Back
[00:01:13] <tomp2> http://tech.blorge.com/Structure:%20/2007/11/15/the-ultra-portable-mac-is-here-leopard-on-asus-eeepc/
[00:03:58] <dmess> hi all
http://www.geekologie.com/2008/03/post_10.php
[00:04:30] <dmess> could emc2 handle that???
[00:08:21] <tomp2> dmess: i think hal could, but orocos may be better suited ( not emc2, what gcode is there to a dog walking?)
[00:09:02] <tomp2> impressive! scary looking beasty
[00:11:49] <toastydeath> how would you put EMC on a planer
[00:12:14] <toastydeath> it's like a lathe, but the feed has to occur after specific conditions are met
[00:12:42] <SWPadnos> use feed-hold and CL or HAL to implement the logic
[00:13:29] <toastydeath> so like, if my feedrate was .050" per stroke, there IS provision in hal to hold the feed until a stroke is complete?
[00:13:43] <toastydeath> index by .050, and let the table go again?
[00:14:17] <SWPadnos> there's a feed-hold pin, which will stop motion
[00:14:27] <toastydeath> hmm
[00:14:33] <SWPadnos> you can hook whatever you want to that, so theoretically yes, you can make something to do that
[00:15:02] <toastydeath> but something seems like the feed would not work "properly"
[00:15:13] <toastydeath> since it's more of a g0 rather than a feed
[00:15:59] <archivist> still need to stroke at the correct speed
[00:16:18] <SWPadnos> do you actually want to have the workpiece jerk forward little bits, or have it run synchronized to the planer blade?
[00:16:43] <toastydeath> no, the workpiece doesn't move - the whole table needs to complete a stroke
[00:16:57] <toastydeath> well it does move, but it's the "spindle"
[00:17:05] <toastydeath> the table needs to stroke once, then feed by the increment
[00:17:15] <SWPadnos> ok, so you're moving the tool, a bit like a wide horizontal mill?
[00:17:25] <toastydeath> moving the work, under the tool
[00:17:40] <toastydeath> the work is the y axis, which is a dumb axis (controls SFM)
[00:17:47] <SWPadnos> yes, the table moves, once it gets to the end, you move it back then index the tool down a notch
[00:17:51] <toastydeath> yep
[00:17:53] <toastydeath> or across, or whatever
[00:17:56] <SWPadnos> sure
[00:18:05] <SWPadnos> typewrites, planer, what's the difference? :)
[00:18:10] <SWPadnos> typewriter
[00:18:13] <toastydeath> exactly!
[00:18:29] <SWPadnos> I'd do a subroutine
[00:18:44] <tomp2> call it clapper()
[00:18:46] <SWPadnos> possibly even lift the blade a little for the "return"
[00:18:56] <toastydeath> there's a hydraulic tool lifter on this machine
[00:19:16] <SWPadnos> and?
[00:19:29] <SWPadnos> is it "click relay and tool goes to full up position"?
[00:19:39] <toastydeath> no, it springs out and kicks the tool back
[00:19:41] <toastydeath> so it clears the work
[00:20:03] <SWPadnos> automatically when you hit the end of travel?
[00:20:04] <toastydeath> it's movable independantly of the rail head so that you can pick which direction the tool clears in
[00:20:09] <toastydeath> so you can undercut, etc
[00:20:13] <SWPadnos> ok
[00:20:24] <SWPadnos> so you don't need to worry about that with the subroutine
[00:20:29] <toastydeath> yar, just the feed/etc
[00:20:51] <SWPadnos> there are several examples of subroutines (like the face.py ones) to go to a specified depth in specified increments
[00:21:07] <SWPadnos> it seems like a pretty easy subroutine to write
[00:21:25] <toastydeath> hmm
[00:21:31] <toastydeath> it is good to know it should be doable.
[00:21:45] <SWPadnos> ("index" down, move across Y, wait a little for head to get out of the way, move back at G0, index down, move at feed ...)
[00:21:48] <BigJohnT> YEA face.py
[00:21:51] <SWPadnos> heh
[00:22:02] <SWPadnos> only thing I could think of on short notice :)
[00:23:06] <tomp2> shaper = move tool, planer = move workpiece ?
[00:23:12] <toastydeath> tomp2: yes sir
[00:23:24] <toastydeath> http://www.surplusrecord.com/listphotos/517242.jpg
[00:23:34] <toastydeath> that's essentially the machine except someone took the second railhead off the one we have
[00:24:57] <tomp2> AGie used to use planers on machine tool bases, they said rotating tools caused a lot of pent-up surface torsion waiting to be released on temperature change.
[00:25:07] <toastydeath> yep
[00:25:10] <BigJohnT> wow that's big
[00:25:22] <toastydeath> also milling causes cold working
[00:25:31] <toastydeath> it does a lot of nasty stuff to surfaces
[00:26:07] <BigJohnT> looks like XYZ basicly to me
[00:26:17] <tomp2> yepp, like that pic and it screeched like crazy
[00:26:34] <tomp2> but bigger and a bridge
[00:26:49] <toastydeath> it kind of is XYZ, if you count an uncontrolled Y axis
[00:26:57] <toastydeath> that sets the "speed"
[00:27:18] <BigJohnT> so the slide the material is bolted to is uncontrolled?
[00:27:26] <toastydeath> yep
[00:27:32] <toastydeath> it's a really, REALLY big hydraulic ram
[00:27:34] <toastydeath> under the table
[00:27:45] <archivist> could be controlled if you wanted
[00:27:50] <BigJohnT> it just goes back and forth and you "know" when you are on each end of the stroke
[00:28:03] <toastydeath> it has been controlled but you lose the "planer" ability
[00:28:08] <toastydeath> and basically get "a milling machine"
[00:28:23] <BigJohnT> how is that?
[00:28:24] <toastydeath> top speed on that table is 300 feet/min
[00:28:49] <toastydeath> so 3600 ipm
[00:28:59] <BigJohnT> smokin
[00:29:30] <BigJohnT> so you Z down to the cut level and X across after each stroke?
[00:29:37] <toastydeath> yep
[00:29:56] <tomp2> run it open loop with an amp using current control ( sorta like hydraulic pressure ), just use limit swx & hal... hmmm or limit numbers
[00:30:17] <toastydeath> the problem is that the table is 5000-6000 lbs
[00:30:29] <toastydeath> and a ballscrew just doesn't like moving something that heavy very quickly
[00:30:42] <tomp2> use hall to control the sppol valve on the hydraulics
[00:30:45] <tomp2> hal
[00:31:36] <tomp2> moog/parker/pegasus valve
[00:31:40] <BigJohnT> isn't all you need to "know" is you reached the end of the stroke and do something n number of times?
[00:31:44] <toastydeath> bigjohnt: yep
[00:32:20] <BigJohnT> a loop in G code with an input like I use for my plasma torch
[00:32:22] <toastydeath> it needs to index by the feed in whatever direction it is moving.
[00:32:25] <BigJohnT> hmmm what was that
[00:33:13] <tomp2> bbl
[00:33:33] <BigJohnT> * BigJohnT looking
[00:35:21] <BigJohnT> hmmm I think it was M66 it waits for an input before going to the next line in the code
[00:35:28] <BigJohnT> yep that's it
[00:36:18] <BigJohnT> as soon as you see an M66 input you move the X then wait for the next M66
[00:36:44] <toastydeath> i guess
[00:36:51] <toastydeath> that still seems awfully sketchy
[00:37:18] <BigJohnT> why is that?
[00:37:41] <toastydeath> i was kind of hoping to make it behave, natively, like a lathe
[00:37:52] <toastydeath> except instead of a revolution, you had a stroke
[00:38:31] <BigJohnT> but you don't want the x moving while you are stroking right?
[00:40:26] <toastydeath> bigjohnt:
[00:40:30] <toastydeath> correct, i want motion to STOP
[00:40:35] <toastydeath> and only to go when the table is latched
[00:40:52] <BigJohnT> that is the M66 g code
[00:41:29] <BigJohnT> does the table stay at the end of the stroke until you command it to go again?
[00:42:32] <toastydeath> no
[00:42:42] <toastydeath> the table would not be under control
[00:42:56] <BigJohnT> it just cycles back and forth uncontrolled
[00:42:58] <toastydeath> yep
[00:43:23] <BigJohnT> still doable
[00:43:43] <BigJohnT> if you have time to move the x after you are past the material
[00:44:08] <toastydeath> is there any way to do it in a script or something rather than relying on g-code
[00:44:21] <BigJohnT> well I have to go make some pizza's for the wife and me
[00:44:27] <toastydeath> because it would be super cool if i could just say "g1 f.010 x10 z0
[00:44:31] <toastydeath> cool dude, have fun
[00:44:49] <BigJohnT> I don't have any idea about that someone else might chime in
[00:45:27] <BigJohnT> as in cut 0.010 wide cuts for 10" at Z0
[00:45:41] <BigJohnT> * BigJohnT hears hungry wife
[00:45:46] <BigJohnT> bbl
[01:00:21] <dmess> tomp.... but the motion control... and feedback....coulnd it be handled...
[01:03:08] <dmess> need and end of M code signal
[01:04:20] <dmess> toast see above
[01:34:32] <tomp> dmess, the motion control in 'big dog' isnt like gcode, it's like animatronics, and orocos is better/only suited to robotic motion. there maybe other solutions, but emc2 only has gcode for language as of now.
[01:34:48] <tomp> and making that dog walk with gcode would be , ugly at least
[01:35:59] <tomp> toastydeath: the f.010 is a stepover , or a feedrate?
[01:37:57] <tomp> i think the motion of the planer can be done, but not with 'g1 f.010 x10 z0'. and i do thin it can be 'transparent', but more like a macro call than a gcode
[01:38:24] <tomp> g101 A.010 B10 Z0
[01:38:58] <tomp> g101 A.010 B10 C0 (like a fanucy macro )
[01:39:14] <toastydeath> tomp: the feed rate IS the stepover
[01:39:29] <toastydeath> just like you have inches per revolution on a lathe, it's inches per stroke on a planer
[01:39:36] <tomp> right, not slowing advancing, but incrementally advancing
[01:39:45] <tomp> right, per stroke
[01:39:48] <toastydeath> rite
[01:40:09] <tomp> got a planer to convert ( we gotta get you running emc :)
[01:40:12] <tomp> ?
[01:42:22] <toastydeath> POSSIBLY
[01:42:26] <toastydeath> but it's a slim possibility.
[01:42:30] <tomp> and i would like to see a moog valve and linear encoder axis under emc control.
[01:42:54] <toastydeath> actually if it was Y-controlled that would make it very useful
[01:43:00] <tomp> not much of an amplifier is needed to move that spool ( 25ma :)
[01:43:15] <toastydeath> but again, it has to not only work as a planer, it has to do what we need it to do
[01:43:19] <toastydeath> and those are both way up in the air
[01:43:52] <tomp> np, good to know you asked about emc to do it
[01:45:03] <toastydeath> well i brought EMC up when we did our lathe
[01:45:21] <toastydeath> and they decided to go with some other fuckjob control that is like, no comparison in terms of utility
[01:45:33] <toastydeath> it has some really REALLy nice oddball features but the overall control sucks balls
[01:45:55] <toastydeath> and that lathe is tracking to ~10 millionths of an inch or so
[01:46:02] <toastydeath> and definately positioning under ~10
[01:46:13] <toastydeath> which would be a hot thing to have running EMC
[01:48:54] <tomp> this is funny google result: "The "EMC" Electro Hydraulic Thrustor consists of a cylinder filled with oil containing the centrifugal pump located at the bottom and the electric motor assembled on the top cover."
[01:49:18] <tomp> an Indian company, no separate hyd pump, each cylinder has its own built in motor
[01:49:56] <toastydeath> lol, that's pretty cool
[01:50:43] <tomp> 10 millionths!, whew, better than i ever work over that distance
[01:51:50] <tomp> bbl
[01:53:26] <jmkasunich> regarding the planer discussion earlier:
http://www.youtube.com/watch?v=pzsQnKJGYtk
[01:53:57] <jmkasunich> and
http://www.youtube.com/watch?v=MiBG-Hpd1ZU
[01:55:18] <toastydeath> jmkasunich: yep
[01:57:27] <toastydeath> also tomp that's just what it's tracking to, i doubt it will cut to that
[01:57:50] <toastydeath> it's just cool to be able to watch a .0001" indicator move in those increments
[01:58:02] <toastydeath> well i know it won't cut to that
[01:58:54] <dmess> georgeous hydraulic shaper
[02:01:45] <toastydeath> our machines are very similar, except they're a bit smaller and only have one column
[02:03:06] <dmess> not a single column ... never... maybe .0005" but you would need to know the machine...
[02:03:17] <toastydeath> no, that's for a lathe
[02:03:42] <toastydeath> we have a lathe that has 50 nm scales on it, and it sucessfully tracks at low feed rates to a couple milionths
[02:04:16] <toastydeath> the one planer i am talking about hasn't even been fired up except for when it was first purchased
[02:04:27] <dmess> oh ive seen cnc lathes the would offset to .00001" so i digress
[02:04:48] <toastydeath> offsetting to .00001 and tracking to .00001 are very different
[02:05:01] <toastydeath> ours tracks to .00001
[02:05:31] <dmess> nice.... no bubble to burst..
[02:06:20] <dmess> climatly controlled enviro???
[02:06:59] <dmess> measuring systems??
[02:07:14] <toastydeath> i said it tracks to .00001, which is completely independant of all those items
[02:07:23] <toastydeath> never said it cuts or has an absolute accuracy of .00001
[02:07:28] <toastydeath> we're still building it
[02:07:37] <tomp> jmkasunich: those vids are more like it, notice the head bobble at end of stroke ?
[02:07:46] <jmkasunich> yep - tool lifter
[02:07:56] <dmess> good for you... i like to see it done.. i might pop 1
[02:08:57] <dmess> as i said i am not bashing.... anyone...
[02:09:48] <tomp> toastydeath: 'tracks to .000010' meaning straightness? i understand its not position, but even my best machines vary more than 10uInch ( so sayeth the HP laser ;)
[02:10:00] <dmess> the one i spoke of i cut parts on in 1998 ish
[02:10:06] <toastydeath> no, tracks, as in follows the scales
[02:10:46] <toastydeath> the dude doing most of the work keeps ratcheting the max error up and it keeps still working
[02:11:00] <toastydeath> should be able to do some darn fine profiling on it
[02:11:46] <dmess> ok question of the night..... what is more accurate a laser or autocalimeter????
[02:12:15] <toastydeath> laser
[02:12:37] <dmess> long machines... 30-50 M
[02:12:43] <toastydeath> laser in a helium vacuum.
[02:13:04] <toastydeath> at any distance.
[02:13:12] <toastydeath> unless you have an old, crappy laser.
[02:13:21] <dmess> auto-cal....any day..
[02:13:26] <toastydeath> sorry, no.
[02:14:26] <dmess> quite new renishaw unit.... i saw its error in 15 minutes..
[02:15:06] <toastydeath> fancy, but i'd be willing to bet money that it was operator or environmental error
[02:15:19] <toastydeath> since nobody uses autocollomators anymore for ultraprecision machining
[02:15:57] <toastydeath> there are more factors that go into laser accuracy than an autocollamator but the potential for a more accurate measurement is higher.
[02:16:07] <dmess> i wont rule that out.... but that what i read... and i see alot with that thing...
[02:17:11] <tomp> Autocollimator, like long distance optical flats, i couldnt find it googling :)
[02:17:25] <tomp> http://en.wikipedia.org/wiki/Autocollimator
[02:18:02] <toastydeath> i am absolutely willing to rule out an autocollimator for high accuracy angular measurement
[02:18:35] <tomp> http://www.micro-radian.com/Principles.html
[02:18:46] <tomp> micro radian heh
[02:19:12] <dmess> oh welll... whatever
[02:20:28] <dmess> ours was ALL optics.... no video.... read to microns
[02:20:42] <toastydeath> righto and lasers are measuring in nanometers
[02:21:03] <toastydeath> between the two reflectors
[02:21:21] <toastydeath> you want higher accuracy in tilt, spread the reflectors out
[02:21:43] <dmess> with 30 % diffraction at nano stroke.. good luck for detail...
[02:22:17] <toastydeath> you don't have to argue with me, go argue with the people doing it.
[02:22:52] <tomp> is a planer with a rotary table a turret mill?
[02:23:05] <toastydeath> tomp: a turret mill?
[02:23:10] <dmess> we set up 5 axis machines with indicators and ball bases... for kinematics
[02:23:33] <tomp> dunno, saw something like a planer with a rotary table doing tubine rings ( mixers )
[02:23:55] <dmess> cause the laser set up was never right..
[02:23:55] <toastydeath> was it a planer or was it a planer mill
[02:24:21] <toastydeath> dmess: i hate to say this but machine setup and verification is like the lowest form of precision measurement i can think of
[02:24:38] <toastydeath> and is not a good place to be comparing "the best"
[02:24:48] <dmess> i figure the highest...
[02:25:17] <toastydeath> lowest, because the most accurate machines correct for misalignment dynamically and don't care if an axis is off by .001"
[02:26:17] <dmess> mass is your friend at that point
[02:26:34] <toastydeath> tomp: i don't quite understand what you are referring to, is it milling?
[02:26:37] <toastydeath> or single pointing
[02:27:22] <tomp> single point, tool like a lathe bit, work large diameter (50" +), the machine was a bridge frame with a large (72") rotary table
[02:27:35] <tomp> relatively large
[02:27:52] <toastydeath> so it planed, then indexed the table?
[02:28:12] <tomp> no the work rotated rather than indexed
[02:28:21] <tomp> it was a vert lathe of some sort
[02:28:26] <tomp> doh
[02:28:27] <toastydeath> ohhhh a vertical boring mill
[02:28:44] <toastydeath> http://www.summitmt.com/images/tools/Vertical_Boring_Mill_SCVC48.gif
[02:28:57] <toastydeath> probably more like this:
[02:29:00] <toastydeath> http://cbb.ca/cap3L.jpg
[02:29:20] <tomp> yes, door #2 :)
[02:29:39] <tomp> and a few heads , as in that pic
[02:30:03] <toastydeath> just a regular ol vertical boring mill
[02:32:25] <tomp> http://www.centroidcnc.com/verticallathes.htm this one, where the 'floor' is the table, we joked it was like a fun house
[02:34:25] <toastydeath> hahah.
[02:35:08] <tomp> i see where some folks say 'vert turret lathe'
http://en.wikipedia.org/wiki/Image:Rockford1.JPG
[02:35:51] <toastydeath> vtl's and vbm's are pretty much the same thing, one has a turret, the other has a long stroke ram
[02:35:55] <toastydeath> for deep boring
[02:36:00] <toastydeath> versus general turning
[02:36:42] <JymmmEMC> fuuuuck I'm tired.
[02:39:45] <toastydeath> ?
[02:42:15] <JymmmEMC> We have another commercial bldg (half-vacant) on the other side of town that the fire alarm has been fucking up since the 16th. My boss is out of town, so I head over there, no air movement at all, and the alarm panel in a hot clammy electrical room.....
[02:43:45] <JymmmEMC> So I dont have the properly tools with me and trying to troubleshoot where the problem is, takes 3 hours between scouranging around for what I can find to fix it and fighting the dman alarm panel which only tests the lines every 15 minutes so I have 4 chance per hour to see if I fixed it......
[02:44:03] <toastydeath> owned =/
[02:44:41] <JymmmEMC> Called the alarm company a few times to make sure the system is reporting correctly now, have them check the log files, wait another 15 minutes, then have them place the system back into normal mode.
[02:45:57] <JymmmEMC> drive all the way thru rush hour traffic to get past my home to pick up my new glasses, then head home. So it's been a very long day, not to mention one of the sales admin's "venting" for things way before I even started the job and have no knowledge of.
[02:48:46] <toastydeath> =(
[02:50:21] <JymmmEMC> And now, I'm wlaking around all woozy like trying to get used to the new progressive bifocals to top it off =)
[02:55:32] <ds2> nice
[02:55:38] <ds2> no air movement in a DC
[02:56:04] <JymmmEMC> No, not a DC. Another commercial bldg that's half empty
[02:58:16] <ds2> new business of co-locating offices? :)
[02:58:35] <JymmmEMC> I started a new job in Jan.
[02:58:41] <ds2> ohhh
[02:59:06] <ds2> still up in the east bay?
[02:59:17] <JymmmEMC> S bay
[02:59:37] <ds2> oh nice... what do you do now?
[02:59:50] <JymmmEMC> everything =)
[03:00:00] <ds2> facilities mgr?
[03:00:14] <JymmmEMC> IT
[03:00:50] <ds2> wow... didn't know they still have those jobs in the US
[03:01:42] <JymmmEMC> ds2: bite me!
[03:01:54] <ds2> eh?
[03:02:27] <JymmmEMC> ds2: ds2: wow... didn't know they still have those jobs in the US
[03:04:58] <ds2> and I am trying to figure out what would cause a reaction like that
[03:06:12] <JymmmEMC> youre smartass comment of course =)
[03:06:21] <JymmmEMC> -e
[03:08:17] <ds2> uh okay
[03:11:26] <toastydeath> FIGHT
[03:27:50] <ds2> hmmm hex steel bar is not that common
[03:28:37] <jmkasunich> metal express has it in 12L14
[03:29:03] <ds2> oh? onlinemetals/enco/msc didn't have them
[03:29:46] <jmkasunich> 1/4 thru 1-1/2 by 16ths, then up to 2 by 8ths, and up to 3 by 4ths
[03:32:55] <SWPadnos> damn. it's expensive to ship cameras around (if you want them insured)
[03:32:57] <ds2> cheaper then McMaster
[03:33:09] <jmkasunich> metal express usually is
[03:33:17] <jmkasunich> plus you can buy just the amount you need
[03:33:21] <ds2> how are they for shipping?
[03:33:33] <ds2> it seems it want me to fill out the entire checkout form before it'll tell me shipping
[03:33:54] <jmkasunich> probably cause they don't know where you are
[03:34:16] <jmkasunich> hang on a sec, I just bought some stuff from them, lemme find the invoice
[03:34:17] <ds2> so they are not flat rate to the US
[03:34:43] <SWPadnos> with metals being heavy, and bars being odd shapes, I'd be surprised if they could do a flat rate
[03:34:46] <jmkasunich> no, its based on weight and distange (they use UPS ground mostly)
[03:35:03] <ds2> SWPadnos: USPS has a flat rate package that is prefect for metals
[03:35:14] <jmkasunich> I bought 20 pieces of 2" square 12L14, 1.1 inches long, so it was pretty heavy
[03:35:18] <SWPadnos> for short pieces of metals, yes
[03:35:39] <SWPadnos> a 2 foot bar (or even an 18" bar) wouldn't fit in the USPS package
[03:35:49] <SWPadnos> maybe one 18", diagonally
[03:36:25] <ds2> trying to price out a 12" peice of hex bar for a simple plastic injection molder from an old issue of HSM
[03:36:43] <jmkasunich> $64.87 for the steel, 9.33 handling fee (cutting charge mostly), and 12.38 shipping
[03:36:55] <jmkasunich> that was probably close to 30 lbs
[03:37:54] <SWPadnos> 88 cu.in at 0.39lb/cu.in
[03:38:03] <SWPadnos> over 30 it seems
[03:38:12] <SWPadnos> or is it 0.36?
[03:38:30] <SWPadnos> hmmm. yes, 0.36lb/cu.in seems more correct :)
[03:38:37] <SWPadnos> damn. copper is expensive
[03:38:51] <jmkasunich> steel is more like .32-.33
[03:38:54] <jmkasunich> cu is .36ish
[03:39:01] <jmkasunich> the box is marked 26 lbs
[03:39:19] <SWPadnos> hmmm. I thought I remembered copper being about the same as steel - maybe I was thinking of stainless or something
[03:39:39] <ds2> copper bearing stuff shot through the roof this last few years
[03:39:42] <jmkasunich> copper is amazingly expensive
[03:39:54] <SWPadnos> 1x2 x 12in long = $67.14
[03:39:59] <jmkasunich> I just bought some at work for high current bus bars
[03:40:37] <jmkasunich> after seeing prices, I went; back to the design to see how much copper I could take out of it
[03:40:39] <SWPadnos> I wonder what kind of conduction you'd get from copper plated aluminum - assuming you plate before the oxide layer forms
[03:40:53] <jmkasunich> (and this is just a one-time thing - lab test equipment)
[03:40:57] <SWPadnos> heh
[03:41:10] <jmkasunich> about the same as plain aluminum, unless at very high frequency
[03:41:15] <SWPadnos> you should see the thousands of pounds of copper they have at that power supply company
[03:41:16] <ds2> SWPadnos: if it is RF at high enough free, plated would be same as solid
[03:41:21] <jmkasunich> skin depth at 60 Hz is on the order of 8mm
[03:41:54] <SWPadnos> actually, it's at high freq that the plating really matters
[03:42:01] <jmkasunich> I'm glad we did the lab rewire a couple years ago
[03:42:14] <jmkasunich> we bought a roll of 500 MCM that must have weighed over a ton
[03:42:23] <SWPadnos> heh - it would probably cost more to wire than the building itself today
[03:42:51] <jmkasunich> we are giving very strong consideration to aluminum in our new designs
[03:42:56] <jmkasunich> weight and cost both
[03:43:51] <SWPadnos> yeah - the weight is 1/3, the conductivity like 80% or something
[03:43:54] <jmkasunich> that test equipment I'm working on is to test busbar at high voltage and low current
[03:43:59] <SWPadnos> and cost 1/8 :)
[03:44:04] <jmkasunich> more like 60%, but still....
[03:44:25] <jmkasunich> we got a pair of surplus spot-welding transformers
[03:44:39] <jmkasunich> 50KVA (or more) at 7.9V
[03:44:45] <SWPadnos> that sounds high current for some reason :)
[03:45:14] <jmkasunich> we tested it by plugging an ordinary benchtop variac into 120V and hooking to the primary (rated at 460V)
[03:45:28] <jmkasunich> shorted the output with a piece of 10AWG about a foot long
[03:45:53] <SWPadnos> bye bye wire
[03:45:53] <jmkasunich> 10AWG glows quite nicely at 300 amps or so, and the variac wasn't even breaking a sweat
[03:46:07] <jmkasunich> actually I had a heck of a time burning it out
[03:46:19] <jmkasunich> with the low input V, I could only get about 2.5V out
[03:46:25] <SWPadnos> you need to invest in a copper vapor reclamation system ;)
[03:46:26] <jmkasunich> as the wire got hot, its resistance increased
[03:47:02] <jmkasunich> if I turned it up while cold, I could get 500-600 amps for a few seconds, but then as it heated up the current dropped to 250-300, and stabilized
[03:48:49] <SWPadnos> for some reason that reminded me of the current sense "resistor" they were using on one power supply a few years ago
[03:49:12] <SWPadnos> this thing was a bunch of flat plates that were bolted together, with room between them (as a heatsink)
[03:49:25] <SWPadnos> it was actually for "low current" supplies - only 2000A
[03:49:41] <SWPadnos> and at 2000A, it had a potential of 50 mv
[03:49:50] <SWPadnos> which is 100W dissipation
[03:49:53] <jmkasunich> sounds like a shunt
[03:50:11] <SWPadnos> they used a scope with isolated probes to measure directly across it
[03:50:19] <SWPadnos> I think it was in series
[03:51:03] <jmkasunich> like this only bigger?
http://webpages.charter.net/jasonr/pictures/Shunt5.jpg
[03:51:29] <SWPadnos> yes, pretty much
[03:53:08] <ds2> isn't a hall effect sensor better at those currents?
[03:55:36] <tomp> heh, those are like the shunts i use 300A = 50mV but i think the picture is a 600A
[03:57:06] <tomp> pulsed dc current is hard to measure, a shunt is just for a meter ( for a slow averaging device )
[03:58:44] <ds2> shouldn't a hall effect sensor be able to do as well?
[03:58:53] <jmkasunich> shunts are simpler
[03:59:11] <jmkasunich> hook a 50mV analog meter movement across it and you are done
[03:59:24] <jmkasunich> no power supply or anything
[03:59:30] <ds2> hmmm
[03:59:34] <SWPadnos> I think this was a pulse mode supply, and they were looking at milliseconds-long waveforms
[04:00:21] <jmkasunich> hence the scope
[04:00:46] <jmkasunich> for that I would expect a Hall would work as well, or better
[04:01:03] <jmkasunich> although I bet a shunt is more accurate
[04:01:05] <SWPadnos> I don't remember the specifics - that was about 7 years ago
[04:01:22] <ds2> it just seem dangerous to use a shunt
[04:01:34] <jmkasunich> Halls are a percent or two, and often have some DC offset, they can also vary a bit with the placement of the current in the window
[04:01:35] <tomp> we had problems with the mfctr mismatching the ammeter and the shunt, it always read high, made his stuff look good, but was a sham
[04:01:51] <tomp> the shunt is just in series with the load
[04:01:57] <tomp> no danger
[04:02:05] <ds2> yes, it is not galvanically isolated
[04:02:14] <SWPadnos> hence the isolated csope
[04:02:15] <jmkasunich> its a danger if the load is more than a few volts above ground
[04:02:18] <SWPadnos> scope
[04:02:18] <tomp> it appears to be a very small resistance
[04:02:24] <jmkasunich> Halls have isolated outputs
[04:02:29] <jmkasunich> tomp: yes, they do
[04:02:50] <SWPadnos> it's too bad, you pretty much can't buy an isolated scope these days, except for the little handhelds
[04:02:57] <jmkasunich> a 500A -> 50mV shunt is 0.050/500 = 0.1 milliohms
[04:03:04] <SWPadnos> LeCroy used to have that as an option, but not any more
[04:03:07] <SWPadnos> he
[04:03:09] <SWPadnos> h
[04:03:16] <jmkasunich> SWPadnos thats cause people manage to kill themselves with them
[04:03:16] <ds2> til it opens and you get the full voltage on you
[04:03:32] <SWPadnos> sure - you have to know what you're doing
[04:04:35] <tomp> the shunt open? its more likely the wiring to it would open, it's like bus barr
[04:16:22] <JymmmEMC> LawrenceG: jmkasunich you guys have a schematic for a chargepump?
[04:17:18] <jmkasunich> not really
[04:17:30] <JymmmEMC> k
[04:17:47] <jmkasunich> its a simple circuit, simple enough that I design it when I need it, but don't tend to write it down
[04:17:57] <jmkasunich> if it wasn't after midnight I'd give it a shot
[04:18:02] <SWPadnos> heh
[04:18:14] <JymmmEMC> jmkasunich: seriously, dont sweat it =)
[04:18:30] <jmkasunich> they tend to be designed around whatever chips and parts I have on hand, so everybody's circuit is differnent
[04:18:35] <SWPadnos> heh
[04:18:44] <SWPadnos> diodes + capacitors, baby! :)
[04:19:02] <jmkasunich> yeah, but something has to turn that into a nice solid 1 or 0
[04:19:05] <SWPadnos> and a comparator, and a schmitt trigger and ... :)
[04:19:32] <jmkasunich> I've used a transistor, or a '14 schmidt inverter, or a 339, or....
[04:20:15] <jmkasunich> I think the one in the shoptask is a '04 driving the diode/cap, with a transistor on the output that turns on a relay
[04:21:39] <SWPadnos> looks like a '14 in the photo ;)
[04:22:10] <SWPadnos> http://willepadnos.net/jmkasunich/perfboard-front-1779m.jpg
[04:22:50] <jmkasunich> ah, right
[04:23:05] <jmkasunich> thats taking the parport signals and driving the pump cap
[04:23:14] <jmkasunich> the output of the pump is driving a mosfet
[04:23:29] <jmkasunich> (that SIP part below the relays is a 4x mosfet array)
[04:23:48] <SWPadnos> oh - odd form factor
[04:24:03] <jmkasunich> I scrounged a tube of them a couple years ago
[04:24:45] <SWPadnos> those ivory chips look like old BOSFETs
[04:24:53] <jmkasunich> they're optos
[04:25:00] <SWPadnos> ok - about the same
[04:25:12] <jmkasunich> three dig-ins on the bottom right connector
[04:25:25] <SWPadnos> I think the BOSFET was an isolated bidirectional-output switch
[04:25:31] <SWPadnos> ah
[04:25:34] <jmkasunich> two 120V relay outs on the two right connectors (parport pin -> '14 -> mosfet)
[04:25:52] <jmkasunich> the charge pump used a third fet, open-drain, to drive an external relay
[04:26:45] <SWPadnos> and one spare ;)
[04:27:09] <jmkasunich> yeah
[04:28:15] <jmkasunich> woo-hoo - part #4 of 20 done
[04:28:26] <jmkasunich> first one that didn't involve program tweaking
[04:29:01] <jmkasunich> time for bed
[04:29:07] <SWPadnos> heh
[04:29:16] <SWPadnos> should be quick tomorrow then
[04:44:11] <Guest971> who is on
[04:44:25] <Guest971> i need the commprotocol of the ppmc
[04:46:05] <Guest971> anyone is here?
[04:46:35] <SWPadnos> what comm protocol?
[04:46:49] <SWPadnos> there is a register set definition at the pico-systems website
[04:46:56] <SWPadnos> it communicates with the PC via EPP
[04:47:00] <SWPadnos> (parallel port)
[04:51:11] <Guest971> i want to build a board with avr to use istead of the pico
[04:51:48] <SWPadnos> heh - that's a challenge
[04:52:39] <SWPadnos> instead of the stepper controller?
[04:53:06] <Guest971> i just can't buy pico (it is expensive) and far from me (middle east)
[04:53:12] <SWPadnos> hmmm
[04:53:42] <SWPadnos> unless you need much lower performance than the pico board can give, an AVR will not suffice
[04:54:04] <SWPadnos> if you do not need all the functions of the pico boards, then an AVR may be good enough
[04:55:08] <Guest971> why it is not enough?
[04:55:42] <SWPadnos> one of the functions of the pico systems board is to cound encoder pulses. this is something that is very fast in an FPGA, but not very fast in a microcontroller
[04:56:08] <Guest971> i never count using a micro
[04:56:17] <SWPadnos> also, generating step pulses is easy if you can use a hardware PWM to do it (like the mega162 or something - it has to be the advanced PWM in the mega chips)
[04:56:27] <Guest971> i will mix a ic/counter with a micro
[04:56:27] <SWPadnos> but it's not as easy in software
[04:56:50] <SWPadnos> the counter chips that are available would cost nearly as much as a pico systems board
[04:57:00] <Guest971> no!
[04:57:01] <SWPadnos> they're $35-75 for each channel
[04:57:06] <Guest971> a 8 bit counter
[04:57:07] <SWPadnos> yes, you can't just use a CD4040
[04:57:30] <Guest971> for lower byte and micro will count the upper
[04:57:31] <SWPadnos> there is another board you might want to try - the MEsa Electronics 7i43
[04:57:45] <SWPadnos> that's a $80 board that connects to the parallel port and has an FPGA on it
[04:57:58] <SWPadnos> http://www.mesanet.com/
[04:58:57] <SWPadnos> (click on "Parallel I/O - Misc I/O" in the upper left)
[04:59:23] <Guest971> i am a sw programmer and i am not expert on fpga
[04:59:29] <SWPadnos> heh - me too :)
[04:59:41] <SWPadnos> there is already software included in EMC2 that will run that board
[04:59:48] <SWPadnos> you can look at the source to get more ideas
[05:01:18] <Guest971> i read the ppmc driver source, but if i had the diagram , i will be able to build a more same board
[05:01:48] <SWPadnos> all it is from the PC side is an EPP peripheral
[05:01:57] <SWPadnos> the FPGA source isn't published
[05:01:59] <Guest971> also i can build my own board and driver but it is some time consuming
[05:02:43] <Guest971> can i communicate with epp easyly?
[05:03:20] <SWPadnos> it's not too hard on the PC side
[05:03:20] <Guest971> i read some documentation but i could not what exactly happends
[05:03:38] <Guest971> on the other side , connect to a micro
[05:03:50] <SWPadnos> the AVR side would need to use an interrupt pin for the strobe, which would potentially screw up step generation or encoder counting
[05:06:15] <Guest971> i thing i am not expert to work in epp mode,but i used parallel port as standard mode
[05:06:40] <SWPadnos> actually, you can do things however you want
[05:06:56] <SWPadnos> I don't think you'll be able to exactly emulate the USC, so you won't be able to use the existing driver
[05:07:12] <SWPadnos> that gives you the freedom to do anything, since you get to design the hardware and write the driver too :)
[05:07:59] <Guest971> what does the address and REGISTER IN PICO SYSTEM MEANS?
[05:08:22] <SWPadnos> are you asking what each register does?
[05:09:36] <Guest971> no how epp select a board and how send data to the board
[05:09:37] <Guest971> a channel of the board
[05:09:57] <Guest971> http://pico-systems.com/encdregs.html
[05:10:01] <SWPadnos> oh - they have jumpers to select which block of addresses each board will respond to
[05:10:30] <SWPadnos> EPP has an address cycle and a data cycle - you send the address in software
[05:11:50] <Guest971> the first 4 bit is for slot next 4 is for seome thing else
[05:12:21] <Guest971> can u send me a link to read more
[05:12:27] <SWPadnos> not really
[05:12:42] <SWPadnos> that link you pasted is where I would send you to get pico systems information
[05:13:05] <SWPadnos> a google search for parallel port programming will get you a lot of data on EPP/SCP/normal modes
[05:13:16] <SWPadnos> uh - ECP, not SCP
[05:14:23] <Guest971> i want to know how pico handle these addreses
[05:14:43] <SWPadnos> you can email them and ask. I don't know
[05:15:52] <Guest971> it does not pwm?
[05:16:30] <SWPadnos> they sell several kinds of device. one does PWM, another does step generation, another has a very nice 16-bit DAC
[05:17:58] <SWPadnos> are you talking about the PPMC only, or have you looked at the Universal Stepper Controller and Universal PWM Controller?
[05:18:04] <SWPadnos> http://pico-systems.com/motion.html
[05:19:37] <Guest971> i want to use with ppmc driver an EMC
[05:20:35] <SWPadnos> the PPMC is a privately made product, with no schematics published
[05:21:13] <SWPadnos> the only information you will have will be (1) the EMC driver, (2) whatever is on the pico systems site - like register maps, and (3) whatever the designer decides to tell you about it
[05:21:15] <Guest971> iw ant to use whit emc
[05:21:21] <SWPadnos> I understand
[05:21:50] <SWPadnos> the information you need is owned and known by someone other than me - you can email him from the links at pico-systems.com
[05:22:10] <Guest971> so i think the ppmc and the driver does not support the PWM
[05:22:34] <Guest971> software pwm is working as well as i remember
[05:22:57] <SWPadnos> I'm not sure about the PPMC specifically, I know there are other products from pico systems that do PWM
[05:23:22] <Guest971> how is the usual xilix?? FPGA speed?
[05:23:44] <SWPadnos> heh - there are a lot of things that affect the speed of an FPGA
[05:24:10] <SWPadnos> with any modern Xilinx chip, you can get to at least 30 MHz, depending on the complexity of your code
[05:24:17] <SWPadnos> usually quite a bit faster
[05:24:58] <SWPadnos> the older mesa cards (using the Spartan 2) can use a 100 MHz PWM clock
[05:25:14] <SWPadnos> newer ones can probably go to 150-200 MHz if you needed it
[05:26:11] <Guest971> for sampling in an encoder the 8mhz is sufficient
[05:26:23] <SWPadnos> yes, for an FPGA it certainly is
[05:26:43] <Guest971> AVR i am talking
[05:26:55] <SWPadnos> for an AVR, you'll need about 30-50 clock cycles (assuming you have a 2-byte counter in SRAM) for the interrupt service routine
[05:27:17] <SWPadnos> if you want any noise canceling, it's higher
[05:27:51] <Guest971> 10k sample in a second is ok for sampling
[05:28:03] <SWPadnos> oh. in that case I'd just use the PC to do it
[05:28:13] <SWPadnos> you can get to 25KHz or better on most PCs
[05:28:25] <Guest971> :)) you are ok
[05:28:29] <SWPadnos> heh
[05:29:32] <Guest971> as i am a software programmer so i am not going to trust to any os even linux :)
[05:29:39] <SWPadnos> I have servos that can spin at 2500 RPM, and have 4000 count encoders on them - for that I need external hardware since it's 166 kHz count speed
[05:29:58] <SWPadnos> well, the real-time kernel is very very good
[05:30:22] <SWPadnos> if you set it up correctly, the RT parts will continue to run even if the kernel crashes
[05:30:26] <SWPadnos> I know this from experience
[05:33:11] <SWPadnos> time for me to go to sleep. good luck
[05:48:01] <Guest971> ok
[06:25:14] <Guest971> who is online
[06:31:41] <Guest492> how can i extend my parallel port simple like (ppmc) for using in EMC
[06:51:27] <Guest411> John Kasunich, Jon Elson, Stephen Wille Padnos are here?
[08:59:16] <archivist> * archivist is amused at the "cnc data" mail on the emc-users list
[10:21:22] <alex_joni> Guest706: hello
[10:21:33] <alex_joni> they usuually are around a bit later (it's still night in the US)
[11:30:36] <Guest706> hi alex
[11:30:42] <Guest706> i have to go
[11:30:49] <Guest706> C u later
[12:49:38] <skunkworks_> Ok - looks like our tosvert vfd has 0-10v input and 'at speed' output.
[12:50:07] <alex_joni> sounds perfect :)
[12:50:44] <skunkworks_> Yes :)
[13:34:15] <tomp> freshmeat today: 'rejoystick' maps joystick/pad buttons to keyboard keys. maybe some Axis keys can be on a joystick. sez it's invisible to X and SDL
[15:06:19] <skunkworks_> hmm it will also do 0-5v - simple rc filter might be nice to test.
[15:53:38] <Beege_> So I have to replace my drives... the ones I have are non-functional, having firewire communications.
[15:54:00] <skunkworks_> aww. Have you found any compatable with your servos?
[15:54:12] <Beege_> What is the difference b/n analog and digital drives, and what's better for a milling machine/router
[15:55:57] <Beege_> trying to get educated here, not knowing much about electronics, etc.
[15:58:08] <Beege_> Servos are 230VAC, .75kW, Sine AMP type, Max 5000RPM, 23.3 in/lbs stall torque
[16:31:05] <skunkworks_> email jonE and see if your servos will work with his new drives
[16:43:11] <skunkworks_> seems to me his drives are sin - not trapizoidle
[16:44:54] <skunkworks_> http://jelinux.pico-systems.com/acservo.html
[16:49:50] <skunkworks_> boy - I sure have been pushing Jons hardware a lot lately
[16:50:18] <skunkworks_> boy - that sounds wrong.
[17:07:49] <Beege_> checked out your link, will email Jon
[17:15:21] <Beege_> How do I know if my servos have hall sensors or something else?
[17:26:53] <jacj_live> hi peoples
[17:27:18] <alex_joni> 'lo
[17:27:22] <jacj_live> waths best post processor for use with artcam?
[17:36:03] <alex_joni> ROFLMAO..
[17:36:15] <alex_joni> lady of the night-lady of the night-lady of the night
[17:36:53] <alex_joni> that's the political correct greeting of santa :D
[18:06:01] <BigJohnT> LOL
[18:06:23] <BigJohnT> we say hoe hoe hoe round ere
[18:06:36] <BigJohnT> and I have a big yellow hoe
[18:07:29] <alex_joni> errm :P
[18:08:56] <jacj_live> HI PEOPLES
[18:09:24] <jacj_live> WATH best posprocessor for use artcam and emc2?
[18:10:53] <BigJohnT> jacj you plasma cutting?
[18:11:16] <jacj_live> no
[18:11:46] <jacj_live> i use desktop mill
[18:12:03] <cradek> To be sure, you should ask the vendor. If they don't know that people are using EMC2 you should be sure to tell them.
[18:13:08] <BigJohnT> ok, any post that will output the RS274NGC
[18:13:31] <BigJohnT> I've found that some generic fanuc will work with some programs some times
[18:15:32] <jacj_live> Before I used mach 3, I am now in doubt choose which pos
[18:15:58] <cradek> since mach is based on EMC code you will probably not have to change very much
[18:17:37] <jacj_live> humm ok tanks
[18:18:12] <jacj_live> sorry for my english, i´am from Brasil
[18:19:15] <cradek> your english is fine
[18:21:51] <skunkworks_> better than mine :)
[18:22:40] <BigJohnT> jacj, are you switching a mill over to EMC2?
[20:12:54] <jacj_live> sorry i work in machine
[20:13:19] <jacj_live> not understan the last question
[20:14:02] <skunkworks_> what didn't you like about mach?
[20:19:09] <SkinnYPupp> I'm sure you have all heard about the mortgage problems at GMAC and Countrywide but sometimes ya just gotta hear it from the horses mouth....this is too funny. HTFC ceo deposition
[20:19:10] <SkinnYPupp> http://consumerist.com/370052/htfc-mortgage-company-ceo-has-a-potty-mouth
[20:19:27] <SkinnYPupp> http://www.portfolio.com/views/blogs/daily-brief/2008/03/06/the-r-rated-ceo\
[20:29:56] <jacj_live> tanks peoples bye-bye
[21:19:24] <micges> BigJohnT: hello
[21:19:41] <BigJohnT> hello micges:
[21:20:22] <micges> could you tell me what is about patch to home vel or sth, that you try(ed) compile in to emc ?
[21:20:29] <BigJohnT> I have tramming my bridgeport head to under 20 minutes... too much practice... too many broken endmills
[21:21:04] <BigJohnT> it is a patch to add HOME_VEL so you can set the speed that it moves from latch to home position
[21:21:50] <BigJohnT> I was unable to get the compile to work and ran out of time to try :(
[21:21:53] <micges> why ?
[21:22:11] <micges> what for settable home_vel ?
[21:22:16] <BigJohnT> yes
[21:22:23] <BigJohnT> instead of rapid
[21:22:41] <micges> ok thaks
[21:23:22] <micges> I asked becouse I found strange behavior while homing and remebered that
[21:23:34] <BigJohnT> it seems to me that as you get faster > 400IPM the home move jerks the machine more than a G0 does
[21:23:49] <BigJohnT> but I can't prove it : (
[21:24:39] <BigJohnT> I could get my steppers to stall by increasing acceleration during the home move but still do a G0 without stalling...
[21:25:06] <micges> mhm
[21:25:26] <micges> you tell 400 inch/min ?
[21:25:27] <BigJohnT> ?
[21:25:45] <BigJohnT> 405 IPM at the present time
[21:25:54] <micges> what is distance of that (error) move ?
[21:25:57] <BigJohnT> up to 500 IPM but didn't really need it
[21:26:24] <BigJohnT> to home?
[21:27:32] <micges> from latch
[21:27:38] <micges> to home
[21:28:13] <BigJohnT> 12.7mm
[21:28:15] <alex_joni> HOME_OFFSET<->HOME
[21:28:16] <BigJohnT> 1/2"
[21:28:24] <alex_joni> that small?
[21:28:29] <BigJohnT> yes
[21:28:42] <alex_joni> and it reaches 400IPM?
[21:29:08] <BigJohnT> yes it only takes 0.1" to reach 400IPM
[21:29:48] <alex_joni> mean machine :P
[21:30:11] <BigJohnT> * BigJohnT has a BIG GRIN on his face
[21:35:42] <BigJohnT> time to go
[22:47:08] <dmess> im back... screen shot locks me up SOLID
[22:49:20] <dmess> tomp... getting anywhere???
[22:50:18] <dmess> on driving tool along more than 1 part surface.. but the same line...
[23:45:28] <gene> Hey guys!
[23:45:46] <jepler> hi gene
[23:45:48] <gene> with a big grin here too.
[23:46:31] <gene> I just put some ball bearing thrust washers in the Y axis, backlash is now about .0006"
[23:47:33] <gene> x is next obviously
[23:48:33] <jepler> sounds good
[23:49:01] <gene> I also drilled into the bore, and set a grease fitting. It will now do 15 ipm from one end of the screw to the other without a oeep
[23:49:11] <gene> oeep/peep
[23:49:33] <jepler> all I've been up to is tedious technical ubuntu stuff
[23:50:15] <gene> but its getting a bit chilly for a t shirt. ubuntu? Next LTS release?
[23:51:26] <jepler> yes -- trying to lay some groundwork so that it'll be possible to run emc on the next LTS release. but don't hold your breath...
[23:52:05] <jepler> (actually, the initial results are promising, but nothing beats the proven stability and reliability of Dapper)
[23:52:13] <gene> Sounds like I ought to pack my /home/gene up to another box, so I can re-install on this one eventually.
[23:52:43] <gene> I agree, this except for my hardware, has been bulletproof.
[23:53:20] <skunkworks> Jepler: one thing I had mentioned to alex - when you shut down - the graphic screen stays up. So you never get the 'system haulted' or what ever it says on dapper. You can hear the harddrive shut down...
[23:53:43] <toastyde1th> toastyde1th is now known as toastydeath
[23:54:30] <gene> uptime isn't so hot, I had it down to put the whole thing back on the UPS a week ago. Thats odd, something to do I suppose with the no powerdown on account of the realtime stuffs?
[23:55:58] <gene> Gotta run Jeff, getting the shivers, tomorrow maybe after I do the x too, and not make so many mistakes...
[23:56:02] <jepler> see you gene
[23:56:35] <jepler> skunkworks: huh that's an annoyance
[23:57:06] <skunkworks> I am assuming it isn't my hardware..
[23:57:13] <skunkworks> :)
[23:57:14] <jepler> I doubt it
[23:57:53] <skunkworks> before shutting it down and banging on it pretty hard the jitter hit max of 8.2us
