. Odd.
[19:51:36] <SWPadnos> every millisecond (by default, this can be changed), the PC reads feedback positions, calculates a new target position, and runs the PID calcualtions to move to the new position
[19:51:44] <gene> And the delay had to be 0.8 secs to be effective.
[19:52:03] <gene> s/syb/sub
[19:52:07] <A4ndy_> i pc willing to stop on 100 count ..what would it do? going further and get back?
[19:52:25] <SWPadnos> I don't understand your question
[19:52:56] <SWPadnos> if there is enough of an error between commanded and actual positions, then a following error occurs and the machine stops
[19:53:05] <SWPadnos> the software looks for that
[19:53:07] <A4ndy_> u said before that if pc read 100 count, mean realworld goes more than that
[19:53:26] <SWPadnos> no, I said that there isn't "buffering" of encoder counts, there's a counter for that
[19:53:56] <kirk_wallace> Hello, I have an AXIS problem. When I select "Show hidden files" in the Open (file) window, AXIS locks up. I wanted to see hidden files because my Samba share is mounted on ~/.gvfs.
[19:53:58] <SWPadnos> so the card doesn't store up 100 edges for the PC to read, it adds or subtracts from a counter, and the PC reads the resulting count
[19:54:42] <A4ndy_> i get it
[19:56:46] <jepler> kirk_wallace: yuck
[19:57:19] <A4ndy_> could u tell me which part of file source .c that do adds or subtracts? ... is it motion?
[19:57:33] <jepler> mu
[19:57:39] <SWPadnos> adds or subtracts what?
[19:57:40] <jepler> it's a function of the hardware
[19:58:08] <A4ndy_> sorry ...do inputing encoder
[19:58:21] <A4ndy_> and process it
[19:58:29] <jepler> it is in the hal hardware driver (e.g., src/hal/drivers) to perform I/O to a specific hardware interface
[19:58:40] <A4ndy_> all movement critical there
[19:59:20] <A4ndy_> thx
[19:59:57] <jepler> kirk_wallace: say you make a folder ~/test, and in it make another folder .test
[20:00:08] <jepler> kirk_wallace: if you go to ~/test, then click "show hidden files and directories", does it lock up?
[20:00:15] <A4ndy_> 3d cartesian is perfect room for robotic
[20:00:43] <kirk_wallace> I'll try ~/test
[20:00:49] <A4ndy_> :)
[20:00:58] <jepler> I don't have a ~/.gvfs (I don't use gnome) but creating a .test directory, everything seemed to be fine
[20:02:21] <archivist> jepler, just replicated the lockup on 2.2.8 seems worse
[20:03:15] <archivist> lost mouse click response but the pointer is moving around the screen
[20:03:56] <jepler> can you tell me what steps you did?
[20:04:44] <kirk_wallace> Locked up for me, I'll try to dismount my Samba share and try again.
[20:06:22] <archivist> for me loaded axis, clicked open file, clicked show hidden...dead ish box, not responding to keyboard or mouse clicks, is serving ok
[20:06:25] <jepler> kirk_wallace: my suggestion with ~/test/.test ?
[20:06:48] <archivist> I have no shares or samba
[20:08:12] <kirk_wallace> Dismounted share. Then opened AXIS (okay so far), selected open file, changed to ~/test, see no files (okay so far), click veiw hidden, locked up.
[20:08:53] <A4ndy_> im going log off, thanks alot to you all.
[20:09:01] <A4ndy_> :)
[20:09:13] <jepler> I believe you
[20:09:17] <jepler> unfortunately, it won't lock up for me
[20:09:42] <jepler> after starting it but before locking up axis, open up a terminal and execute: strace `pidof -x axis`
[20:09:46] <jepler> (you may have to install strace first)
[20:09:58] <jepler> make it lock up, then show me the last few lines of output (say, 20 lines) using pastebin.ca
[20:10:14] <archivist> looks like I have to power down pulling the power to get control back
[20:10:26] <alex_joni> archivist: try ssh first
[20:10:33] <alex_joni> should be ok for strace
[20:10:35] <jepler> or ctrl-alt-f1 or ctrl-alt-backspace
[20:10:50] <jepler> but if you can't get it back you'll have a hard time putting lines of strace on pastebin for me :-/
[20:10:57] <alex_joni> ctrl-alt-backspace would kill AXIS..
[20:11:15] <archivist> ctrl alt backspace killed it
[20:11:37] <jepler> hardy? dapper?
[20:12:37] <jepler> 2.2.8? 2.3.0?
[20:12:46] <jepler> (I see archivist mentioned using 2.2.8)
[20:12:55] <archivist> mine is 8.04 from the live cd plus upgrades
[20:13:32] <archivist> ok got it back up...try again
[20:13:35] <kirk_wallace> I'm a bit slow, sorry. I'll look into the trace and get back. Ctl-Alt-F1 then killall -s 9 axis works for me. I'm using 2.3.0 and axis sim config.
[20:13:44] <jepler> strace -o /tmp/trace.out `pidof -x axis`
[20:13:57] <jepler> ^^ will save it to a file in /tmp so you can read it after using ctrl-alt-backspace again
[20:14:39] <jepler> hm, I tested 2.3.0~beta2 on hardy and CVS TRUNK on dapper and still haven't had a lockup..
[20:16:57] <jepler> kirk_wallace: maybe you can get working again by creating a symlink to the share you need
[20:17:12] <kirk_wallace> Another funny thing, while at the command line, I "ls -al" and get"d??????? ? ? ? ? ? ./gfs" other dir and file listings seem okay.
[20:17:32] <jepler> e.g., cd ~/emc2/nc_files; ln -s ~/.gvfs/somethhing/something short-name
[20:17:46] <jepler> kirk_wallace: hm that doesn't sound good
[20:23:42] <kirk_wallace> Dooh, I used Nautilus to open the share then copied my file to my ./emc2/ngc_files folder (unhidden). This worked okay and gives me time to look into the lockup. Thanks for the help.
[20:24:35] <archivist> repeatable !
[20:26:21] <jepler> archivist: since I can't make it happen here, the nearest useful thing I can think of is that strace debug output..
[20:26:27] <archivist> dam lcd on the box goes mental when the system shuts down
[20:27:02] <archivist> may have to kick it again to get back in but strace was running
[20:27:37] <archivist> * archivist used the power switch
[20:28:09] <archivist> Im guessing it could be deeper and not axis
[20:29:27] <alex_joni> good night all
[20:29:31] <jepler> see you alex_joni
[20:31:20] <archivist> jepler, no trace.out waiting for me on reboot
[20:31:27] <jepler> archivist: crud
[20:33:04] <jepler> if only it would crash for me as well
[20:33:18] <archivist_emc> such a serious block to the system I cant help thinking its somewhere else completely
[20:35:00] <jepler> but surely some other program would tickle it
[20:35:14] <archivist> just looking for one
[20:35:15] <jepler> you and kirk_wallace are in the same boat in that axis is the only program known to cause this problem
[20:35:28] <jepler> axis uses the standard Tk file browser
[20:35:38] <jepler> let me see if I can tell you how to invoke the Tk file browser for yourself
[20:35:40] <jepler> (hmmm)
[20:35:46] <archivist> know of any others ?
[20:35:56] <jepler> ubuntu doesn't come with any standard programs that use Tk
[20:36:05] <SWPadnos> tkemc?
[20:36:26] <archivist> something non emc would be best
[20:36:45] <SWPadnos> well, if tkemc does it, then it's probably not AXIS ...
[20:37:00] <SWPadnos> if not, then it probably is axis
[20:37:04] <SWPadnos> since they both use tk
[20:38:10] <jepler> yes that's certainly one to try
[20:38:23] <jepler> though now that I look, there's apparently a hack in axis to get the "show hidden files" button
[20:38:38] <jepler> down http://emergent.unpy.net/files/sandbox/files.tcl and run it: wish files.tcl
[20:39:14] <crice> it hangs me up too
[20:39:26] <jepler> apparently it does for everyone in the world but me :-/
[20:40:51] <skunkworks> heh - happens here too.
[20:41:08] <skunkworks> I never noticed the 'show hidden' check box before
[20:41:14] <SWPadnos> has anyone mentioned which OS version they have?
[20:41:20] <archivist> files.tcl behaves
[20:41:29] <archivist> 8.04
[20:41:43] <jepler> well nuts
[20:41:57] <crice> Linux Millie 2.6.24-16-rtai #1 Tue Sep 30 22:54:33 EEST 2008 i686 GNU/Linux
[20:42:09] <skunkworks> 8.04 latest livecd
[20:42:25] <skunkworks> aj12
[20:43:16] <jepler> and the lockup happens after you click "show hidden files and directories"
[20:43:17] <jepler> ?
[20:43:21] <skunkworks> yes
[20:43:24] <jepler> before you do anything else, or the screen contents refresh?
[20:43:29] <invite> hi guys, does anybody know an electronic specific Irc channel on freenode?
[20:43:34] <archivist> ok now fired up emc and axis stops the show
[20:43:35] <skunkworks> the check doesn't actually check either
[20:44:22] <archivist> and the contents dont change in the window
[20:44:33] <skunkworks> you can go right into axis - file open - then try to check the checkbox.
[20:44:35] <jepler> aha! finally it did it for me too
[20:44:41] <crice> I just went to my home folder using axis open and clicked on show hidden.
[20:44:47] <jepler> you guys all click the box. I was clicking the words next to it
[20:45:02] <skunkworks> heh - odd
[20:45:22] <skunkworks> clicking on the words works?
[20:45:26] <archivist> odd bug of the year
[20:45:33] <jepler> skunkworks: it does for me
[20:45:53] <jepler> well now I know how I'll be spending my evening
[20:46:09] <jepler> thanks all for crashing your computers for me
[20:46:13] <skunkworks> does here also
[20:46:50] <skunkworks> What ever you do - DON'T CLICK ON THE CHECK BOX ;)
[20:48:02] <kirk_wallace> Don't click on "Click Me"? Sorry, if I caused a fuss.
[20:48:22] <jepler> kirk_wallace: it's a very real bug, and thanks for reporting it when you discovered it
[20:48:49] <archivist> at least we could all reproduce
[20:50:23] <jepler> besides the workaround of clicking on the words instead of the box, I have a fix: http://emergent.unpy.net/files/sandbox/file-open-dialog-crash-workaround.patch
[20:51:13] <jepler> but it requires a recompile, unfortunately
[20:51:31] <jepler> anyway, if I can't figure out a real fix I'll put this in the next version
[20:51:36] <archivist> ive never needed to load a hidden file I can wait for an update
[20:55:45] <SWPadnos> strange. I have an strace from Jaunty if you'd like it
[20:55:53] <SWPadnos> strace aborted though, which I thought was odd
[20:57:00] <jepler> SWPadnos: I'm not sure -- when I asked for the strace, I thought it was probably some file access problem
[20:57:11] <jepler> now I think it's an event handling problem, and strace won't show that very well if at all
[20:57:15] <SWPadnos> well, there is some sort of file access problem
[20:57:18] <jepler> and I can reproduce it, so I can get whatever output I want
[20:57:42] <SWPadnos> there are a bunch of lseek calls at the end of my log
[20:58:11] <SWPadnos> with lots of octal characters instead of usable names
[20:59:28] <SWPadnos> http://www.linuxcnc.org/wiki/uploads/axistrace , if you want to take a look
[20:59:29] <JymmmEMC> SWPadnos: !!!
[20:59:35] <SWPadnos> hi JymmmEMC
[21:00:03] <JymmmEMC> SWPadnos: Haven't seen you around much
[21:00:16] <SWPadnos> no, I've been on vacation the last week
[21:00:24] <SWPadnos> net access is expensive from a cruise ship :)
[21:00:27] <JymmmEMC> SWPadnos: Cool, where to?
[21:00:40] <archivist> port to port
[21:00:53] <SWPadnos> Nassau, Half Moon Cay (best beach I've ever seen), and Grand Turk Island (fun diving there)
[21:00:55] <JymmmEMC> SWPadnos: Your first cruise?
[21:00:59] <SWPadnos> nope, third
[21:01:17] <SWPadnos> I now look like I haven't been in a basement for hte last 10 years :)
[21:01:22] <JymmmEMC> SWPadnos: Are those places in the US?
[21:01:26] <SWPadnos> nope
[21:01:35] <SWPadnos> we left from and returned to Miami though
[21:01:39] <SWPadnos> which is more or less in the US
[21:01:49] <JymmmEMC> ah
[21:02:07] <SWPadnos> hmmm. I should make a smaller strace log
[21:02:59] <JymmmEMC> SWPadnos: Yeah, it's too early in the year to make a Yule Log
[21:03:33] <SWPadnos> heh
[21:04:25] <SWPadnos> so that's interesting. clicking the hidden files button causes strace to die with a glibc error (malloc(): memory corruption ...), but axis works correctly
[21:13:50] <invite> Can someone give me a hint about an IC that can drive a stepper
[21:14:10] <invite> at about 2 A / 40 V thus delivering a power of almost 80 w
[21:14:39] <SWPadnos> there are two commonly used ones, though I don't know that either goes to 40V
[21:14:52] <SWPadnos> one is by Allegro, the other by IR(?)
[21:15:05] <SWPadnos> I can't remember the chip numbers at the moment
[21:15:22] <archivist> L297/298 is ST
[21:15:25] <invite> They can drive motors at 80 w (I mean big stepper for plasma cutter tables for example )
[21:15:43] <SWPadnos> yes, but not that way. I think they're 2.5A, 35V
[21:16:06] <SWPadnos> archivist, thank you. I couldn't remember those
[21:16:17] <SWPadnos> I think the Allegro is the 3977 or 3985 or something like that
[21:16:50] <archivist> Allegro datasheet looks useful
[21:17:05] <SWPadnos> I think people had trouble with the chip, but I don't remember the details
[21:17:31] <jepler> A3977 is the driver chip in the board I use
[21:17:34] <jepler> it works well for me
[21:17:42] <invite> seems from what you say that both of you don't use steppers
[21:17:47] <invite> :)
[21:17:50] <invite> oups
[21:17:59] <archivist> I do use steppers
[21:18:02] <invite> and what motors do u use jepler
[21:18:07] <invite> and archivist
[21:18:28] <MrSunshine> somehthing abut microstepping problems on a allegro chip but i dont remember the model
[21:18:53] <MrSunshine> * MrSunshine would give his right nut for 3 strong servo motors and controllers =)
[21:18:54] <archivist> http://www.arceurotrade.co.uk/Catalogue/Stepper-Motors
[21:19:07] <jepler> invite: my motors are small, NEMA 17, 5W
[21:19:14] <archivist> I got the drivers as well from there
[21:19:19] <invite> Thanks alot guys
[21:19:49] <jepler> 2A / 40V doesn't mean you get 80W of power
[21:19:56] <archivist> Im running about 28 volts 2.4 amps
[21:20:01] <invite> heh ?
[21:20:06] <jepler> usually with stepper drivers the motor voltage is much lower than the supply voltage
[21:20:28] <jepler> e.g., my motors have a nominal voltage of 2.42V but I use a 27V supply
[21:21:04] <archivist> we only use a high voltage to improver the response
[21:22:00] <invite> and how much power do you get from them?
[21:22:20] <invite> (without taking into account reducers)
[21:22:36] <jepler> the motor is rated 75 oz-in = 0.53 N-m
[21:22:42] <archivist> I dont think in terms of power but torque at a speed
[21:22:46] <jepler> presumably that's holding torque
[21:23:29] <archivist> mine being 180Ncm at 2.4 a holding torque
[21:27:10] <invite> I see no maximum voltage specified for the motos
[21:27:24] <invite> Is it then chip driver dependant?
[21:27:43] <archivist> yes
[21:28:02] <invite> aheuh things are getting more and more clear
[21:28:14] <archivist> and they limit the current by pwm
[21:29:00] <jepler> http://www.geckodrive.com/upload/Step_motor_basics.pdf discusses the issue of maximum supply voltage on page 13.
[21:29:10] <invite> another mysterious thing that remains
[21:29:20] <invite> how a small IC like the allegros
[21:29:30] <invite> can handle as much current without
[21:29:36] <invite> getting HOT !!
[21:29:52] <invite> They dissipate almost 5w
[21:29:53] <jepler> generally you'll heatsink the driver chips
[21:30:30] <jepler> they are switching devices, not linear, so they won't dissipate as much as you may be thinking
[21:30:53] <invite> aheuh so basically they only dissipate when switching
[21:31:15] <jepler> look at datasheet figures like "output on resistance" to gauge power dissipation in the driver
[21:31:43] <jepler> A3977 = 0.57 + 0.43 ohm max = 1 ohm max
[21:32:14] <jepler> so at 2A motor current you'll dissipate 4W in the A3977
[21:33:22] <jepler> not a huge amount, but enough to imply you should heatsink
[21:35:11] <jepler> bbl
[21:35:44] <invite> Another question Does anyone use microstpping and does it really value the détour
[21:36:06] <jepler> I'm not sure what you mean by "détour"
[21:36:11] <invite> or should I keep with old good 1/2 step
[21:36:21] <archivist> I use halfsteps
[21:36:23] <invite> worth it (soory french guy)
[21:36:25] <jepler> I run my boards in 1/8 mode, but mostly to show off the high step rate I can do
[21:37:17] <jepler> I wouldn't cross the street for it
[21:37:42] <invite> :) crystal clear
[21:38:02] <invite> I'm really gratefull guys
[21:38:15] <invite> you really eally AND I mean REALLY helped me a lot
[21:38:26] <jepler> you're welcome
[22:17:07] <Jon_geo01005> SWPadnos: Somebody said that you have tried out the new gecko G251/G250 stepper drivers?
[22:24:52] <JymmmEMC> How do you keep a 8ft leadscrew from whipping? Or does it really matter
[22:25:08] <JymmmEMC> We're talking a hot wire foam cutter here.
[22:30:10] <Jon_geo01005> you could rotate the nut instead of the screw. That is what I'm doing on my machine.
[22:30:40] <JymmmEMC> how are you doing that?
[22:31:37] <archivist> Ive also seen long leadscrews with supports that move out the way of the nut as it passes
[22:32:03] <Jon_geo01005> lets see, I have a video on youtube, but I can't get to it here. Youtube is blocked on our campus.
[22:32:18] <JymmmEMC> utl or name?
[22:32:53] <Jon_geo01005> this isn't the video, but just look at my other videos : http://www.youtube.com/watch?v=txx586M2r9s
[22:33:45] <invite> Jon_geo01005: use a http proxy ;)
[22:34:31] <Jon_geo01005> I suppose I could. Usually I just wait till I'm home.
[22:35:09] <Jon_geo01005> You can't really see how it works from the video if I recall.
[22:35:50] <Jon_geo01005> Some of the easiest designs I have seen is drilling a tapping a thru hole in a stepper motor shaft.
[22:36:05] <Jon_geo01005> Then the threaded rod just goes right thru the motor.
[22:36:13] <Jon_geo01005> You can buy them that way.
[22:39:11] <archivist> JymmmEMC, under the walk way rocking assemply and u cups http://www.collection.archivist.info/archive/DJCPD/PD/2001_04_23_Claymills_roof_before_painting/apr20010009.jpg
[22:40:32] <archivist> JymmmEMC, that one is 40ft long but only rotates slowly
[22:43:05] <archivist> another at 30ft but not visible http://www.collection.archivist.info/archive/DJCPD/PD/2001_09_27_Claymills_Open_Day/P1010345.JPG
[22:49:50] <Jon_geo01005> Hot wire cutter, you could use that really neat dual bipod design :)
[22:50:07] <Jon_geo01005> Then you don't even have to worry about a lead screw :)
[22:58:32] <JymmmEMC> dual bipod?
[22:58:37] <JymmmEMC> archivist: thanks
[23:00:37] <JymmmEMC> archivist: Why didn't you climb up there and get close-up pics too
[23:01:20] <archivist> I helped free then when they were seized
[23:02:00] <JymmmEMC> archivist: I only see the mechanics on one side, how does it move the support when traveling the opposite direction?
[23:02:26] <archivist> lever gets hit and it rocks
[23:02:56] <JymmmEMC> Oh, is that what the added "nubs" on the gear wheel are for?
[23:04:01] <JymmmEMC> Well, either way, I think a lil too elaborate for this. Maybe belts instead. it's only moving a wire.
[23:11:52] <archivist> on that particular shaft the gear is driving via a slot in the shaft and the trolley moves left right
[23:17:57] <Jon_geo01005> JymmmEMC: A bipod sort of like this one : http://www.youtube.com/watch?v=DAPTdw_BTF0
[23:42:07] <tomp> tomp is now known as tomp3
[23:44:48] <JymmmEMC> Ok, I've seen that before (with a spray can of paint). What are they using for the straps... toothed belt?
[23:46:50] <Jon_geo01005> I'm not sure. I think that timing belts are a good idea though.
[23:47:26] <JymmmEMC> I guess ion the motors, a toothed gear and a tension pulley of some sort.
[23:47:43] <JymmmEMC> to prevent the belt from coming loose
[23:49:08] <Jon_geo01005> Sure. There was a video posted on here a while ago that actually showed a dual bipod hot wire foam cutter, I think it only used wire wound around a pulley.
[23:49:16] <SWPadnos> Jon_geo01005, I have made a config for hte Gecko G540, which contains 4 G250 drivers in it
[23:49:54] <Jon_geo01005> SWPadnos: I see, I was just wondering how well they work.
[23:50:03] <SWPadnos> they seem great
[23:50:32] <JymmmEMC> SWPadnos: What you think... toothed belt or wire on a pulley?
[23:50:36] <Jon_geo01005> much cheaper than the others.
[23:50:39] <SWPadnos> I happened to be using a very good latency PC for testing, and I had the stepper running at close to 50ksteps/sec
[23:51:06] <SWPadnos> JymmmEMC, without knowing any more than that question, I'd say toothed belt. wire-on-pulley could slip
[23:51:51] <JymmmEMC> SWPadnos: bipod http://www.youtube.com/watch?v=DAPTdw_BTF0
[23:52:08] <JymmmEMC> SWPadnos: for a cnc foam cutter
[23:52:20] <Jon_geo01005> SWPadnos: do you remember somebody putting a link up for a dual bipod foam cutter?
[23:52:33] <SWPadnos> Jon_geo01005, no, not really
[23:52:44] <SWPadnos> JymmmEMC, is that what you'd want the wire/belt for?
[23:53:37] <JymmmEMC> SWPadnos: yeah
[23:54:00] <SWPadnos> hmmm. I don't know then :)
[23:55:23] <JymmmEMC> SWPadnos: bipod just seems reasonable for a foam cutter
[23:55:43] <SWPadnos> you'd never be able to cut accurately going downward
[23:55:59] <JymmmEMC> SWPadnos: weighted of course
[23:56:03] <SWPadnos> unless the hot wire literally cuts the foam like melted butter
[23:56:15] <SWPadnos> ok, weights could make it work
[23:56:38] <Jon_geo01005> The one I saw had largish weights.
[23:57:15] <SWPadnos> you also have tension on the cutting wire itself
[23:57:29] <JymmmEMC> ah, crap...
[23:57:31] <SWPadnos> there's no particular reason why the ends will move in a plane
[23:59:32] <LawrenceG> someone did post a video link on here in the last couple of weeks.... it used a bow holding the hotwire tight that was suspended at each end by 2 wires