#emc | Logs for 2009-12-01

[00:00:28] <Jymmm> does anyone know why it's common practice to nickel plat something before chrome plating it?
[00:00:38] <andypugh> I used to.
[00:00:52] <archivist_emc> so it sticks
[00:01:22] <Jymmm> chrome wont stick to steel?
[00:01:34] <andypugh> I think it is because nickel bonds to steel a lot better than chrome does, and chrome bonds to nickel better than it does to steel. I think there is also something technical about electrode potentials and porosity too.
[00:02:08] <Jymmm> Ok, then why chrome instead of two layers of nickle?
[00:02:35] <andypugh> Chrome is shinier. (also rather harder).
[00:02:38] <archivist_emc> I think there is a porisity problem with the chrome so the nickel underneath helps corrosion resistance
[00:02:48] <andypugh> I rather like nickel plating though, it looks classy.
[00:03:22] <Jymmm> Ok, do the same rules apply to silver and gold plating?
[00:04:01] <andypugh> Ah, another thing. I think that Nickel is self-levelling and chrome very much isn't.
[00:04:42] <andypugh> Industrial gold plating, for contacts etc, I would suspect might have a bonding layer. But that is rarely on to steel.
[00:05:51] <andypugh> Electroplating is a huge subject.
[00:05:54] <andypugh> http://www.amazon.com/Grahams-Electroplating-Engineering-Handbook-Set/dp/0412741105
[00:05:59] <archivist_emc> sure is
[00:06:07] <andypugh> Is a 3-volume handbook on the subject.
[00:06:12] <archivist_emc> see Cannings books as well
[00:06:27] <andypugh> That's the name! That was what I was looking for.
[00:06:53] <Jymmm> Ok, I need to make/find/buy some contacts that can handle like a million+ make/break (lo voltage DC) without tarnishing or wearing out. No clue wth I need for that.
[00:07:43] <archivist_emc> buy a microswitch
[00:07:54] <Jymmm> I dont know what chrome would be rated for.
[00:08:11] <archivist_emc> do not use chrome for a switch
[00:08:12] <Jymmm> archivist_emc: They need to be contacts, not a switch
[00:08:32] <andypugh> Steal the contacts out of a switch
[00:08:36] <archivist_emc> switches are contacts in a box
[00:08:56] <andypugh> As are relays, and contactors.
[00:08:58] <Jymmm> too expensive to do that, I need a coupel thusand
[00:09:20] <andypugh> In that case, ask one of the switch manufacturers.
[00:09:37] <archivist_emc> I still say use microswitches as the are cheap and rated for the operations
[00:09:54] <andypugh> Hmm, thingy pins.... We used to use them in test rigs...
[00:10:16] <andypugh> wb Casainho. Sorry, I went out to the workshop
[00:10:20] <archivist_emc> current will derate the number of ops
[00:11:03] <Casainho> andypugh, I am still in problems with EMC2 startup...
[00:11:49] <andypugh> Jymmm: Coda pins.
[00:11:53] <andypugh> http://www.coda-systems.co.uk/gb/index.html
[00:11:58] <Jymmm> andypugh: pogo pins?
[00:12:12] <archivist_emc> he wont like their prices :)
[00:13:03] <Casainho> I am still getting this error: /home/cas/scripts/repstrap-extruder.hal:16: pin 'halui.machine.off' does not exist
[00:13:19] <andypugh> OK.
[00:13:27] <andypugh> I think you need a line
[00:13:31] <andypugh> HALUI - halui
[00:13:39] <andypugh> Sorry, = halui
[00:13:44] <andypugh> In your INI file
[00:13:50] <Casainho> where do I put that line?
[00:14:17] <andypugh> I could tell you if my machine wasn't in bits
[00:14:19] <Casainho> ok, on init file, but in wht section?
[00:14:27] <archivist_emc> Jymmm, 40c ish 500 off and rated at a million http://www.rapidonline.com/Electronic-Components/Switches/Microswitches/V3-Microswitches/73516
[00:14:56] <andypugh> go to www.linuxcnc.org and search for halui in the search box on the left.
[00:15:06] <andypugh> (I will do the same and get back to you if I find it first)
[00:16:00] <andypugh> It goes in the [HAL] section
[00:16:51] <andypugh> that line creates a whole load of pins and parameters that lets you connect things such as actual physical switches to the user interface controls
[00:17:49] <Casainho> andypugh: ok, thanks. I got EMC2 running...
[00:17:57] <andypugh> Great!
[00:18:27] <Casainho> however, It should run a python program to update the buttons and etc... but looks like it do not run that Python program :-(
[00:18:33] <andypugh> HAL is brilliant when you get the hang of it, it will let you do almost anything, with anything.
[00:18:48] <andypugh> You commented something out?
[00:19:30] <Casainho> if I try to opn/run the python program using EMC2 open file command, looks like it works! but do not work when it should work...
[00:19:44] <Jymmm> https://gist.github.com/6947cc3f9f8b885c71df
[00:20:05] <Casainho> I tried to follow the author instructions, but EMC2 crash if I follow them...
[00:20:07] <Casainho> http://github.com/sam0737/hrepstrap/blob/master/README/README.txt
[00:23:09] <andypugh> OK.... That's a lot of custom stuff
[00:23:53] <andypugh> EMC doesn't crash, it just gives up, and generally says what the problem is.
[00:24:25] <andypugh> if you type dmesg after a failure at the command prompt you should be able to figure out what the problem is.
[00:24:30] <archivist_emc> Jymmm, whats the application
[00:25:47] <andypugh> Casainho: I am not sure that step 3 has come out right.
[00:26:41] <Casainho> andypugh: if I do it, EMC2 do not run and gives up
[00:27:09] <SWPadnos> Casainho, run from a terminal and look at the error message(s) printed there
[00:27:22] <andypugh> I think that line is saying that you need to paste the contents of repstrap_extruder.hal into custom_postgui.hal
[00:27:24] <Casainho> andypugh: tha'ts because I am not doing exactly as author says that I think it is not working as I expect...
[00:28:23] <andypugh> As SWP says, run emc from the command line and read what it says (just type EMC)
[00:28:33] <andypugh> Sorry, unix. type emc
[00:29:31] <Casainho> SWPadnos: ok, I did it now.. and EMC2 runs ok and no message on terminal... but it should run this python file: http://github.com/sam0737/hrepstrap/blob/master/hal/repstrap-extruder.py
[00:30:58] <andypugh> If you open that in emc then it probably will fire up a gui, which will then dump a load of G-code into EMC2
[00:31:10] <andypugh> (But I am guessing now)
[00:31:13] <SWPadnos> no
[00:31:21] <SWPadnos> that's a userspace HAL component
[00:31:22] <andypugh> Listen to SWP, not me
[00:31:56] <SWPadnos> which has to be explicitly listed in the ini file or a HAL file for it to load
[00:32:09] <Casainho> I did a 'print "running"' inside main() of that python code and I get nothing on command line... I was expecting seing "running" message on commnd line...
[00:32:20] <SWPadnos> in the ini file, you could do something like [HAL]HALCMD=loadusr repstrap-extruder.py
[00:32:45] <SWPadnos> have you added repstrap-extruder to one of your hal files?
[00:32:54] <SWPadnos> loadusr repstrap-extruder[.py]
[00:33:09] <SWPadnos> (I suspect you need the .py part)
[00:33:56] <Casainho> no, I didn't... I will try
[00:34:50] <Casainho> I got now this message: <commandline>:0: execv(repstrap-extruder.py) failed
[00:35:02] <Casainho> maybe I need to put full path for that file? will try..
[00:35:07] <SWPadnos> could be
[00:35:32] <SWPadnos> you may be able to "install" that component using comp, if you have the emc2-dev package installed
[00:35:50] <Casainho> It's running!!!! happy happy :-)
[00:35:54] <SWPadnos> yay
[00:36:33] <Casainho> well...
[00:37:04] <SWPadnos> now you probably need to connect a lot of the pins that component exports to other things, to get anything useful to happen
[00:37:11] <Casainho> at least a debug LED it's blinking on my hardware, which means that "repstrap-extruder.py" code file it's running...
[00:38:37] <Casainho> EMC2 is talking by serial port to a custom Arduino board I have with me, it will control a heater element and a stepper motor, to extrude plastic, so I can print using my CNC :-)
[00:40:24] <Casainho> yeah, EMC2 looks porwerfull ;-) -- here the project page: http://objects.reprap.org/wiki/Builders/EMCRepStrap
[00:42:16] <Casainho> thanks andypugh and SWPadnos :-)
[00:43:21] <andypugh> Aye, I can see one of those reprap printing heads in my future.
[00:43:23] <SWPadnos> good luck with it
[00:44:36] <andypugh> I am not sure why the extruder head can't be controlled direct by EMC. Perhaps there is too much analogue stuff going on.
[00:45:10] <MattyMob> you can get 12V sealant guns. instant extruder if you want a nice bead of silicone grout or Sticks Like Sh*t (tm)
[00:46:08] <MattyMob> the heater coil has a dedicated board to do the pwm
[00:47:13] <MattyMob> I'm looking round for one of them or a grease gun in all-steel, so it can be used as a crucible too for bulk melting
[00:47:59] <andypugh> I think the issue is using the parallel port, just not enough pins. With a Mesa card I reckon you could do it all in EMC
[00:49:09] <andypugh> I am not convinced a grease gun would work. Well, not without modification. The floating piston would hate you.
[00:50:25] <andypugh> The Repstrap (I think) works like a glue gun, so you have only the very minimum of mechanism getting filled with molten plastic
[00:51:05] <MattyMob> yep. filament is forced by a cog into a heated nozzle
[00:51:11] <andypugh> Ah, reading that page, they say pretty much what I said: Note: If you have the hardware (like interfacing card) providing all the additional input and output needed, you could interface the EMC2 directly to the extrusion motor, the heat element and temperature read-out. Potentially, you could translate the extrusion and temperature command as another "virtual" axis, and hook them up with the PID loop in the EMC2, just like how yo
[00:51:11] <andypugh> a normal axis. EMC2 comes even-better-than-RepRap algorithm and processing power to handle all these.
[00:54:03] <Casainho> andypugh: eheh :-)
[00:54:20] <Casainho> the question is that I even don't know how to do it ;-)
[00:54:33] <andypugh> I bought a whole load of crimp terminals for PCB headers from Maplin (all the stock of two shops) Then when got home I realised that they are the type that only work in a housing....
[00:54:47] <Casainho> Sam Wong made that Pyhton code, AVR microcontroller code, and hall UI...
[00:55:20] <Casainho> I am just trying to follow his steps and learn also ;-)
[00:56:22] <andypugh> I can see how it would all work. But I have been at this for a year now. (don't go thinking I am anything other than an EMC2 noob)
[01:01:15] <andypugh> I am seriously considering mounting a solderless breadboard permanently in my CNC controller case as a way of having reconfigurable pin wiriing. Like a hardware version of HAL
[01:01:34] <Jymmm> archivist_emc: morse code keyer
[01:02:42] <MattyMob> andypugh, rip up an old ide or floppy cable
[01:03:09] <andypugh> They are IDC, not discrete wire
[01:03:27] <MattyMob> I'd even reuse the terminals, but those IDC type really need soldering after the first use
[01:03:35] <andypugh> http://www.coolcomponents.co.uk/catalog/product_info.php?cPath=48_49&products_id=199
[01:03:46] <andypugh> Is what I want, but they are way too long
[01:05:15] <andypugh> (Also 26p each x 48 pins x 2(ish) is a lot of money.
[01:05:42] <MattyMob> you can grind a 40 pin idc into 20 housings
[01:06:10] <MattyMob> grind/hacksaw
[01:06:43] <andypugh> I suppose so. I could just extract all the IDC bits and solder them on.
[01:07:03] <MattyMob> use the old housings with the new contacts
[01:07:12] <andypugh> But at that point £25 for a hundred wires looks cheap again
[01:07:17] <Jymmm> andypugh: http://www.seeedstudio.com/depot/cables-c-34.html
[01:08:20] <andypugh> Thanks, just the job except for too long, in the wrong continent, and sold out.
[01:08:39] <andypugh> (Sorry, that came out more sarcastic sounding than intended)
[01:08:50] <MattyMob> how long do you want them?
[01:09:09] <andypugh> About 50mm
[01:09:25] <andypugh> And 300mm
[01:10:47] <MarkusBec> MarkusBec is now known as MarkusBec_away
[01:11:00] <andypugh> ie short for point to point from the 7i43 breakout to the buffers, pull-ups, PWM convertor, etc. And then long ones to the stepper amps
[01:11:40] <andypugh> I might just hard-wire onto header pins.
[01:12:04] <Jymmm> andypugh: What $5 not cheap enough?
[01:12:20] <MattyMob> do you need a housing with crimp contacts? use heatshrink
[01:12:27] <andypugh> The price is good
[01:12:53] <andypugh> The country, length and sold-out-ness less so.
[01:13:18] <Jymmm> andypugh: just look for a ardunio distributor in your area
[01:14:00] <andypugh> That was the first link I posted, coolcomponents.co.uk
[01:14:30] <andypugh> (which looks jolly interesting, and in need of a full exloration)
[01:16:46] <MattyMob> * MattyMob already doing it
[01:17:08] <Jymmm> LawrenceG: I got a new toy yesterday
[01:17:10] <MattyMob> it's pointless looking at shinies right now. cashflow crisis
[01:17:47] <andypugh> Hmmm...
[01:17:49] <andypugh> http://www.instructables.com/id/Sponge-Ferric-Chloride-Method-Etch-Circuit-Bo/
[01:17:53] <LawrenceG> Jymmm, well.. thankfully back to linux... got dvdrw flashed... man is windows stupid
[01:18:05] <Jymmm> LawrenceG: lol
[01:18:15] <MattyMob> I've always had a developer tray handy
[01:18:26] <MattyMob> both parents had darkrooms at home
[01:18:48] <LawrenceG> on this box, it did not have drivers for network, video or sound.... all standard on ubuntu
[01:18:51] <Jymmm> LawrenceG: I picked up a Bencher BY-2 and a MFJ-407D yesterday for $50 USD
[01:19:00] <andypugh> Yes, but this claims to be quicker and better.
[01:21:02] <andypugh> That looks incredibly simple.
[01:21:11] <andypugh> I will be trying it tomorrow.
[01:23:03] <MattyMob> doesn't everyone mill their pcbs these days? :)
[01:23:27] <andypugh> Not if etching is as simple as that looks
[01:24:07] <MattyMob> coating with etch resist is still tricky, or do you use transfer?
[01:24:23] <andypugh> Though I recall reading on the intertubes about someone CNC raster sperk-eroding PCBs
[01:24:38] <MattyMob> ooh gotta try that
[01:24:48] <andypugh> I use a flatbed plotter and OHP pen
[01:24:48] <LawrenceG> Jymmm, pictures please
[01:24:52] <MattyMob> micromig
[01:25:03] <Jymmm> LawrenceG: JAS...
[01:25:54] <andypugh> My sister used a micro oxy-fuel welder. The nozzles are hypodermic needles, 1mm flames @ 1500 degrees
[01:26:00] <Jymmm> LawrenceG: http://www.bencher.com/ham/images/by2.jpg
[01:26:21] <Jymmm> LawrenceG: http://www.universal-radio.com/catalog/morse/1324.html
[01:27:15] <andypugh> <missing the point> Can't a computer convert ASCII to morse far better than a human? </missing the point>
[01:27:26] <MattyMob> my friend brother made metal-in-water welders for dental techs
[01:28:32] <Jymmm> archivist_emc: BTW, I need contacts sorta like the ones shown here (gold plated): http://www.bencher.com/ham/images/by2.jpg
[01:28:59] <LawrenceG> Jymmm, cool.... I have the same bencher in black... most of the new hf radios have a keyer built in.... I dont have a separate keyer
[01:29:09] <andypugh> Sounds like exactly the same thing. Electrolyse water to O2 and H2, bubble through acetone flux, and burn in a tiny, teeny gas torch?
[01:29:33] <Jymmm> LawrenceG: they were sold together off of craigslist
[01:31:15] <Jymmm> LawrenceG: I dont have an HF rig, gonna build this... http://www.sparkbangbuzz.com/easy-ten/easy-ten.htm
[01:31:34] <Jymmm> LawrenceG: I still need to find a receiver
[01:32:49] <LawrenceG> Jymmm, cool.... build the transmitter.... maybe I can hear you if the band co-operates
[01:33:39] <andypugh> Why do folk still use morse? Is it that it is still intelligible at greater distances at moderate power, or just an aesthetic thing?
[01:34:04] <Jymmm> LawrenceG: Maybe... I still have nfc on a receiver though.
[01:34:57] <Jymmm> andypugh: It's not required to get a license anymore, Yes it does allow to communicate thru noise better than voice. Sorta becoming a lost art
[01:35:44] <LawrenceG> andypugh, still very effective mode... the brain still works better for decoding than any computer program I have seen so far
[01:35:49] <Jymmm> andypugh: look at that linkabove for what you need for a transmitter to BUILD yourself
[01:35:58] <andypugh> I am looking.
[01:36:45] <andypugh> What Marconi would have given for a quartz oscillator and a transistor
[01:36:55] <Jymmm> eh
[01:36:57] <Jymmm> heh
[01:37:51] <andypugh> Curiiously enough I was reading this a few days ago.
[01:37:52] <andypugh> http://en.wikipedia.org/wiki/Audion_tube
[01:38:23] <Jymmm> LawrenceG: I do have a question on external iambiac keyers though...
[01:39:50] <Jymmm> LawrenceG: Does the output from the external keyer that you connect to the transmiter need anything special since I'm using an iambic paddle?
[01:41:21] <LawrenceG> no... the output of the keyer should emulate a staright key... look in the manual for your keyer to see what the output stage looks like... it is probably a transistor that switches to ground
[01:41:43] <Jymmm> Yeah, it looks that way inside
[01:42:20] <Jymmm> ok cool, maybe I can just use the output of the keyer as a straight key for that circuit above
[01:42:39] <danimal_garage> i have no clue what you 2 are talking about
[01:42:53] <Jymmm> danimal_garage: Morse code
[01:43:05] <danimal_garage> oh ok
[01:43:21] <danimal_garage> is that like a cell phone
[01:43:26] <Jymmm> In the first pic, that is a "straight key" http://www.sparkbangbuzz.com/easy-ten/easy-ten.htm
[01:43:30] <danimal_garage> j/k
[01:43:49] <andypugh> Do the two paddles do long and short, or are they identical and just a way of getting an ergonomic rhythm?
[01:44:19] <Jymmm> danimal_garage: I have an "iambiac key" which is two paddles, one is dit, the other is dah
[01:44:48] <danimal_garage> i just checked that link, pretty cool
[01:44:50] <Jymmm> if you hold down the left paddle, you get dit's evenly spaced.
[01:45:12] <andypugh> I know a rude limerick about Iambic Pentameter...
[01:45:28] <Jymmm> so instead of tap tap tap for get three dits, you just hold it down till you get as many as you want.
[01:45:40] <Jymmm> andypugh: yes, long and short.
[01:46:05] <danimal_garage> man i actually worked hard today. i had 4 machines running at once at one point, and none of them were cnc
[01:46:30] <Casainho> SWPadnos: do you know why in have "self.c['fault.communication'] = 1" like that "hal pins" on Python code but none of them seems to change the custom UI?
[01:46:44] <andypugh> You could combine morse and twitter to send music. There is a way to encode music into 140 characters...
[01:46:46] <andypugh> http://www.newscientist.com/article/dn18173-best-of-twitter-tunes-album-released.html
[01:47:16] <danimal_garage> lol i wish i had that sort of free time
[01:47:44] <danimal_garage> as it is, i dont even have time to put up a real website for my own business
[01:48:10] <Jymmm> andypugh: http://www.youtube.com/watch?v=vsMiTKeTook
[01:48:20] <andypugh> Busy is better than the alternative
[01:50:03] <danimal_garage> i dunno
[01:50:16] <danimal_garage> i had more $$ when i was employed
[01:50:20] <Jymmm> http://www.youtube.com/watch?v=OYRsymVnuAU&NR=1&feature=fvwp
[01:50:21] <danimal_garage> and more time lol
[01:51:28] <andypugh> I found it interesting that the ww2 codebreakers could recognise individual morse code practioners, and so work out inportant information about the message (U-boat, Kireigsmarine, land-based, weather report..0
[01:52:16] <danimal_garage> life will be easier when i get servos on my mill, and get my toolchanger working
[01:52:31] <skunkworks> that looks so staged it is unreal.
[01:53:21] <MattyMob> danimal_garage, turn out some commodity items that don't need much tooling
[01:53:37] <MattyMob> reprap parts are in high demand atm
[01:53:58] <danimal_garage> http://homebrewedcomponents.wordpress.com/product-pictures-and-pricing/
[01:54:02] <danimal_garage> thats what i make
[01:54:06] <andypugh> I picked up a nice drawer slide at work today and got to thinking toolchangers.
[01:54:29] <danimal_garage> not too crazy, but i gotta make alot of them
[01:54:51] <danimal_garage> plus ti and stainless can be fun to machine
[01:55:15] <danimal_garage> i need to get creative with the tooling so i can be efficient
[01:55:29] <MattyMatt> mm, een fietsmakker. could you make a bakfiets like this? http://graffitiresearchlab.com/?page_id=99#video
[01:55:40] <danimal_garage> i gotta make up a quick change pallet system
[01:57:07] <andypugh> Nurse! Matts gone all Dutch again!
[01:57:59] <MattyMatt> daar in dat kleine cafe on the haaaaaaven
[01:58:31] <danimal_garage> que?
[01:58:53] <danimal_garage> un gato en mis pantalones
[01:58:56] <andypugh> Which confirms my opinion that dutch is just very, very, badly spelt enlglish
[01:59:00] <MattyMatt> daar zijn de meeeeensen gelijk aaaaaand tefree
[01:59:39] <MattyMatt> halfway between german & english
[02:00:27] <danimal_garage> i'm a fan of the dutch oven myself
[02:00:33] <danimal_garage> one of my favorite moves
[02:01:17] <MattyMatt> anyway, a bakfiets like that would work on english roads
[02:01:32] <MattyMatt> in a flattish town with good brakes
[02:02:03] <andypugh> Danimal, have you seen the SRAM Red casstte? That's a lovely piece of work, 10 speeds machined from solid.
[02:02:41] <andypugh> (Hollow in the back with a pressed-in drive plate)
[02:02:58] <andypugh> http://www.cambriabike.com/shopexd.asp?id=52552&page=SRAM%20RED%2010%20SPEED%20OG-1090%20ROAD%20CASSETTE
[02:03:05] <danimal_garage> yea, i have one on my commuter
[02:03:34] <danimal_garage> oh wait not that one
[02:04:04] <danimal_garage> neat
[02:04:10] <andypugh> Not inexpensive
[02:05:10] <danimal_garage> yea i wouldnt pay 200 for a cassette
[02:05:11] <andypugh> It looks more impressive withthout the infill plate. There is really only just enough metal in there to hold the teeth in place
[02:05:42] <andypugh> (I wouldn't pay $200 for a _car_)
[02:05:46] <danimal_garage> although i sell $140 chainrings
[02:06:01] <danimal_garage> haha me neither
[02:06:07] <danimal_garage> well
[02:06:21] <danimal_garage> i paid 400 for my truck, but i've made a ton off of it
[02:06:21] <andypugh> Though I have never owned a car...
[02:06:52] <danimal_garage> never?
[02:06:52] <andypugh> 2 wheels good, and 180hp better :-)
[02:06:59] <MattyMob> you could have one split cog that inserts extra teeth to change gear. /me patents
[02:07:33] <danimal_garage> matty, i only do single speed stuff
[02:08:05] <andypugh> Matt, it's been tried, there was one version on Tomorrow's World in about 1983
[02:08:06] <MattyMob> I'd like to make fancy hubs eventually
[02:08:36] <andypugh> Hubs are on my list too.
[02:08:40] <danimal_garage> andypugh, i've done the whole sportbike thing, i have too heavy of a throttle hand for that
[02:08:45] <danimal_garage> same here
[02:08:54] <MattyMob> brakes, motors, generators, 3/5 speed sturmey archers etc :)
[02:09:32] <andypugh> I want to make a trandem that works. I made one that didn't really. It seems that standard wheel bearings can't take 3 fat blokes
[02:09:55] <danimal_garage> once i get my hardinge finished, i'm gunna tinker with hubs
[02:10:19] <andypugh> My trandem was very slow 1-up, really quite quick 2-up, and very slow again 3-up
[02:10:21] <danimal_garage> matty that link you sent me locked up my pc
[02:10:37] <MattyMob> sorry
[02:11:02] <danimal_garage> not your fault, my computer came out of a dump
[02:11:10] <danimal_garage> just letting you know i couldnt see it
[02:11:21] <MattyMob> it's not I site I know and trust. mea culpa
[02:12:07] <andypugh> You spend 8 hours a day in bed and 8 hours a day on your PC, two places not so skimp :-)
[02:12:38] <MattyMob> it was very nice bakfiets anyway. a low cart in front of back half of a bike
[02:13:12] <danimal_garage> i have 6 computers running in this house, this one is the garage rat
[02:13:23] <danimal_garage> neat
[02:14:22] <andypugh> was it here that someone was discussing the chinese driving test. or elsewhere?
[02:14:43] <danimal_garage> i dunno
[02:14:54] <danimal_garage> i do need to pick up another emc2 box for my lathe
[02:15:08] <danimal_garage> not sure what to get
[02:15:10] <andypugh> ie, "What is the legal maximum height to load a bicycle, 2m, 4.3m or 5.7m?"
[02:16:04] <andypugh> (Thinking about it, it was at work, the commercial vehicle chaps)
[02:16:55] <andypugh> they had to take a chinese driving test to test vehicles there.
[02:17:24] <danimal_garage> what kind of bike do you have andy?
[02:17:25] <andypugh> Anyway, the answer is 5.7m, as otherwise your bicycle will not fit under bridges.
[02:17:44] <danimal_garage> must be 1000cc or above
[02:18:38] <andypugh> I have a Carrera 1-andy-power, a GasGas EC200 and a Yamaha R1
[02:19:56] <danimal_garage> a yamaha guy huh
[02:20:45] <MattyMob> i know someone wants rid of his VFR750. nice V4
[02:20:47] <danimal_garage> my last crotch rocket was a 954rr
[02:21:00] <andypugh> And back on topic, I bought a cheap old 1U server (dual olde-skool Xeon) and that fits nicely in a rack with the VFD and stepper amps etc. Also, with the SMI patch and a custom SMP kernel it runs 2,500nS latency.
[02:21:25] <danimal_garage> nice
[02:21:26] <andypugh> VFR750 are really surprisingly dull.
[02:21:49] <MattyMob> it looks like a jetski
[02:22:02] <MattyMob> could be one, with a bit more FG
[02:22:15] <danimal_garage> i'd never buy another v motor
[02:22:20] <andypugh> http://picasaweb.google.com/bodgesoc/Gibbs#5398215381172597314
[02:22:27] <danimal_garage> my tl1000r was a load
[02:22:48] <andypugh> Server on the left, amps and PSUs in the middle, VFD on the right.
[02:23:12] <danimal_garage> clean
[02:23:20] <andypugh> Rack made of extruded Bosch profile and Mdf
[02:26:32] <andypugh> If I did it again, the stepper connectors would all be the Speakon connectors on the bottom of the VFD module. You can't argue with 20A/250V in a £3 connector with positive locking and a real sense of quality
[02:27:45] <danimal_garage> i need to find a use for my old steppers and drives
[02:27:48] <danimal_garage> or sell them
[02:28:11] <danimal_garage> i doubt i'd get anything for them though
[02:29:04] <MattyMob> make a robot arm to load your mill
[02:29:19] <MattyMob> no need for qc palette then :)
[02:29:26] <MattyMob> pallet
[02:29:53] <danimal_garage> i should make a robot arm to hit the cycle start button too
[02:30:14] <MattyMob> and to preorder fresh stock
[02:30:58] <danimal_garage> the possibilities are endless
[02:31:58] <MattyMob> has anyone made a dot matrix seed sower?
[02:32:11] <MattyMob> say it with flowers
[02:32:23] <andypugh> Probably not.
[02:32:36] <danimal_garage> cant say that i have
[02:32:39] <andypugh> Now _that_ is worth a patent
[02:33:27] <MattyMob> worth building a custom bakfiets for maybe. they are quieter in public parks
[02:33:28] <andypugh> It's not even hard.
[02:33:55] <andypugh> I am thinking acre-scale artworks
[02:35:10] <MattyMob> I'm think hit and run graffiti attacks
[02:35:19] <MattyMob> >:)
[02:35:34] <danimal_garage> i'm thinking bratworst and beer for dinner
[02:35:35] <MattyMob> football slogans etc
[02:35:43] <danimal_garage> with a side of macaroni and cheese
[02:35:51] <MattyMob> no sleep till brakfast
[02:36:11] <andypugh> People are playing at "big art"
[02:36:15] <andypugh> http://www.jimdenevan.com/jim.htm
[02:36:19] <danimal_garage> no sleep till brooklyn
[02:36:30] <andypugh> http://vowe.net/archives/011027.html
[02:37:05] <andypugh> But google-maps scale full-colour images?
[02:38:17] <MattyMatt> you have to make them permanent. google maps still shows my car on the road on a nice sunny day in 2006
[02:38:23] <andypugh> You could probably even get an Arts Council grant
[02:39:11] <danimal_garage> google maps finally updated my house's picture
[02:39:27] <danimal_garage> it's only about a year old now
[02:39:41] <danimal_garage> my truck is dent free
[02:39:57] <danimal_garage> and my palm tree is still alive
[02:40:06] <danimal_garage> ah those were the days
[02:40:15] <danimal_garage> those days of early 2009
[02:40:17] <andypugh> They could easily be pirennial. It would be fascinating to watch a vast Dali (for example) evolve over the years.
[02:40:54] <MattyMatt> animate it by staggered flowering times
[02:41:01] <Neo_The_User> does anybody know RustamKh? He had some problems with my HOWTO on building emc.
[02:42:58] <andypugh> Just on field saying Manic.... Miner..... at 3.17E-8Hz...
[02:43:40] <MattyMob> ooh yeah I could do it in a garden at 16x16
[02:46:28] <andypugh> Neo_The_User: Never seen him here, and I am one of the very few people who would be using that link, I suspect
[02:47:02] <Neo_The_User> ah ok thanks
[02:48:47] <andypugh> Although, I think I have linked to it from the mailing list, so there is no way to know who might have seen the link.
[02:52:32] <andypugh> EEk! 3am. Work in 5 hours. Must leave.
[02:53:03] <MattyMob> thinkgeek.com btw
[02:53:16] <MattyMob> don't go there :)
[02:53:26] <andypugh> Sage advice
[03:10:36] <danimal_garage> 3am huh
[03:10:39] <danimal_garage> 7pm here
[03:13:53] <danimal_garage> i have no idea how this endmill is still going
[03:14:03] <danimal_garage> i'm waiting for the snap any second
[03:45:25] <ChanServ> [#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/, wiki at http://wiki.linuxcnc.org/"
[06:28:20] <eric_mill> I think my drives may be smoking my serial ports
[06:30:02] <eric_mill> I can see the bytes going out on pin 3, but loopback doesn't work
[07:02:54] <tomp> tomp is now known as tom3p
[09:09:31] <EbiDK> EbiDK is now known as EbiDK|AWAY
[10:41:58] <MarkusBec_away> MarkusBec_away is now known as MarkusBec
[10:56:59] <MrSunshine> I LIKE MY MACHINE =)
[10:57:07] <MrSunshine> never before has fabrication of parts been so simple =)
[10:57:16] <MrSunshine> to me that is =)
[10:57:51] <MrSunshine> MDF sucks to mill in.. atleast with 4 flute endmills
[11:00:33] <roh> hm?
[11:13:07] <roh> MrSunshine any reason? works fine here with a 6-flute endmill
[11:13:38] <MrSunshine> roh, packs it in the endmill etc
[11:13:42] <MrSunshine> so tha tit doesnt cut like it should
[11:13:56] <MrSunshine> might be my low rpm tho ... i think its like 1500 or something .. half burned motor on the mill :P
[11:14:42] <roh> eh.. yes... wrong rpm is a recipe for desaster and ugly cuts
[11:14:52] <MrSunshine> mm
[11:14:58] <MrSunshine> gonna switch the motor on it when i can afford a new one
[11:15:12] <MrSunshine> removing the whole damn housing for gears etc and just put a motor on there =)
[11:15:17] <MrSunshine> less weight and higher rpm =)
[11:15:24] <roh> less torque
[11:15:38] <roh> depending on how you dimension it.
[11:16:07] <roh> we will also need to switch ours sooner or later. some people switched to vfd and 3phase motors on a bf20
[11:16:17] <roh> so i know its possible.
[11:16:27] <MrSunshine> well its going to have cogged belt from th emotor to the chuck thingie and dunno what motor yet .. some strong motor should suffice :P
[11:16:37] <roh> we have afaik a 2:1 gear, which we can switch to 1:4 for 'slow'
[11:17:25] <roh> no belts at all. only direct drive for the steppers and gears for the spindle
[11:21:32] <piasdom> g'mornin all
[13:02:27] <KA1> hello
[13:08:52] <KA1> www.zanner.sbg.at
[13:09:53] <MattyMatt> ein fabriek!
[13:10:39] <MattyMatt> is that a mechmate in the front?
[13:11:02] <MattyMatt> ooh :) http://www.zanner.sbg.at/default_e.htm
[13:11:47] <KA1> we've our own products
[13:12:36] <MattyMatt> "Big Autumn Action goes ahead even by the *new generation of CNC-controlled Routers!"
[13:12:56] <MattyMatt> I'd run that past a marketing person
[13:13:32] <MattyMatt> a C coder sees "pointer to new" in there :)
[13:13:57] <MattyMatt> sorry, no offence mate
[13:14:23] <eric_unterhausen> that's not pointer to new, it's dereferencing new
[13:16:50] <MattyMatt> sorry, I'm really an asm coder at heart. ld a,(new) is read as "get a from address pointed at by new" and in mental shorthand "get a from pointer (to) new"
[13:19:14] <eric_unterhausen> I was reading some of my assembly code for the Rabbit a while ago, it drives me nuts because it's such an asymmetric instruction set
[13:19:40] <eric_unterhausen> mixing c/assembly was fairly nice though
[13:20:27] <MattyMatt> the 80188 support chip in this printer is labeled Rabbit. I don't suppose that's related
[13:22:20] <MattyMatt> after all these years, when I make an example, I still do Z80 :)
[13:22:55] <MattyMatt> your first asm is always the strongest ;)
[13:24:18] <MattyMatt> mm speaking of which, gotta look at the ones in mp3 players sometime. z80+dsp+usb for $20 could be handy
[13:24:41] <eric_unterhausen> I figure they are all arm
[13:25:03] <MattyMatt> that's what I guessed too, but it's z80 + dsp
[13:25:12] <eric_unterhausen> what is?
[13:25:30] <MattyMatt> the chip in all the cheap mp3 players
[13:26:19] <MattyMatt> they even have a stereo amp of sorts. could be enough to drive one tiny motor
[13:29:20] <MattyMatt> and afaics, usb is plenty fast enough to control real time machines, provided you don't run it through the general purpose usb framework of a heavy duty OS
[13:30:10] <MattyMatt> usb 2.0 as a caveat, but even usb1 has similar throughput to parport
[13:31:14] <eric_unterhausen> throughput is fine, latency isn't
[13:31:21] <MattyMatt> obviously
[13:32:09] <MattyMatt> but with the right drivers, usb2 would have lower latency than parport
[13:32:48] <MattyMatt> tiny packets with no preamble or discovery
[13:35:08] <MattyMatt> ah, that's a big job. I'd have to learn how Ubuntu kernel handles USB normally, and divert that. and then the hardware (e.g. usb parport adaptors) needs decoding too
[13:51:00] <MattyMatt> nice http://www.coolcomponents.co.uk/catalog/product_info.php?cPath=27&products_id=352
[13:51:54] <MattyMatt> they do bare thumbsticks for 2.50 too. worth bearing in mind for pendant design
[13:55:28] <MattyMatt> bleh only 9 ppr. enough for manual tweaking of params, but no use as a rolling 2d encoder
[14:00:18] <DaViruz> wow, the eurotech pc104+ cpu cards were kind of expensive, $2100 for an intermediate one..
[14:00:49] <DaViruz> $1400 for the intel atom one..
[14:04:50] <MattyMob> intel were selling atom + chip for $45
[14:05:08] <MattyMob> that's enough left over for an SMT oven
[14:05:58] <MattyMob> reference circuit + 3x parport = laughing
[14:07:19] <MattyMob> I wonder if intel supply a reference bios with that
[14:08:06] <MattyMob> if not, you will have to coreboot :)
[14:09:42] <DaViruz> so what do i need to mount atom+chip, like 6 layers at least? :)
[14:09:42] <MattyMob> 8GB bios why not?
[14:10:01] <MattyMob> hmm never thought of that
[14:10:15] <MattyMob> 4 layers is reasonably cheap to get made
[14:10:33] <MattyMob> certainly reasonable compared to $1400
[14:11:24] <DaViruz> hell, you can make 2 single layer and one double layer 0.5mm pcb yourself and get 4 layers at 1.5mm
[14:11:34] <DaViruz> vias are of course a problem
[14:11:35] <MattyMob> how many pc104 on a 12"x12"? about 6?
[14:11:56] <MattyMob> cheaper if you fill a 12"x12"
[14:11:57] <DaViruz> 9
[14:11:58] <DaViruz> actually
[14:12:20] <DaViruz> but i need some other stuff like lvds video
[14:12:47] <DaViruz> and i doubt intel is selling at single quantity?
[14:12:47] <celeron55> also, if you'd make such a board, you could easily sell lots of them :P
[14:13:09] <MattyMob> atom bridge chip has "gpu"
[14:13:33] <DaViruz> hmm well paralell ttl to lvds converters are reasonably cheap
[14:13:43] <DaViruz> i think i have some samples lying around actually
[14:13:44] <MattyMob> yes that was single quantity price of first Atom. special offer really dunno if still available
[14:14:09] <MattyMob> ok 4x parport :)
[14:14:20] <DaViruz> hmm, well this is all very interesting but the idea behind this is to get the damn mill running :/
[14:14:40] <MattyMob> cheap, fast, and good. choose any 2
[14:14:53] <DaViruz> i'm thinking mesa 4i65, pc104 version of the 5i20
[14:16:28] <MattyMob> yep 72 bits should suffice :)
[14:16:37] <DaViruz> any guesses on how fast a computer is needed to run emc2 with axis?
[14:17:14] <DaViruz> with reasonable latency for 40kHz or so step output
[14:17:45] <archivist> I use an old 800meg athlon forr one and a 1g athlon for the 5 axis
[14:18:13] <MattyMob> no idea. but if you get a cheap pc104cpu on ebay and try 1 axis, then 2, then 3, you may get an idea
[14:18:14] <archivist> and the 5 axis does other work in the background
[14:18:33] <DaViruz> i run my current one one an intel atom 230 and it works pretty well
[14:19:08] <DaViruz> the reseller of the eurotech board said he'll try to hook me up with some cheaper board if i'd make a list of my requirements
[14:21:28] <MattyMob> afaik, a slower one will always work with steppers, just not always as fast as possible
[14:22:22] <DaViruz> will the mesa AnyIO have any speed advantages over the parallell port?
[14:23:16] <DaViruz> archivist: what kind of step frequencies are you getting?
[14:23:22] <DaViruz> if you use step/dir at all
[14:23:38] <archivist> I have 5 steppers on the 5 axis
[14:23:59] <MattyMob> when you've programmed the fpga, that thing could take over some of emc's duties
[14:24:02] <archivist> I only use half step
[14:24:40] <DaViruz> MattyMob: could yes, but i'm really not good enough with verilog to do something like that
[14:24:52] <MattyMob> VHDL it says :)
[14:25:04] <DaViruz> it's a xilinx, you can use either
[14:25:06] <MattyMob> and they provide stepper and servo controller code
[14:25:15] <DaViruz> but will it play with emc2?
[14:25:21] <DaViruz> without having to do major coding
[14:25:57] <DaViruz> archivist: oh, ok. i'm fixed at 10 microsteps
[14:26:13] <MattyMob> I expect so, it could be programmed as genuine parports
[14:26:29] <MattyMob> hopefully that's one of the examples they provide
[14:27:04] <DaViruz> there is the hostmot2 code the mesa-people seem to be using, but i'm not sure how it works
[14:27:14] <DaViruz> i don't think it unloads the cpu in any way though
[14:27:40] <MattyMob> that could be it. just plain parport emu in hw
[14:27:43] <MattyMob> in fpga rather
[14:27:47] <DaViruz> i'd really like to implement the whole motion controller in the fpga
[14:28:54] <DaViruz> maybe i'll just try the live cd in a couple of different old computers to get an idea of what kind of power i need
[14:29:17] <MattyMob> do mesa integrate their stuff with emc?
[14:33:20] <DaViruz> how do you mean?
[14:34:53] <DaViruz> the hostmot firmware seems to give 4 hardware encoder inputs, 4 DAC outputs, 32 digital in, 16 digital out
[14:35:22] <SWPadnos> up to 4
[14:35:33] <SWPadnos> or up to 8, if you use different firmware
[14:36:24] <DaViruz> how does emc2 communicate with it, Port-IO just like the parallel port?
[14:36:35] <DaViruz> for the general purpose IO
[14:36:47] <SWPadnos> it's mapped into the PCi address space
[14:36:57] <DaViruz> oh ok.
[14:37:06] <SWPadnos> I think it may be a memory map, not an IO map, though I'm not positive
[14:37:43] <DaViruz> so the speed advantages over parallel port would be little to none?
[14:38:00] <SWPadnos> no, that's a bad conclusion
[14:38:04] <DaViruz> that is, latency is still the limiting factor
[14:38:19] <DaViruz> yeah i realized that :)
[14:38:20] <SWPadnos> CPU latency is almost always the limiting factor
[14:39:05] <SWPadnos> the first speedup you get is that each read/write accesses 24 I/Os at a time, rather than 4,5, or 8 on the parport
[14:39:20] <SWPadnos> the second is that it's a PCI cycle, which theoretically runs at 33 MHz
[14:39:27] <SWPadnos> rather than 1 MHz
[14:40:21] <SWPadnos> on the worst PCI chipset I've ever seen, PCI accesses were still done about 3/microsecond
[14:41:55] <DaViruz> maybe i'll just get a step multiplier, i don't really need 1µm resolution..
[14:42:29] <SWPadnos> oh - as far as "advanced functions" go, the Mesa hardware is even better
[14:42:42] <SWPadnos> encoder counting at several MHz, and step generation speeds to match
[14:42:47] <SWPadnos> also fast/accurate PWM
[14:42:58] <SWPadnos> all with basically no CPU usage
[14:43:25] <DaViruz> hmm, how does the step generation work?
[14:43:28] <SWPadnos> with the Mesa cards (and other similar products), you don't need a base thread - you only run a servo thread at 1ms or so
[14:43:37] <SWPadnos> the FPGA has step generators in it
[14:43:40] <DaViruz> oh
[14:44:07] <SWPadnos> so latency becomes much less of an issue, since 50 uS latencies are pretty much irrelevant to a 1ms thread period
[14:45:36] <DaViruz> so, any guesses on minimum cpu to achieve decent latency?
[14:46:01] <SWPadnos> nope, it's not really CPU dependent
[14:46:11] <SWPadnos> it seems more chipset and
[14:46:18] <SWPadnos> "other stuff" dependent
[14:46:29] <SWPadnos> like SMI, power savings, ACPI, etc.
[14:46:46] <archivist> particular motherboards are bad
[14:47:06] <MattyMob> it'll be the bios 99% of the time, although linux bypasses that
[14:47:23] <DaViruz> i'm looking at pc104 stuff, i'm guessing there should have been some consideration to these issues when they were designed?
[14:47:39] <SWPadnos> you'd hope so, but you really have to experiment
[14:48:18] <MattyMob> pc104 design = "hey these chips are cheap, can we cram them on a board?"
[14:48:49] <DaViruz> and then sell the thing for $2800 and hope nobody notices?
[14:49:08] <MattyMob> it$ a p£an!
[14:49:46] <DaViruz> anyway, cpu needs are basically what i need to run x11 and render axis and such?
[14:50:36] <celeron55> pretty much yes
[14:51:07] <SWPadnos> depending on how large the NC cod efiles you expect to run are, and how smooth a display uodate you like, anything from a ~600MHz P3 to a few GHz will be OK
[14:51:25] <SWPadnos> ugh
[14:51:30] <SWPadnos> I typo a lot when I'm sick
[14:51:56] <celeron55> and something like 256MB of RAM is the minimum if you use the default ubuntu desktop
[14:52:09] <SWPadnos> yep, though 512 is better for the install
[14:52:18] <DaViruz> gnome will probably have to go either way
[14:52:47] <SWPadnos> you can install from the server/alternate install disc, then install emc2
[14:53:13] <SWPadnos> but I don't know if the dependencies are set up to give you a minimal X environment (since there are text-mode UIs for EMC2, X isn't strictly needed)
[14:53:14] <celeron55> actually, 128M is too little in any case, 192M is enough with a minimal desktop
[14:53:19] <DaViruz> i actually run it on crux now
[14:53:33] <archivist> having a mysql build and test running in the background proves to me that cpu needs are not large
[14:54:01] <MattyMob> do latency problems vanish with text mode UI?
[14:54:19] <DaViruz> i think i've underestimated rtai
[14:54:20] <eric_unterhausen> no
[14:54:21] <celeron55> most probably no
[14:55:15] <eric_unterhausen> a video driver can screw things up, but a lot of computers are bad in text mode
[14:55:51] <DaViruz> what about throwing the video card out alltogether :)
[14:56:18] <eric_unterhausen> I've done that
[14:56:21] <celeron55> i guess not many computers boot like that
[14:56:28] <archivist> text mode used to only write to video ram in the sync period
[14:56:28] <celeron55> or do they
[14:56:46] <tom3p> the gnome desktop applet 'system monitor' has a display called 'load', when my 'processor' monitor is hitting 100%, this 'load' display shows 'the system load average is 2.5'... wtf is it talking about?
[14:56:47] <SWPadnos> I actually have a HAL-only system running with no X and usually no network
[14:56:54] <SWPadnos> it has <1 uS latencies
[15:00:02] <MattyMob> archivist not usually on early PC. writing vram disrupted output so you got snow
[15:00:21] <celeron55> tom3p: i don't know how it's calculated but you can see it with various commands, like 'uptime', too
[15:01:03] <celeron55> http://immike.net/blog/2007/07/27/what-exactly-is-a-load-average/
[15:01:03] <archivist> MattyMatt, I know, hence writing in sync time
[15:01:10] <MattyMob> afaiaa anything up to VGA can only work that way, you have to choose to only write in vsync period to avoid snow
[15:01:11] <celeron55> that explains it, i guess
[15:01:22] <tom3p> celeron55: cool, never heard of that tool, ("1.05,0.74,0.71")
[15:02:25] <tom3p> celeron55: should 1 see 2 users, both me?
[15:02:39] <celeron55> i see 15 users on my personal laptop
[15:02:44] <tom3p> haha
[15:02:59] <tom3p> dopplegangers
[15:03:25] <celeron55> it's quite random, some systems count the bash windows in screen for those etc, some not
[15:03:48] <celeron55> you can see the users with the command 'w'
[15:04:43] <celeron55> umm... actually, it says 15 but lists only 13... strange :P
[15:05:49] <tom3p> w also tells the load, users, and TTY (2 different tty's... where the doppleganger comes from )
[15:06:08] <tom3p> thx
[15:09:03] <tom3p> what SWPadnos said about hardware handling cpu load, and latency vs thread speed, and 'other stuff dependant'... oughtta be on page 1 of linuxcnc.org & wiki.
[15:11:50] <tom3p> and his <1us xless, netless system... is that the future of emc2? a dedicated device that is monitored?
[15:13:45] <MattyMob> that seems to be right. parports are dying for a start
[15:14:14] <MattyMob> not all laptops even have one
[15:14:29] <tom3p> is the idea of plugin cards dying?
[15:14:54] <DaViruz> i'd say no laptops from the last 5 years have a parport
[15:14:55] <tom3p> fer sure fewer than my ancient isa bus boxes, with up to 11 slots
[15:15:04] <MattyMob> for machine control the isolation of a serial link is handy
[15:15:44] <MattyMob> sharing a cpu between motor control and status display is so 1980s
[15:16:20] <tom3p> dunno, commercial cncs i work on a parallel in nature ( not printer ports at all, but parallel vs serial )
[15:16:32] <DaViruz> a dedicated motion controller in an fpga would be nice
[15:16:51] <MattyMob> it's broken out to parallel eventually
[15:16:59] <DaViruz> not sure if it could be made versatile enough though
[15:17:25] <MattyMob> with 200k gates I'm sure it could be
[15:17:29] <eric_unterhausen> the whole idea that drives emc is that a computer can do the job and is cheaper and more general purpose
[15:18:41] <tom3p> if emc uses pcs, but headless netless thinking, then it can be 'canned' into vhl at some point
[15:18:42] <MattyMob> PC is ideal for setting up, but a machine could run on an embedded one too
[15:19:03] <MattyMob> yeah implement emc on mesa
[15:19:05] <MattyMob> :)
[15:19:17] <tom3p> yes
[15:19:22] <eric_unterhausen> that doesn't really make sense to me
[15:19:31] <eric_unterhausen> you can always move stuff to the fpga if you want
[15:19:34] <MattyMob> VHDL is just like C but with when(x){y;}
[15:19:35] <tom3p> thats a pmac or magellan
[15:19:59] <DaViruz> well if the mesa cards can contain the step generators i suppose that's enough
[15:20:03] <tom3p> i meant very high density logic, not the language
[15:22:56] <MattyMob> if the emc PC was embedded in the machine, then CAM programs could talk to that directly
[15:23:41] <MattyMob> and there'd be no worries about the CAM program upsetting latency
[15:23:43] <eric_unterhausen> hopefully the pc manufacturers can standardize on pcie for as long as they have used 5v pci
[15:25:05] <eric_unterhausen> I keep thinking it would be nice to have a separate PC for the display, emc seems to work fairly well that way
[15:25:05] <SWPadnos> there aren't issues with CAM programs upsetting latency anyway
[15:25:23] <MattyMob> even blender? :)
[15:25:37] <SWPadnos> you can do more or less anything you want in userspace without screwing up the realtime stuff
[15:26:02] <SWPadnos> blender shouldn't be more of a problem than Axis is (they're both openGL programs)
[15:26:24] <MattyMob> blender grabs the sound output too
[15:26:48] <tom3p> bbl
[15:27:04] <SWPadnos> well, you can play MP3's (or movies for that matter) while machining
[15:27:26] <SWPadnos> if you want to use certain programs while the machine is running, you should use those during your latency test
[15:27:52] <MattyMob> I wouldn't. I don't do that stuff when burning CDs even
[15:28:02] <celeron55> and play a driving simulator with a usb wheel and pedals
[15:28:09] <MattyMob> I remember when coasters cost $20 each
[15:28:24] <DaViruz> i don't even dare to use xcalc when machining ;)
[15:28:58] <DaViruz> and you'd be hapy if they'd work at 2x speed
[15:29:15] <celeron55> someone should make a video of crazily abusing a computer running emc while machining
[15:29:23] <eric_unterhausen> I'm not really that tempted to use my mill machine for other stuff, I have another computer in the next room
[15:29:32] <tom3p> SWPadnos: are these things voodoo? shouldnt they just slow down (user perspective) and leave emc running full speed?
[15:29:51] <celeron55> that's what realtime means
[15:30:15] <celeron55> the realtime stuff will take it's cpu time when it needs it and the rest is left for userspace
[15:30:34] <MattyMob> more programs = more risk of a system crash
[15:30:36] <SWPadnos> the realtime part would run at full speed, though it might be possible to run so many programs that the EMC2 userspace processes (namely the interpreter) mught not be able to provide motion commands to the RT code fast enough
[15:30:43] <MattyMob> divide and conquer
[15:30:47] <SWPadnos> but I think you'd have to work at it
[15:31:02] <tom3p> thx
[15:31:39] <MattyMob> and embedded one could have a totally dedicated kernel. no floppy drivers etc
[15:34:01] <MattyMob> I always intended to embed mine, although it's a full headed PC atm
[15:34:55] <eric_unterhausen> you can build your own kernel, it's really a fairly decent idea
[15:36:06] <MattyMob> yep. a rite of passage for gentoo fans >:)
[15:36:41] <MattyMob> debian's main attraction to me is there's none of that needed
[15:36:51] <eric_unterhausen> well, if you want a rite of passage, patch and build a kernel from the rtai source
[15:37:17] <MattyMob> that's what a proper embedded one would need yeah
[15:37:40] <MattyMob> so it'd better be a board that's easily replaced, to make it worth the effort
[15:38:54] <MattyMob> one off kernels for one off machines is NASA's bag
[15:40:36] <eric_unterhausen> I don't know why they can't build a stable version of firefox every release
[15:41:20] <MattyMob> 2.x was OK for me, but 3.x is constant fail
[15:41:26] <MattyMob> even on windows
[15:42:35] <skunkworks_> http://www.electronicsam.com/images/KandT/testing/pent4-26ghz1.png
[15:42:40] <skunkworks_> look at the task bar....
[15:43:03] <MattyMob> atom on pc104 seems like a risky choice for a LTS chip. 386SX has been available for nearly 20yrs
[15:43:23] <celeron55> skunkworks_: that looks fun 8)
[15:43:27] <MattyMob> I think they just stopped making them
[15:45:19] <MattyMob> skunkworks_, I have 4x desktops like that often :)
[15:46:07] <MattyMob> workstation != RT machine controller
[15:46:18] <MattyMob> I'm still sure of that, in my bones
[15:47:14] <MattyMob> run the sims on the workstation sure
[15:47:32] <MattyMob> even The Sims if you want
[15:48:05] <celeron55> of course it's not the best idea but it can be done and it is cheap
[15:49:11] <MattyMob> It's an aesthetic choice as much as anything. A friend with a server packs 12 hdd in one box and runs VMs. I'd use the same money+electricity to make 6x machines with 2 hdd each
[15:50:48] <MattyMob> he doesn't use hot spares tho, so every disk breakage is effectively downtime
[15:51:12] <MattyMob> and he gets a lot of breakage, with 12 in a box
[15:51:28] <MattyMob> 3x array of 4
[15:52:52] <DaViruz> wow the amazon kindle dx is pretty nice
[15:53:27] <DaViruz> for documentation and fictional litterature alike
[15:54:21] <MattyMob> no backlight is a dealbreaker for me, for fiction-in-bed
[15:56:30] <celeron55> it probably tries to be as much as possible like a real book for the eyes :P
[15:57:02] <MattyMob> it's the nature of digital paper. I wish IBM had kept it and used it for keytops
[15:57:25] <celeron55> an e-ink screen probably wouldn't support backlighting very well
[15:57:50] <MattyMob> not at all unless they changed the tech
[15:58:18] <celeron55> backlighting might consume notable amount of battery power also
[15:58:20] <MattyMob> the white was Ti oxide when I last looked
[15:58:33] <DaViruz> even considerable amount of battery power
[15:58:34] <celeron55> not a problem if you're reading in the bed, though...
[15:58:47] <DaViruz> the two weeks would quickly turn to two hours i suspect..
[15:59:05] <celeron55> maybe
[15:59:26] <DaViruz> or 20 at least, but what about frontlighting?
[15:59:36] <DaViruz> *at most
[15:59:38] <MattyMob> yeah but it's either that or hold a torch or fall asleep with the light on and get crap in the morning from momma
[15:59:58] <DaViruz> i don't do much reading in bed so it's not an issue for me
[16:00:11] <DaViruz> dim rooms though, i wonder how the contrast is
[16:00:23] <DaViruz> and how bright the white is..
[16:00:34] <celeron55> the screen is probably very good
[16:00:37] <MattyMob> not so much since I turned 18, but it's still a cherished memory, reading under the covers :)
[16:01:22] <MattyMob> these days, I just leave the light on. CFD bulbs don't ignite bedclothes
[16:01:28] <DaViruz> i just found out that the thing runs linux
[16:01:43] <DaViruz> probably exist a thousand hacks to run various things on it
[16:02:03] <MattyMob> there's a challenge :) emc
[16:03:04] <danimal_garage> my tablesaw runs linux
[16:03:07] <MattyMob> it's like a games console, you know there are limitations but you also know if you work around it it'll work on every other kindle too
[16:03:13] <DaViruz> i don't think rtai is ported to ARM :/
[16:03:45] <DaViruz> or maybe it's pretty much platform independant
[16:04:14] <MattyMob> dunno. to do a good job it'd need to mess with x86 runlevels
[16:04:26] <MattyMob> rings
[16:04:36] <celeron55> rtai has been ported to arm
[16:04:40] <celeron55> https://www.rtai.org/
[16:05:58] <MattyMatt> sweet. Nin DS is my first target
[16:06:27] <celeron55> hmm
[16:06:54] <MattyMatt> that must wait until I can mill PCB :)
[16:06:55] <celeron55> "StrongARM; ARM7: clps711x-family, Cirrus Logic EP7xxx, CS89712, PXA25x"
[16:07:04] <celeron55> kindle seems to have arm11
[16:07:25] <MarkusBec> MarkusBec is now known as MarkusBec_away
[16:07:30] <MattyMatt> probably upward compat, like arm9 will run arm7 code
[16:07:38] <MarkusBec_away> MarkusBec_away is now known as MarkusBec
[16:07:39] <MattyMatt> * MattyMatt checks
[16:08:01] <celeron55> ds has one arm9 and one arm7
[16:08:38] <MattyMatt> http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0344c/index.html
[16:08:41] <MattyMatt> yep
[16:09:51] <DaViruz> arm has grown really huge
[16:09:53] <DaViruz> really really huge
[16:09:59] <MattyMatt> yep
[16:10:23] <DaViruz> nice documentation resource
[16:10:32] <MattyMatt> yeah I like that site
[16:10:39] <DaViruz> i wonder if there's a kindle version available ;)
[16:10:50] <MattyMatt> should wget it before it goes the way of all ether
[16:11:15] <DaViruz> maybe thing's like that is why it's huge
[16:11:19] <DaViruz> and the developers realize that
[16:11:33] <DaViruz> (good documentation that is)
[16:12:03] <MattyMatt> it was designed for a proper home computer, the kind with BASIC in ROM and the cpu instruction set in the manual
[16:13:33] <MattyMatt> context change by cassette tape. by eck them were the good old days
[16:15:07] <MattyMatt> the computer was a flop unfortunately
[16:15:45] <DaViruz> yeah the world market is like 7 or 8 computers in total, tops
[16:16:16] <MattyMatt> not in 1985 it wasn't :) the writing was on the wall already tho. ibm or die
[16:17:29] <MattyMatt> I spent thousands on TRS-80s and got a harsh lesson in Moore's Law
[16:23:25] <L84Supper> http://www.rtai.dk/cgi-bin/gratiswiki.pl?RTAI_On_ARM
[16:24:01] <MattyMatt> Current state 2005-08-04:
[16:24:13] <MattyMatt> inactive wiki
[16:24:20] <MattyMatt> get editing :)
[16:24:51] <L84Supper> RTAI ARM Maintainer: Michael Neuhauser is working on the Cortex-8
[16:26:11] <DaViruz> how about MIPS
[16:26:12] <DaViruz> :)
[16:26:27] <L84Supper> I've seen up to Arm9 supported by RTAI so far
[16:26:27] <MattyMob> PSX or OpenWRT?
[16:26:42] <DaViruz> silicon graphics!
[16:27:18] <MattyMob> NDS has both arm7 and arm9 and linux
[16:27:31] <MattyMob> gotta be worth a try
[16:27:33] <DaViruz> and a couple of fun games
[16:27:51] <DaViruz> one dedicated arm for realtime stuff and one for UI
[16:27:54] <DaViruz> that's great
[16:27:55] <DaViruz> :)
[16:27:59] <MattyMob> only lego star wars has actually got me playing
[16:28:21] <MattyMob> limitations on what you can do with each cpu
[16:28:27] <MattyMob> they share memory etc
[16:28:42] <DaViruz> yeah i know
[16:28:48] <DaViruz> have done some developing for the ds actually
[16:29:15] <MattyMob> sweet
[16:29:38] <MattyMob> have you tried running linux on it?
[16:29:48] <DaViruz> yeah
[16:29:53] <L84Supper> ARM Cortex8 and 9 are going into netbooks in high volumes and there is lots of Linux development happening now for ARM, MIPS has stalled out in comparison
[16:30:11] <DaViruz> didn't find much uses though besides irc
[16:30:37] <MattyMob> there's that mips handheld system looks like a PSP, half the price of a DS
[16:30:55] <DaViruz> hmm
[16:31:01] <DaViruz> gp2x?
[16:31:08] <MattyMob> yeah linux w/o a keyb and hdd is kinda pointless :)
[16:31:09] <DaViruz> just guessing here
[16:31:33] <MattyMob> nope cheaper than gp2x
[16:31:35] <DaViruz> 4GB+ compact flash works fine for storage
[16:31:45] <DaViruz> and keyboard on lower screen works pretty well too
[16:32:38] <DaViruz> but even so i can't really think of any uses
[16:32:54] <MattyMatt> these things -> http://cgi.ebay.co.uk/NEW-MP5-Lite-Game-Console-Videos-Camera-Music-all-in-1_W0QQitemZ280430269308
[16:35:11] <MattyMatt> I'm thinking embedding an NDS in a machine. touch screen would be nice
[16:38:31] <MattyMatt> although once the mill is working I'll make a posh console with proper buttons for everything
[16:39:00] <eric_unterhausen> I'm trying to talk myself into the idea that my servo drives fried 4 different serial ports
[16:39:29] <MattyMatt> try #5, then you'll know for sure
[16:39:32] <eric_unterhausen> but it seems like if they would fry the rx they would probably also fry the tx
[16:40:29] <eric_unterhausen> I guess I'm going to have to get over my laziness and get rs485 working
[16:40:37] <archivist> are the serial ports real 12v ones
[16:40:38] <MattyMatt> depends. some level converters were in or out, so one chip =tx1+tx2, another = rx1+rx2
[16:41:25] <MattyMatt> I think the max232 was the first to combine them tho, and that's about base level primitive these days
[16:41:37] <DaViruz> eric_unterhausen: a similar thing actually happened at work
[16:42:17] <DaViruz> or wait, maybe not
[16:49:13] <eric_mill> archivist: not sure. I was looking at them with a scope, but didn't note the voltages
[16:49:26] <eric_mill> I tried a loopback test using hyperterminal and it didn't work
[16:49:42] <eric_mill> but I hate hyperterminal so much it may be lying to me
[16:51:18] <eric_mill> the serial ports have talked to the drives in the past
[16:51:30] <eric_mill> then the mill squished my finger and I lost interest for a while
[16:51:31] <MattyMatt> pc com ports are real 12V, that's the only reason ATX psu have -12V these days
[16:52:04] <MattyMatt> since the bios stopped being eprom
[16:52:35] <DaViruz> hyperterminal is really annoying
[16:52:54] <DaViruz> just the fact that you have to create AND NAME a session to connect
[16:53:02] <eric_mill> good news is my touchpad works
[16:53:31] <eric_mill> then again, that wouldn't make up for frying 4 serial ports
[16:54:07] <MattyMatt> maybe your -12V in the PC is dead. the only way you'd notice is the com ports these days
[16:54:24] <DaViruz> the TX on the computer died?
[16:55:21] <MattyMatt> you should have the driver chips in sockets, IMO
[16:55:40] <MattyMatt> that was common even for printer ports at one time
[16:56:50] <eric_mill> still would be two separate power supplies fried
[16:57:08] <eric_mill> DaViruz: the rx appears to be dead, tx ok
[16:58:08] <DaViruz> OK
[16:58:34] <eric_mill> but it's sooooo much easier to verify tx than rx
[16:58:50] <MattyMatt> check the 5V logic signal behind the rx buffer
[16:59:07] <MattyMatt> if there is a separate one
[16:59:34] <DaViruz> yeah, but loopback really should be dead sure since that will take away any baud rate etc issues
[17:00:17] <MattyMatt> " I tried a loopback test using hyperterminal and it didn't work"
[17:00:18] <DaViruz> a really good terminal for windows is realterm
[17:00:25] <DaViruz> MattyMatt: i know
[17:12:03] <eric_mill> this computer is filthy
[17:17:51] <eric_mill> loopback works on the other computer
[17:18:56] <eric_mill> as much as I hate to admit it, this one probably has a power supply out of spec
[17:24:52] <MattyMob> can you switch the servos to RS-422 mode? that's single ended iirc
[17:25:29] <MattyMob> supposed to be +5V, but the servos will probably still handle +12V tx
[17:26:26] <numen> hi
[17:26:58] <MattyMob> hmm don't suppose that will help much. the PC probably won't handle rx like that
[17:58:17] <micges1> micges1 is now known as micges
[18:24:02] <EbiDK> EbiDK is now known as EbiDK|AWAY
[18:44:05] <geo01005> I'm working on making a delta robot right now. From what I understand it might be the first delta robot to run on emc.
[18:44:19] <alex_joni> it would
[18:45:20] <geo01005> Should I expect any problems with the kinematics? (based on the recent messages on the emc list)
[18:46:44] <geo01005> "the trajectory planner might not be suitable for any machines with non-trivial kinematics"
[18:49:31] <eric_unterhausen> how different are the kinematics from a hexapod
[18:49:32] <eric_unterhausen> ?
[18:52:42] <geo01005> hmm, I haven't thought of that before...The delta robot that I'm making has three rotary actuators with all fixed link lengths.
[18:52:54] <geo01005> The delta robot isn
[18:53:11] <geo01005> isn't as difficult of a problem as the hexapod.
[18:53:57] <eric_unterhausen> is it one to one?
[18:55:28] <geo01005> Here is an example of one way to solve for the kinematics: http://forums.trossenrobotics.com/tutorials/introduction-129/delta-robot-kinematics-3276/
[18:55:50] <geo01005> There are multiple solutions, but only a single set that is used.
[18:56:18] <geo01005> It would be pretty easy to just use the code from that site.
[18:57:31] <geo01005> The code looks like it would be quite fast.
[18:59:26] <geo01005> I obviously won't be able to use the whole workspace because emc requires a rectangular limits on the workspace.
[19:00:37] <eric_unterhausen> it admits the possibility of no solution though
[19:01:54] <eric_unterhausen> typical geek, he spends the first half of his movie showing screenshots, and then the robot moving
[19:03:31] <geo01005> The only place that there is no solution is outside the workspace... I was expecting to limit the workspace to a volume that fits inside the workspace of the bot.
[19:03:42] <geo01005> I was thinking that would work, am I wrong?
[19:03:45] <eric_unterhausen> so it should work then
[19:04:21] <eric_unterhausen> you should be able to get it to work in simulation relatively quickly I would think
[19:04:48] <geo01005> Yeah, I just need to relearn how to do the kinematics code...
[19:07:07] <geo01005> There are some good examples though.
[19:08:03] <geo01005> I wanted to run this bot off an arduino, but the kinematics are way too much for it.
[19:08:39] <geo01005> EMC is much more powerful though.
[19:10:20] <Jymmm> Um, some built a spider using an arduino
[19:15:02] <geo01005> ?
[19:17:07] <motioncontrol> Good morning.a question about arduino.The analog in is 0-10 volt ?
[19:21:16] <motioncontrol> geo01005, a question about arduino.The analog in are 0-10 volt ?
[19:27:32] <Jymmm> motioncontrol: try http://arduino.cc
[19:27:55] <motioncontrol> Jymmm, ok more thanks
[19:28:20] <dimas> motioncontrol, by default it's 0-VCC
[19:29:18] <dimas> you can set external AREF voltage, looking what range is for that
[19:29:36] <motioncontrol> dimas, the default is 0- ?
[19:29:50] <LawrenceG> Jymmm, -.-. --.- -.. . ...- . --... .. - -.-
[19:31:06] <dimas> 0-5v or 0-3.3v if 3.3 is primary voltage for the board
[19:31:56] <dimas> i only have 5v arduino, so only guessing how 3.3v boards doing
[19:33:44] <motioncontrol> dimas, thanks for info
[19:38:02] <Jymmm> LawrenceG: .-.. .- .-- . .-. -. -.-. . / .-- - .... / -.. --- / -.-- --- ..- / .-- .- -. -
[19:39:08] <Jymmm> LawrenceG: -...- -.--.-
[19:39:33] <dimas> motioncontrol, np
[19:42:02] <skunkworks_> geo01005: you just need to know the limitations. Emc could command a 'joint' to move faster than it should because the limits are applied in world space. (I wonder if that made sense)
[19:42:12] <skunkworks_> (before the kins are applied.)
[19:45:11] <eric_unterhausen> inverse kins take world coords and transform them into joint coordinates. Are the accel limits really applied at the world level?
[19:46:12] <alex_joni> geo01005: I don't think there will be major problems applying emc2 to a delta robot
[19:46:30] <alex_joni> sure the TP has some limitations, but for the delta robot joint limits are not really an issue
[19:46:35] <alex_joni> (it's crazy fast ;)
[19:46:56] <alex_joni> and has no singularities in the working space, so that should be fine
[19:49:29] <motioncontrol> alex_joni, good evening.please i study the axis interface and i have see the c.load_tool_table() for reload the table tool.For load the partprogram wath is the fuction ? thanks
[19:50:11] <alex_joni> sorry.. no idea
[19:50:28] <alex_joni> load_program() maybe? :D
[19:50:46] <motioncontrol> ok thanks
[19:51:35] <geo01005> Thanks for the help guys.
[19:51:36] <alex_joni> motioncontrol: I don't know .. and no AXIS sources around right now to look, but I'm sure a grep on load would be conclusive
[19:51:47] <alex_joni> geo01005: bug me when you need help with kins
[19:51:58] <geo01005> Ok, sounds good.
[19:52:05] <motioncontrol> ok thanks
[19:52:15] <alex_joni> I am planning on building a sim for another bot..
[19:52:22] <alex_joni> I should probably document all steps
[19:56:22] <geo01005> That would be helpful.
[20:03:39] <motioncontrol> alex_joni, please have a example in python for open a file parprogram and load in emc2 ?
[20:36:07] <numen> re
[20:52:39] <ries> numen: I could find a second hand PIII here in Ecuador.. however I am highly likely going to miami next week, so I just might pick up something there and bring it back
[20:59:17] <MrSunshine> http://benchmark.20m.com/reviews/TritonRouter/TritonRouter.gif can i take the motor out of something like that and put on my mill to up the RPMS? :)
[20:59:20] <MrSunshine> its a sieg X1
[21:00:25] <Casainho> hello :-)
[21:02:02] <Casainho> I have some code here, made by other developer. It have a custom UI for EMC2 and some python code... the problem is that I can't see the pins that python code updates, updated on pyvcp... does any one knows what can be the problem? I can past bin the small code...
[21:02:50] <frallzor> yo yo
[21:02:55] <Casainho> I can even see in EMC2 the hal pins and I see the outputs from Python code at one state and on the pyvcp with other state, it's like if net connections are missing...
[21:04:54] <frallzor> lo ries =)
[21:08:24] <ries> Hey frallzor
[21:08:26] <ries> How is your hand?
[21:08:41] <frallzor> still treating it :P
[21:08:49] <frallzor> no fun just work atm :
[21:08:57] <frallzor> or well school :P
[21:11:41] <frallzor> but I relized my rails arent that shitty, just unevenly ground
[21:11:47] <frallzor> due to no control of angle
[21:11:53] <frallzor> so im looking for a spirit level
[21:15:29] <ries> I just ripped a spirit level from a old angle (cheapo)
[21:15:32] <ries> that worked wuite nice..
[21:15:44] <ries> you will still wiggle a bit, but you get quote nice results...
[21:15:59] <ries> and what I said on teh forum, try not to grind to much in one run and grind slowly
[21:17:07] <frallzor> yeah I did go all the way down
[21:17:17] <frallzor> so I guess i can practice on the failed ones
[21:20:56] <ries> frallzor: what you can do is cut them off 2 or 3 mm
[21:21:01] <ries> then re-grind them all over
[21:21:07] <ries> this way you don't have to throw anything away
[21:21:18] <ries> shoulds be possible I think..
[21:21:27] <ries> We didn't cut teh rails and they are still at 60mm...
[21:21:39] <frallzor> I did cut
[21:21:41] <ries> so I can practice quite a bit more :)
[21:22:44] <frallzor> pretty sure the cut is needed to the pinions can reach the racks
[21:22:47] <frallzor> *so
[21:27:18] <andypugh> Does a Mesa card remeber the config after power-down?
[21:27:43] <cradek> no, the firmware is uploaded anew every time you start emc
[21:28:25] <andypugh> If not, how should one ensure that the output pins are in a defined state? (I suppose I am asking what value of pullup / pulldown would be appropriate)
[21:29:06] <cradek> the watchdog handles that for you. you should set up the hardware so your safe defined state is the one that you get when the watchdog bites
[21:29:35] <cradek> I believe the watchdog turns all outputs to inputs so they are pulled neither low nor high.
[21:29:44] <andypugh> The pins all go open-drain don't they?
[21:29:46] <cradek> (but check the man page)
[21:30:14] <cradek> man hostmot2, look for section Watchdog
[21:30:22] <cradek> I see that I am slightly wrong :-)
[21:30:59] <cradek> you should make it so this causes an estop, and the estop should already take care of putting everything into a safe state
[21:31:17] <andypugh> I just accidentally hit the "clear" button. What was that previous sentence?
[21:31:21] <cradek> then your machine can not come out of estop until the mesa is programmed and working properly
[21:31:24] <cradek> then your machine can not come out of estop until the mesa is programmed and working properly
[21:31:28] <cradek> you should make it so this causes an estop, and the estop should already take care of putting everything into a safe state
[21:31:45] <cradek> man hostmot2, look for section Watchdog
[21:34:05] <andypugh> Do I believe man/man9/hm2_7i43.9.html or man/man9/m7i43_hm2.9.html
[21:34:22] <cradek> good question
[21:34:32] <cradek> I think man hostmot2
[21:35:16] <andypugh> I was just commenting that the man pages have two versions of the driver documented.
[21:36:19] <PCW> andypugh: the I/O lines are pulled high at startup (3.3K on 5I20,5I22,5I23, FPGA pullups on 7I43 = also about 3K)
[21:36:41] <andypugh> So, they get pulled high by default, which will turn on the connected ULN2003
[21:36:48] <andypugh> Which is bad...
[21:37:10] <andypugh> Can I pull specific lines down?
[21:37:53] <andypugh> Or should I put some inverters on the breakout board?
[21:38:27] <Casainho> andypugh: I found what was my probelm of yesterday :-) -- my .hal file was not being reading, because I commented it, so, I ddn't had any "signals" fron IN to OUT pins and vice-versa
[21:38:31] <cradek> you should have an estop loop that makes this question irrelevant
[21:39:10] <andypugh> Yes, I should. Now given that I haven't...
[21:39:23] <Casainho> andypugh: and turns out that my .hal file was "crashing" EMC2... and I found that 2 lines have something wrong... and there is no problem to me since they are related to some motor that I don't have :-)
[21:39:44] <Jymmm> andypugh: ... you stop and redesign something far safer.
[21:39:46] <cradek> make it so it doesn't matter what the ULN is doing, when the machine is estopped. then make it so you can't come out of estop until the mesa is programmed and configured and the watchdog is watching
[21:40:06] <andypugh> Casainho: I did query whether you had left some lines commented out that you mentioned.
[21:40:11] <cradek> now given that you haven't, that is the way to proceed :-)
[21:40:26] <PCW> I would put inverters. The I/O is designed around the I/O module rack which needs high for "off"
[21:40:28] <PCW> It is possible to add pulldown resistors but then you would need to
[21:40:29] <PCW> 1. Disable the preconfig pullups (a jumper on the 7I43)
[21:40:31] <PCW> 2. make a new bitfile with pulldowns on the I/O pins
[21:41:26] <Casainho> andypugh: thanks, I will now continue with my task, that is trying to control a plastic extruder from EMC2 UI :-)
[21:42:02] <PCW> (the existing 7I43 configs put pullups on all I/Os)
[21:42:02] <andypugh> The E-Stop chain looks like a bit of a catch-22.
[21:43:13] <andypugh> I have plenty of channels on the ULN2003, I can wrap some round to give me inverted logic.
[21:43:28] <roh> Casainho i have that part working basically
[21:44:10] <andypugh> Part of the reason I converted to the 7i43 was to have some pins to link the software e-stop to.
[21:44:28] <Casainho> roh?
[21:44:34] <frallzor> oh anyone know of a nice bob that does 0-10V analog?
[21:44:47] <Casainho> roh: what you mean? are you working also on EMCRepStrap?
[21:44:50] <roh> Casainho reading reprap style gcode with some m-stuff
[21:45:16] <roh> Casainho with emc. yes. my pid loop for the heater still needs a proper D-gain, but the rest works
[21:45:26] <andypugh> I am going round in circles rather wondering how I can bring the 7i43 up without it instantly triggering an e-stop and shutting down..
[21:46:08] <Casainho> roh: did you saw the page? http://objects.reprap.org/wiki/Builders/EMCRepStrap
[21:46:36] <roh> yes.
[21:46:54] <roh> but i do not use any reprap electronics at all
[21:47:06] <roh> besides a power-amp
[21:47:15] <roh> no gcode in any arduino here.
[21:49:38] <Casainho> roh: yes, me and Sam Wong also not!
[21:49:46] <roh> Casainho i tared my configs: http://yamato.hyte.de/tmp/repstrap_roh.tgz
[21:49:51] <Casainho> are you sharing your code files, or pictures, or something?
[21:49:51] <roh> also got a nice pyvcp ui
[21:50:06] <roh> will do a full documentation and open the svn soon...
[21:50:16] <Casainho> please do it!
[21:50:27] <roh> you can see the extruder here: https://m21.hyte.de/blog
[21:50:36] <andypugh> Anyway, on the subject of E-Stops, would it be fair to say that the only danger on stepper machine is the spindle motor? Part of the rationale behind this rebuild was to have the VFD powered through an EMC-controlled relay.
[21:50:41] <roh> the tarball basically contains my current state of svn for the 2 dirs
[21:50:55] <Casainho> and I think you should document in RepRap oficial site! or at least goind there and add a link to some of your site...
[21:50:57] <Casainho> http://objects.reprap.org/wiki/Builders/EMCRepStrap
[21:51:23] <roh> andypugh no. the steppers on a serious machine have enough force for serious injuries
[21:51:47] <andypugh> Unfortuntately, without a relay the VFD is going to power up at full speed the moment the 7i43 sees power. Which even I can see is unwise.
[21:52:47] <roh> andypugh why dont you use a chargepump circuit and and-gates to all your inputs?
[21:52:48] <andypugh> roh: Yes, but if the machine is sending steps, then it should also be monitoring the e-stop inputs? ie no need for a dedicated power disconnect on the amps?
[21:52:52] <roh> thats what i did
[21:52:54] <cradek> is this a desktop or a full size machine tool?
[21:53:06] <roh> andypugh the estop is fed back
[21:53:08] <andypugh> In-between.
[21:53:11] <Casainho> roh: thanks!! I will look at your files later and read your code ;-) -- many thanks :-)
[21:53:14] <ries> andypugh: on teh subject of E-Stop, IMHO a E-Stop should power of the complete machine... no power on anything VFD, Steppers, BOB, Drivers etc...
[21:53:19] <skunkworks_> even a desktop machine is going to have enough force to smash a finger/hand
[21:53:24] <cradek> if it's strong enough to cut metal I bet it's strong enough to smoosh a body part
[21:53:27] <andypugh> It's a 7" lathe with a milling head.
[21:53:30] <roh> Casainho the most ugly part is 'waiting for the temp' on M108
[21:53:41] <roh> basically it polls the hal in 1 second loops from bash
[21:54:05] <Casainho> roh: did you use Python?
[21:54:24] <andypugh> roh: I was going to let the 7i43 watchdog achieve much the same thing.
[21:55:08] <Casainho> roh: you may read that page about EMCRepStrap, there are some good ideas for future, like placing all the logics on Python instead of that AVR (Arduino)...
[21:55:18] <andypugh> ries: I can see your viewpoint, but wonder if it might be better to have the spindle stop under power.
[21:56:00] <ries> andypugh: I do wire my E-Stop through my main relays... so my whole control cabinet is powered down
[21:56:09] <ries> I will have seperate program start/stop buttons of course
[21:57:00] <andypugh> That gives me the problem I mentioned earlier though, as soon as I power up the cabinet, the 7i43 will go into standby mode, and trigger the e-stop.
[21:57:34] <Casainho> roh: you just used some normal copper wire for heater? does it work?
[21:57:46] <ries> seems like you need to have a delay somewhere... so you don't enter that condition, bear in mind... I have no idea what a 7i43 is :)
[21:58:18] <Jymmm> ries: basically a i/o board
[21:58:32] <andypugh> It's an IO card, connects to the P-Port and gives you 48 IO channels and onboard quadrature decoders and step generators.
[21:58:41] <ries> Ahh ic...
[21:59:40] <ries> personally.. savety goes first and E-Stop should power off everything... unless it's a very small machine, but it sounds like you need to prevent the 7i43 going into that mode.
[22:00:17] <Jymmm> So what if it triggers E-Stop, you just reset the estop chain
[22:00:21] <andypugh> That "mode" is what happens when the watchdog bites, so actually what I need to do is make everything safe in that mode.
[22:00:48] <andypugh> Jymmm: And then the 7i43 will power up again, and trip the estop
[22:01:04] <roh> Casainho i used resistor wire
[22:01:07] <Jymmm> When you bring up a system, you WANT the estop chain triggered, and manual reset.
[22:01:35] <Jymmm> Imagine a power failure, then the power comes back.
[22:01:43] <roh> Casainho my logic is completely in the emc hal. i only use one arduino like in the configs as 'gpio box'
[22:02:09] <roh> http://axis.unpy.net/01198594294
[22:02:34] <Casainho> roh: are you taking digital and analog signals to EMC?
[22:02:44] <roh> exactly that (to measure the analog output of the sensor)
[22:02:44] <andypugh> Jymmm: Good point, that is probably the best place to start thinking.
[22:02:51] <roh> Casainho exactly
[22:03:05] <Jymmm> andypugh: just get a safety relay and go from there
[22:03:45] <roh> Casainho i have a hal-component which does the ntc-calibration shit (make the 'curve' meet the reality more
[22:04:07] <roh> in the tarball in arduino_emc-hal/reprap_temp.py
[22:05:05] <andypugh> Right, so, let me run a scheme past you all. 24V power and 12V power through a safety relay, but 5V power permanent-on. That relay can only be closed by a signal from the 7i43, which clearly can only produce that signal after its has downloaded firmware etc.
[22:06:07] <Jymmm> andypugh: Dude, take the PC out of the picture completely in respect to E-Stop chain.
[22:06:26] <andypugh> I think I just did.
[22:06:35] <Casainho> roh: ook, good work but that can be also used by many others! just please document it ;-)
[22:07:24] <roh> Casainho will do as soon as i like it myself ;)
[22:07:37] <roh> need to build a faster extruÂder
[22:08:35] <andypugh> And when you say "safety relay" you mean something like http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=0492092 ?
[22:08:57] <ries> andypugh: I agree with Jymmm , E-Stop chain should be hardware only...
[22:09:08] <andypugh> Which costs more than the PC and the motors and drives together?
[22:09:35] <roh> Casainho thats also why i chose to used jeff eplers arduino stuff.. to make it 'most easy' to replicate for somebody else
[22:10:06] <roh> usually i do everything 'the direct way' .. not caring if somebody else could replicate the same materials and or needs special tools or knowledge
[22:10:26] <Casainho> roh: I am just adding that link to EMCRepStrap page. You could add yourself and the documentation to your work...
[22:10:29] <roh> the drivers are connected to the second parport
[22:11:00] <roh> Casainho i think i will put all documentation in our wiki, and link it from there then
[22:11:22] <Jymmm> andypugh: Look, it's E-Stop as in EMERGENCY STOP, not "oh shit", or "I fucked up button". It's to prevent/stop the loss of life or property, What if aneice of nephy put their hand in the wrong place at the wrong time do yu really want to take the chance that a computer is going to do the right thing at the right time?
[22:11:39] <Casainho> roh: well, I understand.. that's why I am documenting since I can do it, because I am follwoing the steps of others... documenting also helps me, not just for others ;-)
[22:12:29] <Casainho> roh: good idea!!! EMC is great, and looks also the Python is powerfull... I am learning a lot with this tools ;-) thanks ;-)
[22:12:39] <Jymmm> andypugh: I dont give a damn if you lose position or fucks up the workpiece, that's not what E-STOP is meant for, even though ppl treat it that way.
[22:13:09] <andypugh> Calm down old chap.
[22:13:32] <Jymmm> andypugh: If you want a PAUSE button/feature/functionality, you add that, above and beyond E-STOP, but dont treat them both the same as they are not.
[22:13:35] <fenn> relays don't have to be expensive
[22:13:44] <SWPadnos> of course they do!
[22:13:54] <andypugh> I was trying to think of a way to ensure that having a valid config in the interface board was _part_ of the e-stop chain
[22:13:58] <fenn> * fenn puts SWPadnos in charge of the regional sales force
[22:14:02] <SWPadnos> well, maybe not relays, but contactors should be *very* expensive
[22:14:41] <Jymmm> andypugh: Then get the E-STOP chain working PROPERLY first, then worry about add-ons.
[22:15:13] <SWPadnos> andypugh, that would be good, if the 7i43 would forget its config when EMC2 shuts down, but it doesn't
[22:15:14] <andypugh> ie, nothing can power up until the e-stop switches are all closed, and there is a "ready now" from the software
[22:15:32] <andypugh> The manual talks of writing the config to eprom.
[22:15:41] <SWPadnos> so EMC2 can completely crash, the CPU locked up beyond recognition, and the 7i43 will keep its program until it loses power
[22:16:47] <SWPadnos> but you do raise an interesting idea - there's no reason why an output on a Mesa card shouldn't be (able to be) dedicated as an output meaning "the PC is alive and well"
[22:16:55] <SWPadnos> the Pico Systems boards have this
[22:17:09] <roh> Casainho you can link https://m21.hyte.de/wiki/RepStrap on reprap
[22:17:15] <andypugh> And, to wind back up the thread a bit, it was suggested that the 7i43 in boot-up state should trigger the E-stop. That wasn't my suggestion, it was what I was told should happen.
[22:17:30] <roh> Casainho that url stays static and i will make the documentation nice there
[22:17:33] <SWPadnos> I don't know that that's true
[22:17:38] <roh> (and may change some links)
[22:17:59] <andypugh> Isn't that exactly what the watchdog does? Any driven-low input effectively is that.
[22:18:14] <SWPadnos> the 7i43, in a "just applied power to it" state, should have all of its outputs tri-stated
[22:18:19] <SWPadnos> well, sort of
[22:18:36] <SWPadnos> but I don't know that the Mesa firmware knows what to output on its I/Os when the watchdog trips
[22:18:44] <SWPadnos> it does shut down all PWMs and stepgens
[22:18:59] <andypugh> All outputs pull high, PWC just said so.
[22:19:05] <SWPadnos> ok
[22:20:07] <SWPadnos> ok, so as long as high means "bad", you're OK
[22:21:29] <andypugh> OK, is killing 24V and 12V power enough for an e-stop? That will kill the steppers and amps, switch off the VFD main power relay and also mean there is no supply to the 0-10V input to the VFD.
[22:21:55] <SWPadnos> it really depends on the machine
[22:22:27] <SWPadnos> killing stepper power could be good, but it could also allow a Z axis to fall
[22:22:32] <SWPadnos> unless there's a brake
[22:22:48] <andypugh> In this case it seems like enough. The only slight concern is that the VFD main power is a SSR rather than contactor.
[22:23:00] <SWPadnos> do you mean that the VFD would be disconnected from the line?
[22:23:27] <andypugh> Partly disconnected form the line, neutral still connected
[22:23:36] <SWPadnos> that's probably enough, unless 12V is needed for anything safety-related
[22:23:43] <SWPadnos> (like the E-stop chain itself :) )
[22:23:50] <andypugh> (And the Z can't drop, even when I want it to)
[22:24:06] <SWPadnos> heh
[22:25:30] <roh> Casainho dumped more of my brain into the wiki
[22:26:08] <andypugh> I would still have the PSU's up, but their feed to the busses would be cut. (except for one feed to the contactor via the 7i43-mediated, boot the drives now, line.
[22:30:36] <andypugh> Oh, I will probably leave the 12V feed to the LCD monitor out of the E-stop chain too.
[22:32:09] <andypugh> What is it that makes a "Safety Relay" cost $200? What do I lose if I want to be a cheapskate and use a standard relay?
[22:34:36] <alex_joni> SWPadnos: this one's interesting: http://excito.com/bubba/products/technical-specifications.html
[22:35:34] <SWPadnos> cool
[22:37:17] <SWPadnos> andypugh, safety relays monitor at least one sensor input, often several, and will shut off if any of the inputs are not in the right state
[22:37:43] <SWPadnos> they are also certified for safety-related duty, which I'm sure adds a crapload to the price, if not the cost
[22:38:28] <alex_joni> andypugh: the only thing you can lose is peace of mind :P
[22:38:47] <alex_joni> or shall I say a piece of mind .. when the machine goes berzerk
[22:38:58] <andypugh> I had plenty of that until you lot started hectoring me!
[22:40:27] <alex_joni> who? us?
[22:40:34] <alex_joni> we're like sheep amongst the wolves
[22:41:51] <Casainho> roh: ok, I will put that link to your wiki page ;-)
[22:44:52] <Casainho> roh: what is your nick?
[22:45:12] <Casainho> roh: raumfahr?
[22:54:59] <danimal_garage> w00t just ordered my 7i33 board and a 7i37
[22:55:23] <danimal_garage> now i'm committed, i gotta convert my mill to servos
[22:56:51] <danimal_garage> well i guess i can always use that board for my lathe too
[22:57:08] <danimal_garage> but i need servos on my mill badly
[23:01:02] <PCW> "SWPadnos: but you do raise an interesting idea - there's no reason why an output on a Mesa card shouldn't be (able to be) dedicated as an output meaning "the PC is alive and well""
[23:01:04] <PCW> Actually any low output bit will do for this (referenced to 7I43s power)
[23:01:05] <PCW> powerup, justconfigured,watchdog bite will all go to all high states on all I/O
[23:01:42] <SWPadnos> so I learned - thanks
[23:02:09] <SWPadnos> is that true for the 5ixx as well?
[23:02:09] <danimal_garage> thanks for the help Peter
[23:02:19] <PCW> yes
[23:03:17] <PCW> Welcome danimal_garage
[23:05:15] <andypugh> OK, main e-stop breaker relay ordered. Coil voltage will pass through e-stop buttons, chuck guard switch, and finally through a ULN2003 line, controlled by an inverted 7i43 pin.
[23:07:09] <PCW> Make sure inverters are powered by same source as 7I43, so when 7I43 loses power you dont get a high output
[23:07:29] <andypugh> And if my imaginary nephew grows up soft because he isn't using a lathe with no e-stops or guards at the age of 7 (like I was) I know who to blame. :-)
[23:08:05] <andypugh> I was going to power it from the VCC pins on the 7i43 header
[23:09:25] <PCW> Right, thats how the I/O module racks work (LED anodes to 5V on pin 49 of cable, cathodes to I/O bits)
[23:10:20] <andypugh> I have assumed that pin-1 is the one next to the moulded triangle on the header?
[23:10:28] <danimal_garage> here's the machine the servos are going on (the one to the right) http://img.photobucket.com/albums/v20/danielwilcox/DSCN0543.jpg
[23:11:25] <danimal_garage> it's enclosed for flood coolant now
[23:11:28] <andypugh> Ah, proper machines, that really need e-stops. I can (just) lift mine. Not that that helps if it is eating your necktie.
[23:12:09] <danimal_garage> e-stops are for quitters
[23:12:18] <danimal_garage> hehe
[23:13:10] <danimal_garage> speaking of, i got my sweatshirt wrapped up in my bridgeport's x axis handle while the power feed was going
[23:13:21] <danimal_garage> that was ugly
[23:13:38] <danimal_garage> those servo power feed units have a little torque behind them
[23:14:50] <danimal_garage> unfortunately i couldnt reach the lever to turn off the feed so i had to grab the handle and try to stall it
[23:14:57] <danimal_garage> till the breaker poped
[23:24:46] <numen> is it possible, to build only with 3 tr16x4 a linear move?
[23:26:08] <roh> Casainho i have no account in the reprap wiki afaik. you can write raumfahrtagentur to it. thats our hackerspace
[23:29:54] <andypugh> danimal_garage: The Bridgeport at work has special handles to prevent that. It is very clever. (Perhaps too clever). If you grip the little in-you-hand bit as you twirl, the friction torque rotates that part, that operates a rod inside the main handle, that engages a clutch in the hub. Very neat, but I ended up dismantling the lever before I figured it out.
[23:31:36] <danimal_garage> what do you guys think about this servo http://cgi.ebay.com/Mitsubishi-HD-40-10SR-Servo-Motor-from-Nomura-CNC-Lathe_W0QQitemZ120293234077QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item1c0209159d
[23:32:04] <danimal_garage> i can get 3 with matching drives for about 600 or less
[23:32:19] <awallin_> what power are those?
[23:32:23] <danimal_garage> for my shizuoka
[23:32:31] <danimal_garage> 4000watt is all it says
[23:32:47] <danimal_garage> i still gotta do a little research
[23:33:35] <danimal_garage> 200v
[23:34:08] <andypugh> You don't want to try stalling that with a sweatshirt
[23:34:17] <danimal_garage> hahaha
[23:35:03] <danimal_garage> no handles on my shizuoka to catch my shirt
[23:36:55] <danimal_garage> hmm i cant find any info on them online
[23:57:07] <jt-plasma> anyone got any ideas on how to ftp a live web cam to a web page? I have the web page set up, I just need a hint on how to send the image on a regular basis
[23:57:27] <jt-plasma> http://suburb.semo.net/jet1024/Hardinge/hardinge.html