#emc | Logs for 2005-10-20

[00:00:00] <icee> they are probably overkill for your app
[00:00:29] <icee> lg: first, have you looked at: http://www.edn.com/contents/images/52704di.pdf
[00:01:07] <icee> the IGBT i'm looking at using is: http://www.irf.com/product-info/datasheets/data/gb25xf120k.pdf (overkill for 300W)
[00:01:47] <icee> http://www.irf.com/product-info/datasheets/data/ir2131.pdf <- that for an IGBT/FET driver
[00:02:13] <icee> and then for the pulse counting/pwm generation.. undecided right now between microcontroller/DSP and FPGA
[00:02:21] <alex_joni> * alex_joni shivers
[00:02:30] <alex_joni> I tried some IR2111 iirc
[00:02:32] <icee> the encoder count rate is almost 1MHz
[00:02:38] <alex_joni> were the crappiest I ever did
[00:03:01] <icee> alex; really? I think it was an ir2104 i used before; had good luck with it
[00:03:06] <icee> alex: what problem did you experience?
[00:03:30] <alex_joni> they kept blowing up
[00:03:44] <alex_joni> sometimes the driver, sometimes the fets
[00:03:45] <alex_joni> :D
[00:03:57] <icee> lg: I'm planning on rectifying and filtering 220VAC to get 340VDC to chop up
[00:03:59] <icee> alex: any idea why?
[00:04:13] <alex_joni> so I sacked that, and used P & N typed FETs
[00:04:18] <alex_joni> icee: nope
[00:04:43] <icee> lg: though I'm moderately scared of 340VDC :P
[00:05:20] <icee> alex: hm. blowing up $70 IGBTs would suck.
[00:05:23] <les> 240...
[00:05:40] <LawrenceG> icee, thanks for links... I was looking at ir2109 drivers and not sure whether fets or igbts's using 120vac rectified
[00:06:06] <icee> les: yah, 240, my bad.
[00:06:17] <les> heh
[00:06:25] <LawrenceG> icee, search for an889 design on microchips pages
[00:06:58] <icee> lg: I'm personally planning on integrating against the pico systems parallel port servo interface in EMC, btw
[00:07:26] <icee> lg: well, if we do things right we ought to be able to share the control logic at least between our designs
[00:07:29] <LawrenceG> yea... I have had a few fets explode.... les smoke as they mostly just dissappear
[00:07:48] <icee> even if the power stages are completely different
[00:08:44] <icee> lg: the hard thing isn't generating the output waveform.. it's handling the pulses from the encoder
[00:08:58] <les> I had an oops today...the vfd sent a remote keypad that in fact will not work with that particular model...no prob. I can talk to it with rs-285
[00:09:00] <icee> filtering them, coping with noise, dealing with the pulse rate, etc.
[00:09:01] <LawrenceG> I am wondering about a mesa card interface or a cheaper idea that doesnt use an interface card....
[00:09:21] <icee> lg: well, the pico systems interface for their stuff uses an ECP parallel port to read/write registers
[00:09:32] <icee> I'm going to, in whatever my control hardware is.. implement the same interface so i can use their drivers
[00:09:45] <icee> the drivers in emc, that is. none of their hardware
[00:10:06] <les> Jon is claiming 10k servo updates with the pico. I would like to see that.
[00:10:19] <icee> les: well, if emc can do it, my hardware will be able to
[00:10:29] <alex_joni> icee: no drivers on pico in emc2 yet ;)
[00:10:32] <alex_joni> but on the way
[00:10:35] <icee> alex: really? eek.
[00:10:37] <LawrenceG> I was thinking of i2c in software via the parallel port.... ie talk to 3 or 4 drives at one.... 16 bit drive out and a 32 bit position read back with a few extra drive status bits.... about 1us/bit
[00:10:42] <les> we max out at 2k with the stg.
[00:10:49] <alex_joni> * alex_joni knows 2 people working on it
[00:10:53] <icee> lg: will that get you enough update rate? i2c is slow
[00:11:23] <icee> 400KHz / 48 bits.. plus addressing and overhead
[00:11:52] <icee> lg: the nice thing is if we just use the pico protocol, there's no need to implement drivers ourselves in EMC
[00:12:18] <LawrenceG> agreed.... and transfers are low cpu load
[00:13:11] <LawrenceG> icee, question.... are you thinking of smarts on the driver like a cpu or just an fpga?
[00:13:34] <icee> I am not sure i can get the count rates / parallel port interaction rate with a CPU
[00:13:45] <icee> maybe with something like tms320f28xx I could..
[00:14:00] <alex_joni> motorola starcore 8102
[00:14:06] <alex_joni> I would suggest
[00:14:14] <alex_joni> but that could do a bit more
[00:14:14] <LawrenceG> some of the pic and and motorola chips have all the pwm and encoder interface in hardware
[00:14:22] <alex_joni> maybe TP
[00:14:25] <alex_joni> and others
[00:14:34] <alex_joni> hey jepler
[00:15:01] <icee> lg: yah, there's nice timer/counter facilities on atmel avr, too
[00:15:16] <icee> both in mainline parts and things like their at90pwm3 (which isn't quite available yet but will be shipping soon)
[00:15:25] <icee> I am not sure it's beefy enough for this app though.
[00:15:35] <LawrenceG> microchip is sending some dsPIC samples... but I am not sure I can program them
[00:15:59] <icee> then there's the AT91 or LPC21xx parts
[00:16:04] <LawrenceG> the an 889 note uses a pic16f7x7 whatever that is
[00:16:09] <icee> which frankly i think are more trouble than they are worth
[00:16:19] <alex_joni> icee: google the StarCore I mentioned
[00:16:23] <icee> the 889 doesn't need to deal with an encoder
[00:16:41] <icee> alex: i don't really like the freescale starcore parts to be honest
[00:17:27] <icee> if i were to do a dsp, i'd pick tms320f28x, because i already know the assembly language
[00:17:46] <icee> and it's cheap and easy to integrate
[00:17:54] <alex_joni> how about CodeWarrior for StarCore
[00:17:56] <alex_joni> :D
[00:18:02] <alex_joni> not very cheap.. but powerfull
[00:18:16] <icee> well, i own a 28x c compiler too
[00:18:39] <alex_joni> nicee
[00:18:40] <alex_joni> :D
[00:18:40] <dmess> with the correct dardware it sounds feasable and advetantgadeous...
[00:18:49] <LawrenceG> icee, will look at the ti stuff... does it do 3 phase pwm in hardware?
[00:18:52] <alex_joni> dmess can't type today
[00:18:53] <alex_joni> :D
[00:18:55] <dmess> hardware
[00:19:04] <icee> lg: yah
[00:19:13] <alex_joni> dmess: kidding :P
[00:19:15] <icee> lg: look at the atmel stuff too.. at90pwm3
[00:19:17] <dmess> i asked for help on the big word at the end
[00:19:34] <dmess> and sti facked it up..
[00:19:40] <icee> lg: personally i'm inclined towards the FPGA route. everything is easy to implement except the pport interface and even that is not hard
[00:19:53] <alex_joni> http://solaris.cs.utt.ro/~emc/
[00:19:56] <icee> lg: and I could use a CORDIC to get much nicer sine wave data than some sine rom will, etc
[00:19:58] <dmess> still... its the flippin' keyboard
[00:20:02] <LawrenceG> I like micros.... adjustable after the fact....
[00:20:05] <alex_joni> should have daily tar.gz's with emc2
[00:20:22] <icee> lg: micros are adjustable too; synthesize some more HDL and program the config flash
[00:20:25] <icee> lg: er fpgas.
[00:20:34] <icee> lg: I can even include a microcontroller -on- the fpga if you want ;)
[00:21:34] <dmess> make it so # <icee>
[00:21:54] <LawrenceG> yea... mesa does some of that..... I am trying to eliminate addin cards like encoder cards... picos' interface maight work well for reading back positions
[00:22:42] <icee> yah, pico's interface is predicated on a 24 bit position counter
[00:22:51] <icee> which is sufficient for 16 seconds of counts at 4500RPM, so..
[00:23:45] <LawrenceG> even if its expanded to 32 bits, it wouldnt involve a major driver mod
[00:24:07] <icee> I think 24 bits is plenty, with two's complement math
[00:24:27] <icee> you only get ambiguity if you're updating less than once every 8 seconds
[00:24:32] <LawrenceG> icee, what are you planning on sending as a drive signal.... a torque value? a velocity?
[00:25:01] <icee> I believe the pico system sends a torque value
[00:25:08] <icee> I've not read the driver yet, that's one thing i need to do.
[00:25:30] <LawrenceG> icee, I would like to avoid having to tune control loops on the drive and leave them all in emc
[00:25:33] <icee> so the position and velocity loop runs in emc; the torque and current loops run on my hardware
[00:25:51] <LawrenceG> icee, best way I think....
[00:25:51] <les> it really doesn't matter as far as emc...torque, velocity, and voltage will all work fine
[00:26:31] <les> the pid coefficients will be different
[00:26:38] <LawrenceG> les, but having to have a user interface on the drive for tuning it adds a lot of complexity
[00:26:42] <icee> i do want extra gates so i can move additional functionality to the board if needs be
[00:27:35] <les> Lawrence: yeah, I am partial to torque mode. It is the simplest with thw widest bandwidth generally.
[00:27:39] <icee> gates / microcontroller cycles / whatever
[00:27:39] <dmess> all REAL cnc's have a drive controller dealin' with stuff too.
[00:28:00] <LawrenceG> icee, on the EDN page, where you refering to the inverter (fet drive cct)?
[00:28:05] <icee> dmess: I don't care about it being a 'REAL' cnc. I care about what performance i get out of it
[00:28:14] <icee> lg: it's on the last couple pages, yah, the inverter
[00:28:16] <icee> sorry about that
[00:28:34] <icee> lg: the design is kinda sketchy, but is apparently proven at the power levels you're interested in
[00:28:37] <dmess> then make it rock and roll like the rest
[00:28:38] <icee> and simple and cheap
[00:28:43] <LawrenceG> icee, yea saw it... looks cheap to build for the lower power stuff
[00:29:26] <les> fact is...pid doesn't know dynamics or even kinematics. it just knows commands and errors.
[00:29:47] <dmess> but it sounds like you've got a good case of a HYBRID... so run with it.. ; )
[00:30:11] <LawrenceG> with a meas i/o card, one could probably drive 3 of those and use the mesa encoder interfaces... but I really want to get rid of the $200 card in the PC
[00:30:40] <dmess> fine so we add dynamics and kinematicts within emc2
[00:31:10] <les> dmess: it has it!!! it's called feedforward!
[00:31:37] <icee> les: If I really thought I needed that, it wouldn't be hard to do a nice analysis of the control system by root locus and replace the PID crap, i think.
[00:31:48] <icee> but PID is pretty good.
[00:31:57] <dmess> ok the its done.. add masses and speeds and max's
[00:32:04] <LawrenceG> all we need from the drive is "where are you" and a "step on the gas" command
[00:32:33] <dmess> how to step on the gas is another...
[00:32:34] <icee> lg: yah. to write to the drive a nice signed 'effort' value, and to read back the encoder counts
[00:32:44] <dmess> how to slow down..
[00:32:56] <icee> the how to slow down is a good point
[00:33:05] <icee> i mean, I have a brake. I will have to figure out how to best deal with that
[00:33:06] <les> Well we do root locus sometimes. Those use linear operators though. the machine generally ain't.
[00:33:13] <icee> (e.g. have a deadband of effort where it engages, or..)
[00:33:13] <dmess> all proportionate to the length of the move
[00:33:14] <les> but
[00:33:14] <alex_joni> LawrenceG: there are 2 modes to move.. you can set up a position loop
[00:33:22] <les> pid doesn't care!
[00:33:27] <alex_joni> prescribe position, drive must follow
[00:33:34] <alex_joni> or.. speed loop
[00:33:55] <LawrenceG> or torque command like dc motor
[00:34:19] <LawrenceG> its is emcs job to decide what speed or position really is
[00:34:35] <LawrenceG> and how to keep it where its needed
[00:34:37] <alex_joni> * alex_joni suggests http://www.maxonmotorusa.com/files/download/pdf/epos-firmware-specification-e.pdf
[00:34:45] <alex_joni> a very nice info material
[00:34:48] <LawrenceG> looking
[00:34:49] <les> well with a velocity command you would use little or no d in emc. the velocity amp p replaces it
[00:35:11] <dmess> moi aussi
[00:36:12] <dmess> les is a lot more knowledged
[00:36:15] <alex_joni> just found it today
[00:36:24] <alex_joni> dmess: it's an introductory material
[00:36:34] <alex_joni> * alex_joni was looking for info on homing :)
[00:36:56] <LawrenceG> a velocity amp ends up needing tuning of the amp itself.... could be a pot or digital via some ??? interface
[00:37:18] <dmess> right now cnc machines are a dime a dozzen around here places are droppin like flies
[00:37:19] <LawrenceG> Thanks... 168 pages more to read :}
[00:37:25] <icee> lg: well, the parallel port can be used for tuning, too, if the pico protocol is appropriately extensible
[00:38:14] <dmess> everyones learnin' to eat rice with sticks..
[00:38:26] <LawrenceG> yea... I need to have a look at the ecp modes.... its a nice way top move bytes fairly easily with simple strobes and acks done mostly by th pp hardware
[00:39:17] <icee> lg: yah. we'd even have DMA if we wanted it.
[00:39:47] <icee> lg: build a command set and have the parport send it out.. no bitbanging/tying up the cpu in a high priority realtime task.
[00:39:57] <LawrenceG> its interesting how in China.... everyone has a machine in the garage and one guy makes 10-32 screws his whole life...
[00:40:43] <dmess> there your hired for life... here your hired for a nite...SLUT
[00:40:57] <les> yeah...manufacturing in the US is kinda wild now...I know
[00:40:59] <LawrenceG> icee, easy to talk to fpga or micro as well for reading writing position and drive registers
[00:41:24] <icee> lg: yah. so i like the idea, especially with proven drivers already existing for it
[00:41:38] <LawrenceG> icee, sold....
[00:41:49] <icee> lg: so we just need to decide on what to implement it with ;)
[00:41:53] <dmess> europe is the same.. they hire kids at 16 for life with the company
[00:42:28] <les> I am hiring retired people.
[00:42:33] <les> mostly.
[00:42:51] <dmess> well thats for life.. ; )
[00:42:57] <les> heh
[00:43:11] <LawrenceG> pension plans are cheaper if you start at 70
[00:43:38] <dmess> but at 16 you can teach them EVERY aspect of ALL your equipment
[00:43:48] <dmess> from then ON
[00:43:59] <les> yeah. Well I have a very limited labor force out here. But it is a popular retirement area.
[00:44:12] <les> Many are highly skilled...and bored.
[00:44:39] <LawrenceG> what a waste....
[00:44:45] <les> yeah
[00:45:25] <LawrenceG> if one is lucky enough to be able to make work fun, then it makes life in general fun
[00:45:51] <les> I find that they go a little slower of course...but their vast experience makes up for it
[00:46:21] <les> Harder to find high tech though.
[00:46:25] <icee> I don't really 'get' how most of the population-- even the engineering and creative populations
[00:46:30] <icee> has lost the idea of work being fun
[00:46:46] <les> It's fun to me.
[00:46:53] <icee> that's not to say there aren't unfun things about it.. but..
[00:46:53] <les> I just play all day.
[00:46:59] <les> I will never retire.
[00:47:02] <icee> I'm at my best and happiest when I'm doing new things and learning and working.
[00:47:07] <LawrenceG> yep
[00:48:21] <Jacky^> :)
[00:48:37] <icee> les: btw, is it a ridiculous idea to replace the spindle motor on my mini mill with one of the 300W servos?
[00:49:01] <icee> it seems like it has enough oomph to spin the spindle well, and i'd think i could use it for thread cutting and other things because i have precise control over rotation rates and positions
[00:49:04] <icee> maybe even pick and place.
[00:49:08] <les> You see, I seem to be at that dangerous point where I have exponential growth, and can't get staff to keep up with it-so I am turning down work.
[00:49:14] <les> Icee:
[00:49:29] <Jacky^> eh
[00:49:43] <les> 1 kW would not be out of the question for a mini mill (mill drill)
[00:49:45] <icee> (plus I'd get experience with driving these things before moving on to the big stuff)
[00:49:48] <Jacky^> ice velocity is not all
[00:49:56] <Jacky^> :)
[00:50:11] <icee> j: I know what torque is. :P
[00:50:25] <les> oh I suspect those cheap "ac servo" motors would make fine spindles
[00:51:15] <icee> i've never used a machine tool that has an 'intelligent' spindle before.
[00:51:34] <LawrenceG> icee, so to emulate a torque drive system on a bdlc motor.... does full drive just mean rotating the phases at the motor max rpms?... what does one do with the current control.... keep VF constant? how about at stop? hold current like a stepper?
[00:51:39] <les> belt drive or something of course...the bearings are not designed for big thrust loads prob
[00:52:05] <icee> lg: You keep track of the current position from the encoder
[00:52:17] <icee> lg: and then you drive a certain number of degrees 'ahead' of that position in phase angle between the three phases
[00:52:26] <icee> lg: of the right magnitude to get the torque you want.
[00:52:43] <icee> how far ahead you drive varies with speed.
[00:53:10] <icee> les: Yah.. maybe a planetary drive with a nice bearing. I would like to be able to address precise positions without worrying about slippage
[00:53:40] <icee> so that pick & place would become a possibility, or thread cutting
[00:53:47] <les> right
[00:53:58] <dmess> thrust loads on MOST cnc's are quite qritical
[00:54:15] <les> yes sure are
[00:54:53] <les> My new spindle docs reccomend 20 degree entry angle max
[00:54:55] <dmess> a spindle encoder would allow thd cutting
[00:55:22] <dmess> or co-ordinated spindle orientation
[00:55:48] <icee> the current spindle motor is a crummy 1/3 HP DC motor on the thing
[00:55:50] <dmess> to allow you to cut square corners
[00:56:27] <LawrenceG> icee, most of these drive ccts I am looking at are voltage drives with over current protection feedback, not true current drives
[00:56:52] <icee> lg: well, it's easy enough to measure current
[00:56:56] <les> mini mills are usually stiffness limited...so you can use pretty high powewr if the rpm is also high
[00:57:01] <dmess> dc motor ... encoder.. the right system.... your golden
[00:57:01] <icee> lg: and so if it can be measured, why not control current in a closed loop?
[00:57:12] <LawrenceG> current control could be done via cpu and feedback... my thoughs
[00:58:02] <icee> lg: You know, do 30KHz PWM or whatever, and adjust the coefficient you multiply the sine waves out based on whether you're over or under current
[00:58:18] <dmess> current from the amps to the cpu sounds like i need a new a new 'puter
[00:58:35] <icee> the current sense is just a resistor between the low side transistor and ground.. ADCed
[00:58:46] <icee> filtered and ADCed, that is.
[00:58:55] <LawrenceG> icee, np
[00:59:03] <les> yeah my spindle uses shunts to feed back current
[00:59:27] <les> and my servo amps too
[00:59:28] <icee> (it also could be used as an overcurrent trip for things like the ir2131 igbt/fet driver)
[01:00:12] <LawrenceG> icee, so we have the input and the output... now we just need to figure out the best piece to put in the middle
[01:00:15] <icee> i think it would be sexy to just command '2 amps clockwise' and have the microcontroller do it
[01:00:35] <icee> lawrenceg: Yah. And I'm still in the early phases of estimating that.
[01:00:37] <dmess> sounds like your closing the loop guys.... is the CODE really feasable??
[01:00:43] <icee> I don't think an AVR can do it. I think the TI DSPs can, but they're cumbersome.
[01:01:02] <icee> dmess: If we write it for a DSP it'll be under a thousand lines.
[01:01:25] <icee> HDL could be even shorter but the effort per line of HDL is higher
[01:01:41] <dmess> cost??
[01:02:13] <icee> dmess: for the 'control' portion? the biggest piece is spinning the pc board.
[01:02:19] <icee> the big dc power supply and the power stages are more expensive
[01:02:33] <dmess> still an add on card to the PC... right??
[01:02:38] <icee> dmess: nah, parallel port connected
[01:02:42] <LawrenceG> no pc cards
[01:02:53] <dmess> ok gooder
[01:03:24] <dmess> so were runnin' true parallel out
[01:03:28] <LawrenceG> power supply not bad as we can use line isolation transformers in the several kva sizes
[01:04:04] <icee> lg: You've got a good source for those?
[01:04:10] <LawrenceG> write a parallel drive command and read encoder with a parallel input cmd
[01:04:16] <dmess> who can deal with detailed drawings for that...??
[01:04:39] <icee> dmess: well, we need to select a part to implement with before we enter a schematic
[01:04:54] <icee> dmess: so that's the next 'to-do'. more research on parts and part selection.
[01:05:09] <LawrenceG> demes...I'm seriously working on it... money commited to motors already
[01:05:21] <icee> i really wish we could use an AVR or something else in a 'friendly' package
[01:05:23] <dmess> thank YOU for saying that
[01:05:28] <les> how many did you buy?
[01:05:49] <LawrenceG> 3 @ 300w
[01:05:52] <dmess> then YOU pick.. your box is the puinea pig
[01:05:56] <les> ah ok
[01:06:18] <icee> i bought 5 @ 1kW and 1 at 300W
[01:06:18] <dmess> am i outta place here Les??
[01:06:32] <les> naw.
[01:06:59] <LawrenceG> the 1kw are a little big for any of my machines.... would be great for a bridgeport converstion....
[01:07:07] <icee> lg: that's my ultimate goal
[01:07:12] <icee> I don't have anything to run them on yet.
[01:07:22] <les> Making a drive for this population of motors is not a viable commercial venture...but for your own use...why not?
[01:07:26] <icee> but I've been eyeing things when local machine shops go out of business etc
[01:07:53] <icee> les: Yah. I plan to open up the design.
[01:07:58] <icee> There's a lot of AC motors this stuff would work on.
[01:08:09] <les> sounds cool to me.
[01:08:20] <LawrenceG> liability wise... I can deal with small electrical fires better than the average homeowner :}
[01:08:24] <dmess> we nee to stop talkin' and do more CONCURRENT engineering... i plan on retiring with at least fanuc beaten...
[01:08:34] <les> heh
[01:08:44] <icee> lg: haha. stop worrying more than i already am
[01:09:38] <dmess> hey... they dont wanna know wat happened n my house last sunday.. ; )
[01:10:33] <les> ??
[01:10:54] <icee> worrying me more than i already am, that is
[01:11:43] <dmess> the dirt ..
[01:12:56] <dmess> we spoke of ... or was that day all a dream.../...Flasback...
[01:13:36] <dmess> im not sure ... last i saw was yellew smoke...
[01:21:37] <alex_joni> * alex_joni goes to bed
[01:21:41] <alex_joni> night guys
[01:21:55] <Jacky^> night alex_joni
[01:34:18] <Jymmm> dmess: It's Halloween, not 4th of July... quit setting off the smoke grenades INSIDE the house already!
[01:55:39] <les> jymmm: are your steps still slipping?
[02:17:50] <Jymmm> les: Haven't tried a long job today.
[02:18:09] <Jymmm> les Tried a short one and my clamps came loose TWICE!
[02:18:50] <les> ok. I have been dealing with increasing the Q fo PZT piezoceramics today.
[02:19:00] <les> for
[02:19:03] <les> no fo
[02:19:11] <Jymmm> les haws the results so far?
[02:19:14] <Jymmm> hows
[02:19:51] <les> I have a Q of 20. I want a q of 1000. I have to talk to suppiers.
[02:20:06] <les> suppliers
[02:20:11] <les> bluh
[02:20:14] <Jymmm> (thats a serious typo... how did I get from 'o' to 'a'? Not even next to each other or anything)
[02:20:37] <Jymmm> les is what you need available?
[02:20:48] <les> yeah I think so
[02:21:18] <les> thr o to a is temporal not spacial haha
[02:21:21] <Jymmm> cool, I've been finding that logisitic has been my downfall lately. And lately being a couple of years.
[02:21:53] <les> and...watching the storm
[02:21:56] <Jymmm> Even the v cutting I was doign today tourned out terrible
[02:22:02] <les> NHC hasn't a clue
[02:22:06] <Jymmm> really really bad.
[02:22:13] <les> how so?
[02:22:18] <les> fuzzed up?
[02:22:44] <Jymmm> no, the cutting itself. Some of it was even wavey on a straight cut
[02:22:49] <Jymmm> but only on the inside
[02:23:11] <Jymmm> some fuzz too I guess.
[02:23:35] <les> hmm
[02:23:47] <les> what feed and speed?
[02:24:20] <Jymmm> I was trying to v cut some graphics. spindle about 20K RPM, feed was 100IPM
[02:24:35] <les> ok sounds about right
[02:25:01] <les> so...wavy on one side in long straight cuts?
[02:25:11] <Jymmm> The plank was onyl 1/4" thick, so I think the CAM program may have just not gone in deep enough
[02:25:19] <Jymmm> les yeah
[02:26:06] <les> wonder if that was the side where the grain was separating during the cut
[02:26:30] <les> did it do this in any direction?
[02:26:38] <les> was it MOF?
[02:26:39] <Jymmm> it was red oak, and the wave cut was againest the grain
[02:26:53] <les> ok
[02:27:02] <les> deflection
[02:27:06] <Jymmm> heh, beeter not have any grain in MDF =)
[02:27:28] <Jymmm> feed too fast?
[02:27:33] <les> I'll bet it did not hapen in cuts across the board
[02:27:41] <les> no feed seems fine
[02:28:25] <Jymmm> les I can email pics, but it'll take a few to convert them down to a reasonable size
[02:28:39] <les> need to take a roughing cut...and a final cut only about .oo1 deep
[02:28:49] <les> yeah...I'd like a look
[02:29:43] <Jymmm> can your email handle big msgs?
[02:30:18] <les> yeah
[02:30:54] <les> 2 meg I think
[02:31:16] <Jymmm> ok, I'll TRY to keep it under that =)
[02:31:45] <Jymmm> each photo is 6MB and I took like 60 photos =)
[02:32:26] <Jymmm> which email addr?
[02:32:33] <les> you know...tool marks tell a story...like a phonograph record groove
[02:32:50] <les> just leswatts
[02:32:52] <les> at
[02:32:58] <les> lmwatts.com
[02:33:22] <Jymmm> Heh... I think that's actually why I took photos... to be able to look back at things and figure out WTH went wrong.
[02:34:09] <les> personally I prefer to figure out WTF went wrong.
[02:34:30] <Jymmm> lol
[02:34:33] <Jymmm> that too
[02:38:00] <Jacky^> yawwnnsss
[02:38:27] <Jacky^> les: my cousin was talking to me thi evening
[02:38:39] <les> yes?
[02:38:44] <Jacky^> he want to biul an elicopter :/
[02:38:53] <les> wow
[02:39:02] <Jacky^> wait ..
[02:39:09] <Jacky^> a mini-elicopter
[02:39:10] <les> I am a pilot, but not for helicopters
[02:39:15] <les> oh ok
[02:39:16] <Jacky^> i know :)
[02:39:23] <Jymmm> Fly me to the moon.......
[02:39:23] <les> they are very hard to fly
[02:39:36] <les> heh
[02:39:39] <Jacky^> he want to buil a mini-elicopter with a webcam
[02:39:53] <Jacky^> hard project ..
[02:40:00] <les> I have thought about getting a rotorcraft rating
[02:40:32] <Jacky^> good
[02:40:35] <les> jacky I have seen many web sites with video cam mini helicopter
[02:40:46] <les> many have video on the site
[02:41:08] <Jacky^> got any url ?
[02:41:16] <les> let me check
[02:41:21] <Jacky^> thanks
[02:41:26] <Jacky^> great
[02:41:50] <Jacky^> he work for civil protection forces
[02:42:11] <Jacky^> i dont' know the right term in english ..
[02:42:59] <Jacky^> he looks the fire in the forest ..
[02:44:10] <les> I understand
[02:44:13] <les> http://www.rc-cam.com/rc-cam3a.htm
[02:45:20] <Jacky^> les: yes,
[02:45:23] <Jacky^> thanks
[02:46:06] <les> http://www.google.com/search?q=rc+helicopter+on+board+video&hl=en&lr=&start=10&sa=N
[02:46:11] <les> yw
[02:47:04] <Jacky^> hehe, great :) thank you
[02:48:27] <Jacky^> http://www.protezionecivile.it/gallery/index.php?&id=305
[02:48:38] <Jacky^> here's where hes working
[02:49:07] <Jymmm> les : I sent it did you get it?
[02:49:39] <les> looking
[02:50:05] <les> yeah...looking
[02:50:12] <Jymmm> ah, ok.
[02:50:31] <Jymmm> I was messing with you a bit... how about now? and now? any yet? etc
[02:50:34] <Jymmm> oh, hi Jacky^
[02:50:51] <Jymmm> brb coffee
[02:50:58] <Jacky^> hi Jymmm ;)
[02:51:04] <Jacky^> i'm going to bed ..
[02:51:38] <Jacky^> g night
[02:51:44] <Jacky^> night les
[02:51:48] <Jacky^> thanks
[02:52:05] <Jacky^> Jacky^ is now known as Jacky^afk
[02:53:37] <les> ok...good pics
[02:53:54] <les> there are some deflection artifacts
[02:54:18] <les> softer ring porous areas cut deeper
[02:54:27] <les> two passes help that
[02:54:38] <les> actually they look pretty nice
[02:54:52] <les> as for the fuzz...
[02:55:01] <les> normal on oak.
[02:55:32] <les> get thee one of those abrasive filled nylon brush wheels
[02:55:40] <les> you know?
[03:02:19] <Jymmm> It was two passes... I can send you the gcode if you like.
[03:02:32] <Jymmm> thanks on the pics... I love my camera
[03:02:52] <les> I can see ringing in the upper right transition from line to arc
[03:03:03] <Jymmm> "ringing" ?
[03:03:13] <les> tell me the size and I can calculate the first resonant point
[03:03:38] <Jymmm> 1/2" 90 deg vcutter.
[03:03:46] <les> yeah...upper left of the plaque...two little wavies after the arc starts
[03:03:57] <les> the size of the sign
[03:03:58] <Jymmm> which pic?
[03:04:03] <les> #2
[03:06:38] <les> as far as the deflection...try to cut as deep as you can. Notice the deeper cuts have less waviness
[03:06:56] <Jymmm> ok, the width of the grove all the way around it is .128"
[03:07:33] <les> and the sign overall size?
[03:08:54] <Jymmm> the height of the final product from flat-to-flat is 2.830"
[03:09:14] <les> ok let me calculate
[03:09:41] <Jymmm> (excluding the birds of course)
[03:10:47] <les> first resonance is at 10 hz
[03:13:35] <les> all in all...not bad. You should not be too unhappy with that
[03:15:07] <Jymmm> I'm REALLY unhappy with the birds, they turned out horrible. As far as the waves on the left vertical side, I can live with it but I'd like to eliminate them in the future if possible.
[03:16:36] <Jymmm> I'm not sure of that means a different medium, some settings, or a different cutter.
[03:16:38] <les> deeper cuts...sharp bits...a second pass at a different (lower) speed and feed
[03:17:07] <Jymmm> ok, so basically not enough elbow room.
[03:17:27] <Jymmm> I'll just avoid oak for shallow work in the future.
[03:17:28] <les> is horizontal the x axis?
[03:17:34] <Jymmm> yes
[03:17:42] <les> ok
[03:18:02] <les> the resonance is a coupled y and twist
[03:18:32] <Jymmm> say that again, this time in an EARTH language if you please.
[03:18:45] <les> uh
[03:18:47] <les> heh
[03:19:34] <les> the router vibrates a little in y and z simultaneously
[03:19:35] <Jymmm> are you talking about where it is cutout in the top-right?
[03:19:45] <les> yeah
[03:20:00] <les> it seems the cut was clockwise
[03:20:01] <Jymmm> not the grove, but the full penetration...
[03:20:12] <Jymmm> that was done with a 1/8" downcut bit
[03:20:28] <les> ok
[03:20:28] <Jymmm> stepdown was .2
[03:21:07] <Jymmm> do you think the results would have been sharper/crisper with a 60 deg?
[03:21:28] <les> not much
[03:21:37] <les> play with speeds and feeds
[03:22:19] <Jymmm> well it's about 20K RPM now, I can still go up to 25K RPM
[03:22:22] <les> the birds would have been better with 60
[03:22:38] <Jymmm> oh, good to know.
[03:22:52] <Jymmm> HEy what is this you speak of?? abrasive filled nylon brush wheels
[03:23:07] <Jymmm> I've been using an old toothbrush so far.
[03:23:47] <les> yeah...in the hardware store. like fat toothbrush bristles but grit in the plastic.
[03:23:52] <les> put in drill
[03:24:04] <les> good for taking off the fuzz
[03:24:16] <les> quicker than scotchbrite
[03:24:16] <Jymmm> btw... I PU that whole board before milling it. It really brings out the color/grain I think
[03:24:30] <les> I saw
[03:25:04] <Jymmm> I never heard of these brushes
[03:25:22] <les> they are ok
[03:25:22] <Jymmm> what are they normally used for?
[03:25:41] <les> I have an invention for better ones
[03:26:01] <les> they get down in grooves and lightly sand them
[03:26:15] <Jymmm> sounds good.
[03:26:23] <les> they are
[03:26:50] <les> just enough totake off the fuzz
[03:27:00] <Jymmm> yeah, not scratcht eh finish
[03:27:06] <les> right
[03:27:35] <Jymmm> those brush wheels you mentioned.... what are they normally used for?
[03:27:46] <Jymmm> painting? automotive?
[03:27:55] <les> sanding contours!
[03:28:14] <les> they get down in grooves and such
[03:28:25] <Jymmm> I have never even heard of them.
[03:29:26] <Jymmm> OH these? http://www.spiralbrushes.com/abrasive-nylon.html
[03:30:47] <les> yeah!
[03:31:39] <Jymmm> Ah, ok. I used to use those in the locksmith shop for deburring keys
[03:32:12] <Jymmm> lot better on the fingers than the brass wire brushes =)
[03:32:19] <les> right
[03:32:37] <les> and they are fine cut...won't soften the carving
[03:32:44] <les> just take off the fuzz
[03:33:22] <Jymmm> Ok, will check em out to see whats available around here. In the meantime the toothbrussh seems to be doing the job (most of the time).
[03:33:45] <les> toothbrush and sand might work
[03:34:12] <Jymmm> baking soda
[03:34:25] <Jymmm> corn starch
[03:34:28] <les> I sandblast with that some
[03:34:50] <Jymmm> Hmmmm.... salt + toothbrush?
[03:35:07] <Jymmm> I should say table salt
[03:35:16] <les> salt is soft. Just dry loose sand
[03:35:29] <les> crushed silica sand...not river sand
[03:35:39] <Jymmm> oh, ok. heh
[03:35:47] <Jymmm> I'm thinking playsand
[03:36:17] <les> it's usually river sand...the grains are rounded
[03:36:30] <les> crushed sand has sharp edges
[03:36:50] <les> used for sandblasting
[03:36:58] <Jymmm> ah, ok. I didnt know that
[03:37:17] <les> you can find it loccally i'm sure
[03:37:30] <Jymmm> not in SILICON valley =)
[03:37:35] <les> haha
[03:38:30] <Jymmm> I posted a msg on the xylotex board about skipped steps.
[03:38:51] <Jymmm> He said I cna turn on the voltage from 24V to 28V on the PS
[03:39:01] <Jymmm> so well see if that helps any.
[03:39:18] <les> you need to check by hand to see if the ballscrew is physically binding
[03:39:25] <Jymmm> but to also check the temp on the motors next time.
[03:39:26] <les> I think it is not
[03:39:38] <Jymmm> jsut a sec, I'll get the post link
[03:39:43] <les> I think it is mid band resonance
[03:39:52] <les> k
[03:41:40] <Jymmm> les : Here's my original post, also look at the threads at the bototm of it too ----> http://groups.yahoo.com/group/Xylotex/message/4083?threaded=1
[03:41:46] <les> k
[03:47:13] <les> torque seems plenty.
[03:48:05] <Jymmm> Yeah. like I mentioned... I couldn't even jog 2" without stalling. two hours later - no problem.
[03:48:13] <les> at least sall torque
[03:48:20] <les> try slowing down
[03:48:48] <Jymmm> I could do that, but why didn't that occure during the first 2 or 3 hours?
[03:49:05] <Jymmm> that my question
[03:49:24] <les> everything was clean?
[03:49:27] <Jymmm> if it was goign the skip, it should have done it much earlier.
[03:49:30] <Jymmm> yes.
[03:49:55] <Jymmm> well, a lil dust. nothing like three hours later =)
[03:50:20] <les> well, stepper torque goes to hell at speed
[03:50:27] <Jymmm> I vacuum up every night.
[03:50:33] <les> just a fact of life
[03:51:03] <Jymmm> do you think going from 24VDC to 28VDC will make a difference?
[03:51:23] <les> runing constant current (or high voltage with balast resistors) helps
[03:51:43] <Jymmm> I can crank up the Vref (current) a little too
[03:52:12] <les> if you slow feed a little...slow spindle speed too
[03:52:16] <les> se what happens
[03:52:30] <Jymmm> ok
[03:52:58] <Jymmm> 80IPM be slow enough?
[03:53:07] <les> yeah try that
[03:53:10] <Jymmm> k
[03:53:19] <Jymmm> gonna take 5 hours now =)
[03:53:25] <les> and 16-18krpm
[03:53:36] <Jymmm> k
[03:53:42] <les> well
[03:53:49] <les> off to bed for me
[03:53:51] <Jymmm> fuck it, I bought two blanks this time =)
[03:53:54] <les> no music room
[03:53:58] <les> kinda tired
[03:54:05] <Jymmm> music room == guest rooom?
[03:54:28] <les> heh well dad is watching tv in there
[03:54:36] <Jymmm> ah =)
[03:54:51] <Jymmm> Well, have a good night and thanks for looking over things =)
[03:54:58] <les> yw
[03:55:00] <les> later
[03:55:07] <Jymmm> =)
[05:27:59] <CIA-6> 03jmkasunich * 10emc2/src/hal/utils/ (scope_files.c scope_trig.c): some partially completed work on halscope config save/restore
[06:13:09] <CIA-6> 03paul_c * 10emc2-auto/wiki/ (8 files in 7 dirs): "Auto update wiki from a cron job. Thu Oct 20 05:30:01 BST 2005 "
[07:26:08] <jmkasunich> hello
[07:26:17] <grichmond> grichmond is now known as richo
[07:26:57] <richo> hi john
[07:27:00] <jmkasunich> hi
[07:27:23] <jmkasunich> * jmkasunich is sitting here staring at emctask.cc and friends
[07:27:58] <richo> how's eveything going? i noticed that you've been busy checking in code... i just entered another bug :)
[07:28:08] <jmkasunich> I noticed :-(
[07:28:41] <richo> sorry, i'd like to take a look at it myself if i can, but i need to get my debugger up and working first...
[07:28:57] <jmkasunich> the most recent commit is actually stuff that is some weeks old - got tired of it showing up every time I did a CVS diff
[07:28:58] <richo> been a week of getting my head around the code!
[07:29:17] <jmkasunich> not an easy thing to do
[07:30:00] <richo> definitely... can i ask what you're using for debugging your code? i've only used the gcc/gdb tools in a microcontroller environment previously...
[07:30:27] <jmkasunich> I've been trying to grok the code that handles state transitions between estopped, ready, and running (the code uses much more ambiguous names)
[07:30:55] <jmkasunich> I don't know how to use gdb
[07:31:37] <jmkasunich> I usually debug by a combination of printfs, inspection, halscope and halmeter, and other things, depending on what code I'm looking at
[07:32:50] <richo> hmm, ok that's fair enough... i can probably start with the printf statements at least :)
[07:33:05] <richo> i'll have to look at gdb in my spare time...
[07:33:30] <jmkasunich> I should learn gdb, it is very powerfull (for a certain class of problems)
[07:33:38] <jmkasunich> not so good for realtime code
[07:35:06] <richo> John, I'm hoping to add some custom M codes to the interpreter... things like writing to IO at a given address... is this something that I should be putting into the main G-code interpreter or is there somewhere else that I should be handling this sort of thing?
[07:35:29] <jmkasunich> there are already mechanisms for custom M codes
[07:35:40] <jmkasunich> not sure of the details, I don't work on the interp much
[07:36:18] <richo> ok... who should I be talking to on that one? Alex Joni perhaps?
[07:36:25] <jmkasunich> but I think there is even a hook that lets you run an arbitrary program when the interp encounters a specific M code
[07:36:40] <jmkasunich> post on the developers list
[07:37:10] <jmkasunich> the interp is some of the most mature code, nobody works on it regularly
[07:37:18] <jmkasunich> some folks know it well though
[07:37:41] <richo> hmm, ok... i'll look into that one... definitely... i posted a more general question on the users list, but the developers list is probably the better bet...
[07:37:51] <richo> ok, thanks..
[07:38:50] <richo> btw, who started the emc project? just wondering if they're still involved or not :)
[07:39:06] <jmkasunich> initial authors were several folks at NIST
[07:39:15] <jmkasunich> Tom Kramer wrote most of the interp
[07:39:50] <jmkasunich> Will Shackleford and Fred Proctor did RCSLIB
[07:40:01] <jmkasunich> Fred wrote much of the rest
[07:40:13] <jmkasunich> I'm probably missing some parts
[07:40:19] <jmkasunich> it dates from the mid-90s
[07:40:49] <jmkasunich> All three of them have more-or-less moved on to other projects, Fred is the most active of the three
[07:41:13] <jmkasunich> but for the past several years 95% of the work has been by non-NIST folks
[07:42:12] <richo> ok, well nicely shared around...
[07:43:41] <richo> richo is now known as richo_away
[08:16:17] <anonimasu> morning
[08:44:50] <jmkasunich> I'm afraid so
[08:47:23] <ValarQ> yeah, it sure looks like a morning
[10:57:54] <CIA-6> 03jmkasunich * 10emc2/src/ (9 files in 6 dirs): changes to support HALified ESTOP logic - estop can be routed thru classic ladder or other logic, optionally latched to permit the use of momentary estop switches, etc.
[11:21:16] <alex_joni> hello
[11:42:49] <jmkasunich> jmkasunich is now known as jmk_sleep
[12:38:50] <ValarQ> alex_joni: hiya
[13:08:50] <alex_joni> ValarQ: hello
[13:11:55] <Jymmm> alex_joni: bye!
[13:12:06] <Jymmm> * Jymmm chuckles*
[13:16:06] <les> morning
[13:18:25] <alex_joni> alex_joni has kicked Jymmm from #emc
[13:18:41] <alex_joni> les: hi
[13:24:51] <jepler_> argh
[13:25:22] <alex_joni> huh.. what's wrong?
[13:25:27] <alex_joni> wrong button?
[13:25:28] <alex_joni> :)
[13:25:48] <jepler_> oh, part of the problem is that I don't know how to use my IRC client, and part of the problem is that my DSL is unreliable when it's raining.
[13:25:53] <jepler_> Jymmm: wtf?
[13:26:05] <Jymmm> jepler he's playing games
[13:26:31] <Jymmm> mornin les
[13:26:33] <Jacky^afk> Jacky^afk is now known as Jacky^
[13:26:55] <Jacky^> morning
[13:27:28] <Jymmm> jepler you might look at KVirc
[13:27:57] <Jymmm> jepler It's very mirc like, and easy to use
[13:31:19] <jepler_> you can have my text-mode IRC client when you pry it from my cold, dead, un-themed hands
[13:32:22] <jepler_> jepler_ is now known as jepler
[13:32:26] <Jymmm> alex_joni: payback is a bitch
[13:32:32] <alex_joni> :P
[13:33:18] <Jymmm> alex_joni ya damn OpNazi! lol
[13:33:27] <alex_joni> ditto
[13:33:32] <alex_joni> :)
[13:33:45] <Jymmm> Hey, I didn't start this, but I WILL finish it =)
[13:34:42] <alex_joni> right.. so how's your carving going?
[13:34:50] <alex_joni> or engraving, or whatever
[13:35:33] <Jymmm> Eh, it's goin.... My v carving from yesterday turned out like crap.
[13:35:46] <ValarQ> crap is good
[13:35:58] <alex_joni> ValarQ: :P
[13:36:15] <Jymmm> Had les look at the toolmarks
[13:36:25] <ValarQ> hmm, opwar on #emc, that was something new...
[13:36:27] <Jymmm> some deflection goin on
[13:37:15] <Jymmm> and something about harmonics
[13:37:34] <alex_joni> your deflections aren't harmonic
[13:37:39] <Jymmm> les : what was all that about anyway?
[13:38:30] <Jymmm> alex_joni say what?
[13:57:20] <anonimasu> hello
[13:57:34] <Jymmm> howdy anonimasu
[13:57:40] <anonimasu> hows things?
[13:57:53] <anonimasu> * anonimasu will go with �huydralics for the spindle..
[13:57:54] <Jymmm> it's 5am... can't be good =)
[13:58:14] <anonimasu> but I've got a shitload of stuff to re-machine..
[13:58:39] <anonimasu> the morons that designed my mill didnt think about where they placed the spindle seals..
[13:58:56] <Jymmm> oooops
[13:59:20] <alex_joni> Jymmm: not really 5am over there
[13:59:33] <alex_joni> more like 1400 hours
[13:59:44] <anonimasu> but huydralics is a nice way to reduce the weight on the column..
[13:59:46] <alex_joni> or 1300 :)
[13:59:56] <Jymmm> all I know is it's really dark outside
[14:00:14] <alex_joni> Jymmm: go to sleep
[14:00:36] <Jymmm> alex_joni I woke up at 0340, might go back to bed in a few.
[14:03:45] <anonimasu> *yawns*
[14:11:32] <alex_joni> Jymmm: nice
[14:12:38] <Jymmm> alex_joni I hope that was sarcaism
[14:13:23] <alex_joni> depends on why you woke up :)
[14:13:55] <Jymmm> I havne't got a full nights sleep in weeks (months?)
[14:14:10] <Jymmm> I just wake up fully awake
[14:15:04] <Jymmm> ok, back to bed.... cya!
[14:15:04] <anonimasu> Jymmm: got a job or are you unemployed?'
[14:15:06] <alex_joni> seen a doc?
[14:15:23] <anonimasu> if you are unemployed and cant sleep maybe you should try doing something that
[14:15:30] <anonimasu> 'll make you tired ;)
[14:15:52] <Jymmm> I work for myself, but most of my stuff is at the computer
[14:16:19] <Jymmm> anyhow... g'night (2nd time today) =)
[14:16:32] <anonimasu> night
[15:25:05] <alex_joni> cradek: around?
[16:18:17] <Jacky^> Jacky^ is now known as Jacky^afk
[16:25:09] <alex_joni> * alex_joni goes home
[16:25:11] <alex_joni> later guys
[16:30:00] <Jacky^afk> Jacky^afk is now known as Jacky^
[16:43:05] <cradek> alpha: yes
[16:43:09] <cradek> argh
[17:16:20] <Jacky^> anyone know if is possibile to drive 5 phase stepper motor using emc2 and L298 ic ?
[17:17:20] <cradek> L298 is just two H bridges
[17:17:31] <Jacky^> hi cradek
[17:17:34] <cradek> hi
[17:17:44] <Jacky^> looking for 5 phase driver ..
[17:18:10] <Jacky^> seems no schematics available on the net
[17:18:23] <cradek> I think those steppers are rare
[17:18:30] <Jacky^> yeah ..
[17:18:47] <Jacky^> ive one i bought year ago ..
[17:19:07] <Jacky^> it should have a great precision
[17:19:23] <Jacky^> but its hard to find any driver..
[17:20:49] <jepler> all you need is 2.5 L298s and a microcontroller to drive the windings in the right sequence
[17:21:37] <jepler> (but I'm just guessing here)
[17:22:34] <cradek> I think they have to be unipolar with 5 windings, so the drivers do not need to be H bridges
[17:22:45] <cradek> so buying L298 is a waste
[17:23:30] <Jacky^> i friend want to build a machine like this: http://www.carlsondesign.com/
[17:23:36] <Jacky^> its a plotter ..
[17:23:56] <Jacky^> i think maybe servos could be used instead of 5 phase stepper..
[17:24:06] <jepler> http://www.geocities.com/dsaproject/electronics/cnc/cnc_ctrl.html
[17:25:06] <Jacky^> jepler: nice link, thanks
[17:27:54] <jepler> Jacky^: np
[17:45:16] <alex_joni> hi guys
[17:47:23] <cradek> hi
[17:56:03] <alex_joni> hi chris.. wanted to ask you smthg, but figured it out eventually :P
[17:59:50] <alex_joni> hi rcsu
[18:06:01] <rayh> Just booted a puppy-emc. Nice.
[18:06:29] <alex_joni> really?
[18:06:33] <alex_joni> cool
[18:15:42] <rcsu> hi alex_joni
[18:15:52] <alex_joni> what's up?
[18:16:02] <alex_joni> rayh: that's puppy-emc2
[18:16:56] <rayh> I see that. Like it a lot so far.
[18:17:11] <alex_joni> like the ini chooser?
[18:17:20] <rayh> I see a 2.4 kernel
[18:17:27] <rcsu> alex_joni: nothing, emc2 runs (at least in the simulator), ill connect the hardware now
[18:17:36] <cradek> yay for 2.4 kernels
[18:17:41] <alex_joni> rcsu: and that you call nothing?
[18:17:48] <alex_joni> * alex_joni cheers with cradek
[18:18:03] <cradek> is it 2.4.31 built with SATA support?
[18:18:10] <rcsu> alex_joni: yes, nothing really spannendes
[18:18:29] <alex_joni> * alex_joni wonders how many got that
[18:21:59] <rayh> I'm a bit out of the puppy-emc2 loop.
[18:22:19] <rayh> What is planned as far as updates/upgrades?
[18:41:16] <alex_joni> well.. we talked to cncuser yesterday
[18:41:25] <alex_joni> and he said he's doing a developer version
[18:41:30] <alex_joni> with a CVS emc2 in there
[18:41:46] <alex_joni> so basicly you only need to boot it, do a cvs up, and compile / run
[18:42:17] <alex_joni> I set up a site with automatic tar.gz of emc2 source code.. so that might aswell get included
[18:42:19] <rayh> Oh. That is nice.
[18:42:39] <alex_joni> will talk to cncuser, maybe I should include binaries aswell
[18:43:10] <alex_joni> but binaries are a pita, they only work on the same kernel/rtai combo
[18:43:15] <rayh> couple issues. shutdown didn't seem to unmount my hd so reboot required a full check of partitions.
[18:43:28] <alex_joni> hmm.. did it mount your hd?
[18:43:45] <alex_joni> or you did that on your own?
[18:43:53] <rayh> You'd have to make certain that you named the bin distro according to the puppy version.
[18:43:59] <rayh> It did the mount.
[18:44:08] <alex_joni> I see..
[18:44:17] <rayh> That seemed a bit different from knoppix.
[18:44:44] <rayh> it connected to net here with dhclient
[18:44:58] <alex_joni> nice
[18:45:16] <rayh> no web browser. but I'm guessing that the file browser can handle urls
[18:45:21] <alex_joni> but bear in mind it's an early beta, lots of things need changing..
[18:45:32] <alex_joni> rayh: wget should be on it.. I think
[18:45:36] <rayh> I thought it was great. I'll show it around.
[18:45:57] <rayh> can't test now rebooted to 4.30
[18:46:09] <rayh> The test was on an athalon 64.
[18:46:14] <alex_joni> coo
[18:46:20] <rayh> Great stuff for testing.
[18:46:22] <alex_joni> I liked the puppy for beeing small
[18:46:43] <alex_joni> I imagine running it from an USB stick ;)
[18:46:51] <alex_joni> just plug it in, boot from it :)
[19:07:34] <anonimasu> hello
[19:08:17] <anonimasu> alex_joni: what was so special about that url?
[19:08:18] <anonimasu> :D
[19:11:40] <Jymmm> Sears now sells a powder coater
[19:12:23] <anonimasu> hm
[19:12:28] <anonimasu> it looks like my mill will be down for amonth
[19:12:29] <anonimasu> :/
[19:12:39] <anonimasu> I got the bearings but I am going to convert the Z to a ballscret..
[19:12:42] <anonimasu> rew
[19:12:53] <Jymmm> replacment parts not available?
[19:13:31] <anonimasu> what?
[19:13:46] <Jymmm> a month to wait for parts?
[19:13:57] <anonimasu> more like a month of work..
[19:14:34] <Jymmm> ah
[19:18:18] <Jymmm> craftsman.com and search for 917288
[19:20:15] <Jymmm> 400F for 20 minutes
[19:23:52] <alex_joni> hey Jymmm
[19:23:55] <alex_joni> up already?
[19:33:55] <Jymmm> alex_joni yeah
[19:34:03] <Jymmm> sorta kinda =)
[20:00:19] <jmk_sleep> jmk_sleep is now known as jmkasunich
[20:00:56] <Jacky^> Jacky^ is now known as Jacky^afk
[20:01:07] <cradek> was there a url for the puppy distrib?
[20:04:18] <jepler> cradek: there was but it seems to have scrolled off my irc
[20:04:39] <jepler> cradek: I have a copy at home
[20:05:23] <alex_joni> you can find it on my mirror
[20:05:32] <alex_joni> http://dsplabs.cs.utt.ro/~juve/emc/
[20:06:37] <cradek> thanks
[20:06:53] <alex_joni> cradek: anytime
[20:07:18] <jepler> that link is a bit of a trick to find...
[20:07:57] <alex_joni> jepler: it's not for everyone :D
[20:08:11] <alex_joni> but point taken..
[20:12:04] <alex_joni> jepler: better like this?
[20:16:05] <jepler> alex_joni: yeah, that is easier to spot
[20:16:13] <alex_joni> ok
[20:16:35] <alex_joni> * alex_joni is happy to see lots of downloads there
[20:20:20] <fenn> lots of cheap igbt's on ebay
[20:21:07] <Jacky^afk> Jacky^afk is now known as Jacky^
[20:22:47] <fenn> are igbt's suitable for use in a pwm servo amp?
[20:23:35] <alex_joni> fenn: depends what kind
[20:23:55] <alex_joni> I did a DC pwm servo amp with some
[20:27:02] <icee> IGBTs are preferred for high power motor driving
[20:27:07] <icee> but yah, it depends on the IGBTs involved
[20:28:32] <fenn> what specifically are you looking for in a motor amp?
[20:28:43] <icee> me?
[20:28:54] <fenn> i mean, why are some good and others bad?
[20:29:08] <icee> well, it depends on what kind of amplifier you're trying to build
[20:29:18] <icee> e.g. high switching frequency usually means big gate charges are bad
[20:29:47] <fenn> cant you just look at the rated switching speed?
[20:29:58] <icee> the quality of the body diode matters more in a motor application, as do the magnitudes of the switching currents
[20:30:02] <icee> No, you can't
[20:30:22] <icee> because whatever you're driving it with needs to be moving that gate charge back and forth at the switching frequency
[20:32:32] <icee> there's a lot that goes into designing a nice big amplifier
[20:33:18] <icee> This is the best part I've found for my app so far:
[20:33:19] <icee> http://www.irf.com/product-info/datasheets/data/gb25xf120k.pdf
[20:33:37] <icee> reasonable (though still high) gate charge, good switching speed.. 'ultrasoft reverse recovery characteristics' (only 55A!)
[20:34:28] <fenn> what does "ultrasoft reverse recovery" mean?
[20:34:30] <icee> the transfer characteristics also matter, .. though if you're using a high driving voltage not so much
[20:34:58] <icee> fenn: LOTS of current can flow when the body diodes recover from being reverse biased
[20:35:43] <icee> this part is nice in that the peak is 'only' 55A into a 400uH load that you'd been forward driving at 25A
[20:36:13] <icee> those reverse recovery currents are the number one thing you need to design for
[20:36:19] <icee> huge amounts of noise, ground bounce, etc come from them
[20:37:31] <etla> hi
[20:37:44] <icee> i'm thinking of trying to switch at 100KHz or so.. but I haven't done the math on the dissipation from moving the gate charge back and forth
[20:37:51] <icee> might exceed the characteristics of the driver part I've chosen
[20:38:11] <etla> what are you building ?
[20:38:27] <icee> etla: a 1kW/axis servo amplifier and driver
[20:38:46] <icee> AC servo, that is.
[20:38:49] <etla> for sanyo denki/surpluscenter bldc motors ? :)
[20:38:53] <fenn> heh
[20:38:59] <icee> etla: yes.
[20:39:00] <etla> stock=0 today :(
[20:39:05] <fenn> that was quick
[20:39:09] <etla> you bought them all ?
[20:39:26] <icee> i bought 5
[20:39:41] <etla> did you get them already ?
[20:39:45] <icee> nope
[20:39:53] <icee> the tracking number they sent me is invalid, too :/
[20:40:02] <etla> so your amp takes in +/-10V from a servo card ?
[20:40:09] <icee> nope
[20:40:18] <etla> how then ?
[20:40:23] <icee> I'm going to speak the pico systems parallel protocol on the driver board
[20:40:29] <icee> it'll send PWM to the amplifier
[20:40:50] <icee> and the amplifier will feed back current levels and the driver will run a current control loop
[20:41:28] <etla> hmm.. I dont really follow, can you describe the chain from pc to motor ?
[20:41:35] <fenn> question: are you an electrical engineer?
[20:41:58] <etla> I have a msc in physics, done some instrumentation electronics - not an expert
[20:42:05] <jepler> I'm sure not
[20:42:16] <icee> I'm not a EE
[20:42:25] <icee> but I've done board level and gate level design professionally
[20:42:36] <icee> and have done a certain amount of power electronics
[20:42:43] <icee> anyways, the interface is as follows:
[20:42:55] <icee> pc -> my driver board, over parallel port, speaking IEEE 1284 protocol
[20:43:06] <icee> PC sends current levels to driver boards, and reads back 24 bit encoder counts back.
[20:43:26] <icee> My driver board is hooked to the encoder, and maintains the encoder counts (for the PC, and so it knows the phase angle to drive)
[20:44:19] <icee> it receives a current level ADC readback from the amplifier board, and runs a simple PID control loop on current
[20:44:26] <icee> output is PWM to the amplifier board.
[20:45:08] <etla> so the position feedback loop is controlled by the pc ?
[20:45:10] <icee> there's also some digital back from the amplifier board for fault monitoring from the IR driver chip (IR2131, i think)
[20:45:12] <icee> etla: yah
[20:45:28] <icee> position/velocity/etc is all on the PC
[20:45:39] <etla> is commutation done with the same encoder as position feedback ?
[20:45:54] <icee> etla: yes. I don't believe the sanyo servos have halls on them.
[20:45:58] <icee> I will know for sure when I get them in hand.
[20:46:15] <icee> The data sheet I have for the 200vac version of them seems to say not.
[20:46:26] <etla> so you need to generate 3 sinewaves which are a suitable amount phase shifted ?
[20:47:03] <icee> ywa
[20:47:30] <icee> duty cycle of the output will be the current control loop output * cos(phase angle)
[20:48:09] <etla> are there ready asics for this for driving a H-bridge ?
[20:48:11] <icee> I may crudely estimate velocity on the board in order to determine how much to 'lead' the position of the motor by
[20:48:26] <icee> etla: IR2131 is a nice part for driving a 3 phase drive
[20:48:46] <icee> DSPIC/my own software for the actual PWM generation
[20:49:07] <etla> ecoder counting ?
[20:49:14] <icee> DSPIC has a quadrature decoder
[20:49:23] <icee> with filters
[20:49:30] <etla> so you have three of them
[20:49:34] <icee> four, actually.
[20:49:52] <icee> Yah. It's either that or a CPLD, and I think it's probably less hassle to just use a microcontroller per axis
[20:50:09] <etla> how does the 24bit count go over the parport to the pc ?
[20:50:15] <etla> you need multiple parports ?
[20:50:20] <icee> etla: nope
[20:50:30] <icee> EPP
[20:50:51] <icee> the parallel port logic on the PC generates an address cycle and appropriate interlocking strobes
[20:51:03] <icee> It's capable of doing 1.5M 8 bit transactions per second
[20:51:11] <icee> supposedly the pico guys are getting 10KHz servo updates out of it.
[20:51:15] <jepler> http://www.beyondlogic.org/epp/epp.htm
[20:51:24] <icee> I have a better link, one second
[20:51:37] <icee> http://www.fapo.com/eppmode.htm
[20:51:45] <jepler> thanks
[20:51:48] <jepler> I'll take a look at that
[20:52:05] <icee> anyways, it's really easy to interface with
[20:52:21] <icee> two interrupt lines on the strobes
[20:52:35] <icee> (triggered both edges)
[20:53:20] <icee> and it's nicely interlocked as a protocol so there's no real timing concerns
[20:53:48] <etla> you have this pico board http://pico-systems.com/PPMC.html ?
[20:53:52] <icee> nope.
[20:54:01] <icee> the DSpics on my board will speak the pico protocol.
[20:54:07] <icee> I'm not using their hardware.
[20:54:22] <icee> It's nice in that EMC1 already speaks their protocol, and EMC2 will soon; so no integration work for me on the PC side.
[20:54:41] <etla> and the signla to the amp is a pwm=velocity and a logic level for direction ?
[20:54:51] <etla> yes, that sounds good
[20:55:06] <icee> the signal to the amplifier itself is a PWM duty cycle.
[20:55:14] <icee> the signal from pc -> my driver board is a current level-- e.g. torque value
[20:55:48] <etla> so some duty cycle corresponds to stop and less=cw more=ccw ?
[20:56:16] <icee> well, etla, my output is 3 phases of PWM
[20:56:43] <icee> so the duty cycles of those, compared to the angle of the motor, determine whether CW or CCW force is expressed and how much.
[20:57:17] <icee> from computer -> driver board, it's just a number. negative == CCW effort, positive == CW effort
[20:57:32] <icee> (though there may be some offset error, the integrator term in the PC control loop will take care of that)
[20:57:48] <etla> and the pwm goes to 'dumb' voltage2current amps that drive the servo windings ?
[20:58:21] <icee> a dumb power stage and IGBT, yes. current level gets fed back to the driver board
[20:58:29] <icee> which measures it and adjusts duty cycle to get the right current level
[20:58:40] <alex_joni> * alex_joni is dizzy :)
[20:58:43] <alex_joni> you guys talk too much
[20:58:47] <icee> maybe I should draw this :P
[20:58:48] <Jymmm> lol
[20:59:06] <Jymmm> you guys draw too much!
[20:59:11] <etla> are you planning on making your design opensource ?
[20:59:20] <icee> etla: I may
[20:59:37] <icee> I don't really want to have to support it.
[20:59:53] <fenn> heh does open source mean more or less support?
[21:00:08] <Jymmm> icee do like all the commercial stuff does... strt a yahoo group
[21:00:12] <icee> fenn: releasing stuff as open source means lots of people email you and beg you for help ;)
[21:00:27] <icee> I'm thinking I'll release most of it. I may not supply gerbers.
[21:00:41] <icee> i'm also a little concerned about liability if someone burns down their house
[21:00:49] <icee> or kills themselves with 340VDC
[21:01:16] <Jymmm> icee just a standard disclaimer
[21:01:34] <Jymmm> "If you fuck up, it's you're own damn fault"
[21:02:09] <icee> anyways, the driver board is the interesting part. the amplifier itself.. most people will have different requirements and will build something else
[21:02:21] <icee> e.g. lawrenceg and i are collaborating on the driver board, but I think his power stage will look completely different from mine
[21:02:52] <icee> and the driver board is of less concern, .. i should be able to open it and its code up all the way
[21:03:29] <etla> so theres one dspic per axis ? how do these cooperate to share the one epp port ?
[21:03:42] <icee> etla: it's bussed between them.
[21:03:57] <icee> The IEE1284 protocol provides for selecting multiple parts on the same port
[21:04:00] <etla> you need an extra chip for that ?
[21:04:01] <fenn> where did you find the pico protocol? just looking at the emc code or what?
[21:04:07] <icee> fenn: yah
[21:04:18] <icee> nope, no extra chip.
[21:04:37] <icee> and the pico guys were nice enough to use the 'address' to select the individual registers they wanted
[21:04:48] <etla> so the controller is modular
[21:04:53] <icee> so it's easy for part 1 to respond to registers 1-10, part 2 to registers 11-20
[21:05:00] <alex_joni> lol.. like that
[21:05:05] <alex_joni> the "pico" protocol
[21:05:13] <alex_joni> maybe call it pipo ?
[21:05:21] <etla> nano
[21:05:28] <icee> well, perhaps. I'm planning on just putting 4 dspics on one board
[21:05:31] <alex_joni> pimpo ?
[21:05:32] <fenn> "PP PP"
[21:05:42] <icee> and sharing power supply, etc, and bussing all the parallel lines
[21:05:46] <etla> OpenPico
[21:06:06] <icee> if other people build it, they don't need to populate the components for the 4th axis or whatever
[21:06:24] <Jymmm> call it zepto
[21:06:25] <fenn> if you made room for 6 you'd be my best friend
[21:06:36] <icee> fenn: you could just hook two of the boards together *shrug*
[21:06:51] <fenn> oh good point
[21:06:54] <icee> you'd just need to be a little careful with the parallel wiring as that's getting close to the limit of how many EPP devices you want to bus
[21:07:26] <fenn> as far as data rate?
[21:07:33] <icee> yah, as far as bus settling time, etc
[21:07:43] <etla> can you program a dspic with the same hardware as normal pics ?
[21:07:54] <icee> etla: I believe so.
[21:08:06] <icee> I haven't written code for dspic. so still research going on there
[21:08:16] <icee> I have gcc for dspic and binutils installed now
[21:08:57] <icee> there's homebrew programmers for dspic, i may just build one of those since microchip's programmers are expensive and I don't really do much pic stuff (mostly avr/arm)
[21:09:29] <icee> lawrenceg suggested dspic, and it's a good choice with the nice hardware quadrature decoder and the nice multiplier/fractional divider on the part
[21:10:04] <etla> generating the pwm and counting the encoder pulses don't interfere with eachother ?
[21:10:45] <icee> nope, there's a separate 'motor driver' timer/counter
[21:10:59] <etla> they run at 20 MHz ?
[21:11:16] <icee> 20MHz? the highest count rates i'll see are about a megahertz
[21:11:29] <etla> the clock of the dspic
[21:11:35] <icee> 120MHz
[21:11:43] <icee> that's an instruction clock of 30MHz
[21:11:44] <etla> ohh
[21:11:51] <icee> I hope to clock much slower than that.
[21:12:16] <icee> slower clock == higher reliability with noise coupled into the supply, etc.
[21:12:22] <anonimasu> hm, why not generate the pulses with a fpga?
[21:12:33] <icee> anonimasu: That's what I was inclined to do at first.
[21:12:41] <icee> But on second thought, -why- generate the pulses with a fpga?
[21:12:42] <anonimasu> it'd be a nice learning project..
[21:12:50] <icee> Oh, I've written plenty of HDL. *shrug*
[21:12:54] <anonimasu> boardspace..
[21:13:17] <icee> I don't care about board space.
[21:13:26] <icee> 4 DSPICs are small, and it makes things modular
[21:13:39] <fenn> but the little box next to the mill will be 10 cm shorter!
[21:13:41] <icee> easier design rework, more versatility/intelligence in my current control loop
[21:14:01] <icee> ;) i can probably build this board well under 20 sq in. compare to each of my driver boards, which will be 20 sq in per axis
[21:14:04] <icee> er
[21:14:06] <icee> amplifier boards
[21:14:46] <icee> even with 4 dspics, the clamping diodes on the analog inputs and the connectors will dominate the board area
[21:14:47] <fenn> 20 sq in sounds really big
[21:15:00] <anonimasu> yeah
[21:15:11] <icee> fenn: 3x6"?
[21:15:27] <etla> eurocard
[21:16:37] <etla> so for each phase you will have a separate current feedback ?
[21:16:47] <icee> etla: nah. one for each axis
[21:16:52] <anonimasu> * anonimasu looks at dsics
[21:16:57] <anonimasu> avr's seems like old crap..
[21:16:58] <anonimasu> :/
[21:17:10] <icee> anonimasu: Hm? they're very nice to work with
[21:17:28] <anonimasu> I know..
[21:17:35] <icee> very very reliable, simple to integrate with, nice toolchain
[21:17:53] <anonimasu> yep..
[21:17:55] <anonimasu> I program avr's..
[21:17:55] <icee> the dsPIC seems like crap compared to a tms320f28x, which I've used before
[21:18:10] <alex_joni> icee: anything containing PIC is crap
[21:18:11] <icee> but tms320 is multivoltage, vqfp-100/144/176, etc
[21:18:18] <alex_joni> although ValarQ might like it
[21:18:19] <icee> alex: agreed
[21:18:33] <anonimasu> hehe
[21:18:33] <icee> this is the right part for the job though. it has the peripherals.
[21:18:38] <alex_joni> the best project I did with a PIC was to underclock it
[21:18:44] <anonimasu> http://www.avrfreaks.net/index.php?module=FreaksDevices&func=displayDev&objectid=106
[21:18:44] <alex_joni> got it running at 50Hz
[21:18:47] <anonimasu> ^_^
[21:18:50] <anonimasu> YAY
[21:18:59] <anonimasu> I'd love a avr doing 50mhz
[21:19:08] <alex_joni> 50Hz not 50MHz
[21:19:13] <alex_joni> but it was 220V 50Hz
[21:19:18] <anonimasu> you could do some brutal realtime stuff..
[21:19:27] <icee> if I could, I'd use the at90pwm parts
[21:19:28] <alex_joni> and the PIC was so happy, it jumped right to the ceiling
[21:19:36] <anonimasu> LOL
[21:19:38] <icee> but they don't exist yet
[21:20:18] <anonimasu> but it's been several years since I played with avr's..
[21:20:26] <anonimasu> and now I cant seem to get my outputs to work as I want them..
[21:20:33] <alex_joni> it's nice to work with them
[21:20:35] <anonimasu> sbi/cbi is dead :D
[21:20:37] <anonimasu> dead/deprecated
[21:20:44] <anonimasu> _BV() is a horrid thing
[21:21:18] <anonimasu> atleast for modifying a single bit easily..
[21:21:28] <icee> _BV() is nice, actually
[21:21:46] <anonimasu> hm care to explain it to me?
[21:22:01] <icee> well, the constants themselves are the bit positions of the flags
[21:22:28] <icee> so if you have a variable, that's true or false, and you want to set the flag to that value
[21:22:30] <icee> you can do
[21:22:52] <icee> PORTB = leda << PB1 | ledb << PB4;
[21:23:17] <icee> which if leda is 1 it'll turn PB1 on, and if ledb is 1 it'll turn PB4 on
[21:23:36] <icee> _BV(PB1) is just (1 << (PB1))
[21:24:17] <alex_joni> icee: see you scared rcsu away
[21:24:35] <icee> ;)
[21:24:46] <anonimasu> hm still dosent make senese
[21:25:04] <icee> anoni: PB0 is 0, PB1 is 1, PB2 is 2
[21:25:10] <anonimasu> yeah I know that much
[21:25:24] <anonimasu> PORTB = _BV(PB0);
[21:25:29] <anonimasu> should turn the bit on.. right?
[21:25:32] <icee> _BV(PB0) is 1<<0, or 1; _BV(Pb1) is 2, _BV(PB2) is 4
[21:25:34] <icee> yah
[21:25:54] <icee> if you have a series of flags that you're passing to a function, though
[21:26:04] <anonimasu> but if I do, PORTB = _BV(PB0); PORTB = _BV(PB1);
[21:26:11] <icee> you need |= for that
[21:26:45] <icee> PORTB |= _BV(PB1); turns on PB1
[21:26:51] <icee> without affecting other bits
[21:27:04] <anonimasu> hm, I need to toy with it a bit later
[21:27:20] <icee> the compiler may compile that either as a sbi instruction
[21:27:27] <icee> or as a load/or/store
[21:27:33] <icee> depending on which works better in the circumstance.
[21:27:49] <anonimasu> yep
[21:27:51] <jepler> (sbi instructions aren't available for all I/O ports)
[21:27:58] <jepler> (iirc)
[21:28:02] <anonimasu> brb
[21:28:04] <anonimasu> need chips
[21:28:12] <icee> yah, sbi only works for the beginning of memory
[21:28:12] <cradek> * cradek gleefully notices that he doesn't know the avr instruction set
[21:28:21] <anonimasu> * anonimasu buys a bag of avr's at the kiosk
[21:28:28] <anonimasu> at90pwm seems crunchy
[21:29:19] <cradek> the first microcontrollers I haven't programmed in assembler
[21:29:33] <alex_joni> cradek: whose aren't avr instructions :P
[21:29:44] <alex_joni> it's the work of the devil expressed as #macros
[21:29:54] <cradek> oh I know that
[21:30:27] <cradek> I was wading through btfsc last night
[21:30:44] <cradek> I never did like PICs even though I used them because they were cheap
[21:30:50] <jepler> btfsc?
[21:31:06] <cradek> `test a bit in the accumulator and skip the next instruction if it's not set'
[21:31:10] <jepler> that's some PIC instruction?
[21:31:13] <icee> it's amazing microchip is the market leader even though everyone hates using their parts?
[21:31:17] <cradek> yeah
[21:31:22] <jepler> (avr has sbic which sounds about the same)
[21:31:29] <jepler> (or something like that)
[21:31:39] <cradek> I never could get comfortable with skip instead of jump
[21:31:57] <alex_joni> cradek: I so hate pic asm
[21:31:57] <jepler> something something the pipeline cs310 something
[21:32:05] <alex_joni> but mostly their stupid IDE
[21:32:11] <alex_joni> the doze one is horrid
[21:32:18] <icee> i am hoping that i can program the dspic in c.
[21:32:20] <cradek> alex_joni: I never used anything but vi and the assembler
[21:32:22] <alex_joni> I spent a few hours chasing a stupid error
[21:32:23] <icee> probably a futile hope though.
[21:32:34] <alex_joni> and the problem was using lines too long
[21:32:47] <cradek> icee: the avr is a better part and has great gcc support
[21:32:49] <alex_joni> and the compiler simply cut them, without a warning
[21:32:58] <icee> cradek: I know, I've used AVRs extensively
[21:33:08] <icee> no AVR has quadrature decoding, though
[21:33:30] <jepler> oh, dspic is some terrible combination of PIC and DSP?
[21:33:33] <cradek> interrupt on one line, and at the top of the handler, read the other line?
[21:33:35] <icee> and none has a really good 3 phase motor driver.. the at90pwm3 will, but.. it isn't shipping yet
[21:33:37] <icee> jepler: yes
[21:33:49] <icee> pic + a nice multiplier instruction, basically
[21:34:09] <icee> and big 40 bit accumulators and prefetch
[21:34:33] <jepler> FPSLIC (AVR with FPGA)
[21:34:44] <alex_joni> that sounds sooo much better :D
[21:34:55] <icee> jepler: if i were to do that, i'd just ditch the microcontroller entriely
[21:34:59] <icee> and use a FPGA
[21:35:07] <alex_joni> or an ARM
[21:35:11] <alex_joni> * alex_joni digs ARM
[21:35:14] <icee> cradek: at 1MHz?
[21:35:24] <icee> I like at91 and the lpc21xx parts
[21:35:28] <icee> still, neither is well equipped for this
[21:35:30] <jepler> http://www.atmel.com/dyn/resources/prod_documents/DOC3023.PDF
[21:36:01] <jepler> I've still never done FPGA
[21:36:07] <LawrenceG> icee, good morning
[21:36:29] <icee> lg: hey. So I've come around. dspic is OK, even though everyone here is trying to talk me out of it now ;)
[21:37:17] <LawrenceG> well... still looking here, but dspic is a good runner so far... samples are free... havent checked on real price
[21:37:25] <icee> it's like $10/unit
[21:37:35] <fenn> which dspic are you using?
[21:37:35] <icee> so anyways, I can't start designing until I havem otors in hand to check quadrature format, etc
[21:37:44] <icee> but I'm thinking the following for the board i'm designing:
[21:37:49] <icee> 5V linear regulator
[21:37:54] <icee> 4xDSPIC
[21:38:09] <icee> fast comparators for 12 channels of differential input
[21:38:16] <LawrenceG> dsPIC30f2010
[21:38:48] <icee> yah, that's the one
[21:39:05] <anonimasu> hm
[21:39:05] <anonimasu> iab
[21:39:07] <icee> and 2 analog inputs per axis (8) with termination resistors and clamping to both rails and zener clamping
[21:39:11] <LawrenceG> 512 ram, quad encoder i/f, 12k flash, 28 pin, 6 10bit adc and 3 phase pwm
[21:39:17] <icee> on a 4 layer board
[21:39:43] <anonimasu> that'll make it mighty expensive
[21:39:45] <icee> all i/o and power terminated with small screw terminal blocks
[21:39:56] <anonimasu> how much is a 4 layer board?
[21:40:08] <icee> anon: I have a piece that makes them for $15/apiece at qty 5
[21:40:12] <icee> place
[21:40:13] <anonimasu> ah
[21:40:15] <anonimasu> :)
[21:40:16] <icee> www.pcbfabexpress.com
[21:40:22] <anonimasu> that makes it easy
[21:40:37] <icee> they're not good to use for super fine pitch surface mount though, because they use cheap materials
[21:40:42] <icee> but I imagine this will be all through hole
[21:41:16] <icee> lg: oh yah, and parallel port interface-- 26 pin ribbon connector
[21:41:29] <LawrenceG> for the amp I am leaning towards 6 * irg4bc20kd driven by 3*ir2109
[21:41:48] <LawrenceG> as per an889
[21:41:55] <icee> why do you prefer the ir2109 over the ir2131?
[21:42:18] <icee> oh, that's another thing.. i'll have digital i/o for the fault monitoring and management capabilities of the 2131
[21:42:24] <LawrenceG> price is about 1/2
[21:42:45] <icee> * icee checks
[21:43:48] <icee> the 2109 only drives a single phase though, right?
[21:44:08] <LawrenceG> yes... need 3 /motor
[21:44:33] <etla> from where do you get the current feedback ?
[21:44:45] <icee> etla: resistor between IGBT and ground
[21:45:04] <etla> to adc on pic ?
[21:45:09] <icee> yah.
[21:45:15] <icee> after appropriate termination and clamping
[21:46:41] <etla> so each motor phase has a current from 0 to some Imax
[21:46:44] <LawrenceG> icee, you mentioned you have c compiler for dspic installed... which one?
[21:46:47] <etla> and not +/- Imax ?
[21:47:06] <icee> etla: each motor has current measured from 0 to some Imax
[21:47:12] <icee> like i said before, current sensing is per motor, not per phase.
[21:47:21] <icee> I am using 6pack iGBTs arranged in a polyphase bridge.
[21:47:31] <icee> lg: gcc4.0 variant, i believe
[21:47:47] <icee> i still need to get it on my main dev machine, it's on my mac now
[21:47:58] <LawrenceG> cool.... free?
[21:48:02] <icee> yup
[21:48:06] <icee> no clue if it works though
[21:48:14] <etla> thanks guys - I've learned a lot again - catch you later.
[21:48:22] <icee> cya etla.
[21:48:41] <LawrenceG> I gave up on using C with pics..... its just as fast for most low level projects to use asm
[21:49:19] <icee> lg: so what do you think of the general parameters of the board design? if that's fine with you, I can begin entry
[21:49:55] <cradek> LawrenceG: I totally disagree
[21:50:16] <LawrenceG> sounds good..... are you planning isolation on the pwm lines to the power boards?
[21:50:56] <icee> lg: not sure what I will do.
[21:51:07] <icee> lg: what switching frequency are you thinking you'll want to run?
[21:51:13] <LawrenceG> cradek, depends on how much of a code library one has at their disposal!
[21:52:02] <LawrenceG> icee, well, 15khz or higher... what is your high power block rated at?
[21:52:05] <icee> if i do run isolation, matching delays between either side of the bridge for each phase is important
[21:52:05] <alex_joni> LawrenceG: it depends how crappy the c-compiler is
[21:52:17] <alex_joni> I totally dig keil & c51
[21:52:19] <icee> lg: I'm hoping to run 100KHz, but I have not done all the math to determine if it is possible
[21:52:29] <alex_joni> in 5 mins you've got any simple app done
[21:53:34] <icee> lg: I'm planning to keep all the dead times, PID coefficient for current control loop, etc, in eeprom, btw.
[21:53:40] <icee> and enable them to be changed over EPP via a utility prog.
[21:53:57] <alex_joni_> alex_joni_ is now known as alex_joni
[21:54:09] <alex_joni> * alex_joni learned how to kill other instances of himself
[21:54:33] <alex_joni> sounds like a quote from "The One"
[21:54:58] <anonimasu> hehe
[21:55:24] <LawrenceG> icee, one should be able to use the eeprom inside the dspic for non volatile params
[21:55:31] <icee> yah
[21:55:34] <icee> that's what i'm saying
[21:55:42] <LawrenceG> (need to check what it has)
[21:55:45] <icee> 1k
[21:55:49] <icee> ought to be plenty.
[21:55:53] <LawrenceG> yep
[21:56:25] <icee> lg: anyways, if i design this on a 4 layer process, are you up to splitting the order of boards with me?
[21:56:33] <icee> it'll be like $40 apiece after shipping
[21:56:59] <icee> 40 for each of us, that is. one of us gets 2 boards, one gets 3 ;P
[21:57:33] <alex_joni> heh: [23:11] -NickServ- [alex_joni] has been killed
[21:57:38] <LawrenceG> with a 4ch board.... it isolation is an issue, it may be better to isolate at the pc interface side (8lines bidirectional + stobes is a lot... :{)
[21:58:13] <icee> yah.. though isolating 2MHz signals and delay matching is hard on the parallel port
[21:59:14] <LawrenceG> icee, yes on boards.... I would like to review design before sending to board house... I hate cut and hacks on 4 layer boards!
[21:59:24] <icee> yah, of course
[21:59:37] <icee> I figure i'll be giving you daily drops of schematics and layout as we learn more
[22:00:13] <alex_joni> ROFL: http://www.theregister.co.uk/2005/10/13/mp3_breast_implant/
[22:00:33] <icee> lg: and i'll assume the risk, too. if i do something really stupid/shitty i'll eat the board order myself
[22:00:45] <alex_joni> icee: that's no fun
[22:00:49] <icee> but your help in avoiding that would be appreciated with review
[22:00:50] <alex_joni> you need to eat the boards
[22:00:58] <anonimasu> lol
[22:01:15] <alex_joni> anonimasu: want an mp3 player?
[22:01:21] <icee> mmm.. pcboards.
[22:01:24] <anonimasu> alex_joni: I have one already :)
[22:01:31] <alex_joni> mmmm.. 36DD?
[22:01:39] <anonimasu> but not such a cool one ;)
[22:01:52] <icee> lg: and i should also be able to get my eldest brother who is a real EE to review all this.
[22:02:13] <anonimasu> alex_joni: nice :D
[22:02:16] <icee> lg: one thing we need to give thought to is how to test this in isolation
[22:02:18] <LawrenceG> icee, cool.... the circuits may actually fit quite well on a 2 layer board
[22:02:27] <icee> Oh, I'm sure it could work nicely in 2 layers
[22:02:34] <icee> I like the noise immunity etc from a good ground plane tho
[22:03:51] <LawrenceG> desk... chair, generator, scope, pc all mounted on coke bottles?
[22:04:09] <icee> haha
[22:04:11] <anonimasu> * anonimasu rolls around under the desk being careful not to dip any coke bottle
[22:04:12] <anonimasu> s
[22:04:36] <icee> I'm saying, we should prototype something taht 'pretends' to be an amplifier and feeds back current levels
[22:05:06] <icee> i can probably rig something up
[22:05:09] <Jymmm> dip or sniff ?
[22:05:45] <icee> it makes me a lot happier to have stuff working in unit test before plugging it all together, especially at the power densities we're talking about
[22:06:26] <LawrenceG> yes.... I was planning on early testing at 12vdc bus voltage!
[22:06:41] <LawrenceG> with current limited supply
[22:06:54] <icee> i wonder how much voltage it takes to turn the motors at all
[22:07:00] <icee> probably not much.
[22:07:37] <icee> that's one thing i'll try, is hooking up a current limited supply across two of the motor leads to see if it turns
[22:07:44] <LawrenceG> I fugure 12v should still give 300+rpm
[22:08:21] <LawrenceG> it will be easy to test the holding torque that way
[22:08:22] <icee> they need to hurry up and show up.
[22:08:32] <icee> I'm a little nervous because ups still claims to know nothing about the tracking number they gave me.
[22:08:46] <LawrenceG> mine are coming 4th class postage... if the dog sled doesnt get lost
[22:09:01] <fenn> what are you going to do with 1kw motors anyway?
[22:09:11] <icee> fenn: good question.
[22:09:16] <icee> haven't figured that out yet. :)
[22:09:24] <LawrenceG> fenn, I bought 300w versions
[22:09:31] <alex_joni> fenn: use 2 of them
[22:09:37] <alex_joni> one in one direction, one the other
[22:09:39] <icee> http://wwwapps.ups.com/WebTracking/processInputRequest?HTMLVersion=5.0&sort_by=status&error_carried=true&tracknums_displayed=1&TypeOfInquiryNumber=T&loc=en_US&InquiryNumber1=1Z6335880301902539&AgreeToTermsAndConditions=yes&track.x=19&track.y=13 <-- makes me nervous
[22:10:22] <LawrenceG> maybe they mailed them to me!
[22:10:23] <alex_joni> Jymmm: what should that have been?
[22:10:45] <icee> if UPS lost my servos i'll go postal :P
[22:11:00] <LawrenceG> I heard all these motors are from a warehouse in New Orleans
[22:11:26] <fenn> maybe looters got 'em :)
[22:11:49] <LawrenceG> ummm... tough but crunchy
[22:11:59] <Jymmm> alex_joni *shrug*
[22:12:17] <alex_joni> Jymmm: your servos order?
[22:13:10] <icee> lg: seriously? heh
[22:13:18] <anonimasu> I havent ordered any yet
[22:13:19] <anonimasu> :/
[22:13:23] <LawrenceG> much pulling of leg
[22:14:17] <icee> anon: well, the 1kW ones are all gone
[22:14:50] <anonimasu> how much is a 1kw servo?
[22:15:18] <anonimasu> icee: I dont care for thoose I cant ship them here
[22:15:49] <icee> anon: it was only $40 us
[22:15:55] <anonimasu> icee: it dosent matter
[22:16:09] <anonimasu> I cant ship that kind of weight
[22:17:27] <anonimasu> it was 17 lbs right?
[22:17:35] <icee> i don't remember
[22:17:46] <icee> yah.
[22:17:58] <anonimasu> that's a 30 kg package for 4..
[22:18:05] <anonimasu> shipping it from usa to sweden..
[22:18:15] <icee> yah.. that could get pricey.
[22:18:22] <anonimasu> 23kg for 3
[22:18:27] <anonimasu> but the shipping still would kill
[22:18:59] <anonimasu> can the rutex drives run the motors?
[22:21:15] <icee> anonimasu: no idea
[22:21:59] <anonimasu> yep..
[22:23:01] <anonimasu> I ordered 4 servos.. now
[22:23:01] <anonimasu> :)
[22:23:11] <alex_joni> that's a definate ... maybe
[22:23:27] <anonimasu> alex_joni: what?
[22:23:45] <alex_joni> that a rutex can drive the motors
[22:23:45] <anonimasu> is it safe to provide thoose people with your card number=?
[22:23:55] <anonimasu> alex_joni: it's been claimed at the board..
[22:23:55] <LawrenceG> cnczone says rutex runs them, but I'll wait to see what shows up when I get my motors
[22:24:01] <alex_joni> depends on your definition of safe
[22:24:10] <anonimasu> I cant really wait much longer
[22:24:13] <anonimasu> if I want servos..
[22:24:41] <anonimasu> what would be the cheapest shipping UPS ground?
[22:24:58] <alex_joni> don't think UPS ground ships to sweden
[22:25:14] <anonimasu> I dont know what to select..
[22:25:15] <alex_joni> no truck can drive through the ocean :)
[22:25:19] <icee> * icee knows nothing about international mail
[22:25:24] <icee> anon: you should call them up
[22:25:31] <icee> have them quote you shipping
[22:25:44] <anonimasu> it's 25$ to get it out of US..
[22:25:53] <alex_joni> 30 kg might be about 500$ on UPS :)
[22:27:03] <icee> 70 lbs was $300 via DHL
[22:27:06] <icee> but that was 2nd day
[22:27:18] <icee> i wasn't able to find anywhere that quoted it for surface transport
[22:27:52] <anonimasu> I dropped them a mail about it..
[22:29:09] <LawrenceG> icee, with all the motors sold, there are going to be a lot of people looking for drives
[22:29:31] <CIA-6> 03jmkasunich * 10emc2/src/emc/ (motion/command.c task/emctaskmain.cc): clean up handling of amp enables and faults
[22:29:40] <anonimasu> * anonimasu is in no hurry
[22:29:40] <anonimasu> :)
[22:29:55] <icee> lg: Yah. I'm not really interested in selling drives/amplifiers
[22:30:09] <LawrenceG> no.... me either
[22:30:10] <icee> lg: too many regulatory requirements to do it legally, plus liability, etc
[22:30:57] <icee> i do want to post my designs online tho.
[22:31:01] <icee> Maybe I'll put up a wiki
[22:31:10] <LawrenceG> icee, I won they pcb auction.... $13 for 50 boards and $23 for shipping
[22:31:12] <Jymmm> icee: I dont think most apply in a kit form.
[22:31:28] <icee> jymmm: You can still get sued when someone burns down their house
[22:31:44] <Jymmm> icee you can get sued for looking at someone funny.
[22:31:50] <icee> plus, it's a big hassle to make almost no money
[22:32:05] <Jymmm> not THAT's another topic all together =)
[22:32:05] <LawrenceG> icee, start a wiki page on info on the emc pages
[22:32:24] <icee> yah, that might be an option.
[22:32:32] <icee> i've got lyle.org, too.
[22:32:47] <CIA-6> 03jmkasunich * 10emc2/configs/ (sim.ini core_sim.hal standard_pinout.hal xylotex_pinout.hal): added estop loop jumper to hal configs, added simulation config (still needs realtime, just no hardware)
[22:32:47] <anonimasu> 400w should be plenty for most machine apps..
[22:33:10] <anonimasu> atleast anything I can fit yet ;)
[22:33:23] <LawrenceG> ok.... I am going to have a read through the dspic specs to look at the encoder inteface and the best way to interface to the pc
[22:33:28] <icee> so btw lg
[22:33:32] <icee> why did you buy the servos without brake?
[22:34:21] <icee> lg: also, using the dspic is predicated on the encoder working the way i think it does.. A/B quadrature + index
[22:34:23] <LawrenceG> smaller... I dont really need brake for my machine and if they are there, they must be energized all the time from 24vdc
[22:34:56] <LawrenceG> I hope so too...
[22:39:58] <anonimasu> hm I hope I get a reply soon
[22:40:03] <anonimasu> so I can order before they are all sold
[22:45:00] <LawrenceG> brb... got to go do some work outside...
[22:48:56] <LawrenceG> hey.... courier at the door... the dspic samples arrived
[22:49:36] <anonimasu> nice
[22:50:23] <Jacky^> * Jacky^ is waiting servodrivers :P
[23:04:44] <anonimasu> * anonimasu orders 3 servos
[23:08:50] <anonimasu> err 4
[23:11:42] <anonimasu> * anonimasu sighs
[23:11:50] <anonimasu> I am debating if I should order or not
[23:11:58] <anonimasu> the shipping is 180$
[23:27:42] <alex_joni> a bit much..
[23:27:47] <alex_joni> how much are the servos?
[23:37:13] <LawrenceG> see http://www.surpluscenter.com/sort.asp?UID=2005102016521585&catname=&byKeyword=yes&search=sanyo
[23:39:46] <Jacky^> anonimasu: any pic of your lathe online ?
[23:43:02] <CIA-6> 03jmkasunich * 10emc2/configs/ (mazak_rf.hal mazak_rf.ini): ongoing work on mazak - connected amp enables, faults, limits, and homes to motion controller, other minor tweaks
[23:55:08] <CIA-6> 03petev * 10emc2/configs/classicladder.hal:
[23:58:55] <fenn> CNC_Workshop: you probably want to join #emc-mazak also