#linuxcnc | Logs for 2013-08-21

Back
[02:05:11] <DJ9DJ> moin
[03:16:11] <false> Hey all
[03:16:30] <kengu> hey you
[03:16:43] <false> I have a question, or actually a couple :P.
[03:22:13] <ReadError> you may have 1!
[03:23:52] <false> I am in the process of retrofitting an old Matsuura MC710V, I am using Mesa cards (2x 7i33, 1x 7i37). I am reusing the drives which are old fanuc velocity drives with tach input (using the second 7i33 as tach output to drives). Now for the first question, is it normal for the servo to drift when I use the following: P0 I0 D0? I have shorted the analog and tacho to ground on the drives to
[03:23:52] <false> correct for 0 drift on the drive. I also tried to get the 7i33 outputs as close to zero as I can, within 0.0015V. But when I connect the drives to the 7i33 with those pid parameters and enable the drives the servo starts to drift.
[03:31:50] <false> And the second (sorry I need more than one :P), if I use a Pgain of more than 0.5 and jog at low speed (250mm/m) the axis oscilates violently to the point where the fuses blow or the drives go into overload, I am seeing way higher numbers on other configs, is there something I am missing?
[03:34:05] <false> I've followed the tutorial on tuning velocity drives but I can't set my Pgain high enough to correct for initial following error.
[03:38:50] <cncbasher> anyone running camview on development as run in place , i find the onscreen dro's work fine on installed 2.5.3 , but emc.halio plugin fails when used on dev and run in place
[03:45:10] <archivist> cncbasher, been some chatter on the mailing lists with mention of emc.halio
[03:45:51] <archivist> not deeply following as I dont run camview
[03:46:19] <cncbasher> yes i have seen it and thats what prompted me to look further
[03:47:03] <Valen> false: the numbers are irrilivent
[03:47:17] <Valen> also if you are metric vs inches i believe they will be very different
[03:47:35] <cncbasher> i presume it's linking to installed version
[03:48:19] <Valen> i don't know about velocity drives in paticular but generally set P higher and higher till it oscilates then back it off by 1/3rd
[03:48:45] <Valen> add some ff1 to null out any fixed offset (this could be different with a velocity drive)
[03:48:47] <false> Valen: Increase while idle, and don't try to move the axis?
[03:48:52] <Valen> add some I and a little D
[03:48:58] <Valen> move the axis around
[03:49:02] <Valen> slow and fast
[03:49:27] <Valen> my P is like 3 on one axis and 5 on another
[03:49:49] <Valen> also make sure your deadband is larger than an encoder count
[03:49:59] <Valen> otherwise I will build up and make it oscilate
[03:50:15] <false> Ok, I'll go and try those, thanks!
[03:50:52] <false> Did you happen to see my first question to?
[03:51:32] <Valen> if you have no correction random noise and offsets will make it drift
[03:52:12] <false> Ok, thought that was it, but wanted to make sure before I try to tuning something that isn't correct in the first place
[03:52:36] <Valen> i think there is a "bias" term you can put in to correct that?
[03:54:18] <false> I'm going to try again, thanks for your help!
[07:41:18] <false> And we're back. Best results so far are P=1.0 I=0 D=0.1 FF1=0.03, but this way I have a static following error of about 0.6mm. Increasing FF1 does not help, also adding Igain reduces the static error but the loop starts to oscillate by about 0.4mm peak-peak. Any ideas?
[08:03:20] <jdh> I'd ask JT
[08:25:12] <false> Should I start a thread on this or will he be here?
[08:26:06] <archivist> he is behind you
[08:27:54] <false> :P
[08:31:53] <jthornton> lol
[08:32:21] <jthornton> false, is it a velocity drive?
[08:46:49] <pcw_home> Normally you should not need D on velocity mode drives so this is suspicious
[08:48:08] <pcw_home> also is this a MM or inch machine? (mm machines will have P gains lower by a factor of 25.4 if not normalized)
[08:48:25] <cradek> you don't have real tachs?
[08:48:34] <false> Yes it is a velocity drive (fanuc a06b-6035-h422)
[08:48:39] <false> mm
[08:48:58] <false> No tachs, they were not on the motors and were provided by the NC board
[08:49:10] <jthornton> http://gnipsel.com/linuxcnc/tuning/index.html
[08:49:28] <false> It used a f/v converter to generate tach from encoder input
[08:49:36] <cradek> your +-10 fake tach might be MUCH lower voltage than a real tach was
[08:49:54] <cradek> oh hmm I don't know what input range that would be then
[08:49:55] <false> Jthorton: I tried using that, but it did not get me anywhere unfortunatly
[08:50:12] <false> No the tach was 6v/1000rpm
[08:50:38] <false> the tach is giving the correct output and is stable
[08:51:02] <false> I bumped up the servo-thread to 5Khz to make sure it was updating fast enough
[08:52:27] <false> I also measured the outputs when the orignal NC board was still connected, then simulated a turning encoder using a mpg wired to the encoder input of the 7i33, the tacho output was correct
[08:55:21] <false> pcw_home: what do you find suspicious about this, could this be a hint as to what is causing the issue?
[08:59:52] <false> I was thinking that the drives may be configured in torque mode, but this seems not logical as the board need a VMCD (as opposed to TCMD) and a torque mode drive doesn't need tacho's right?
[08:59:57] <pcw_home> A velocity mode drive needs no D
[09:00:25] <false> Ok, I will set D=0
[09:00:34] <pcw_home> large static errors say torque mode to me
[09:01:25] <pcw_home> so something about the velocity feedback is not working (possibly backwards?)
[09:04:36] <pcw_home> if they are smaller motors an you can decouple them you should be able to feel the
[09:04:37] <pcw_home> velocity feedback with 0 commanded velocity and PID = 0
[09:04:39] <pcw_home> (it should attempt to hold the shaft still if you turn it, though it may creep slowly)
[09:06:54] <false> Feedback is correct, it is inverted. The drive uses a vcmd of 7v/1000rpm and a tsa (tacho) of 6V/1000rpm but a positive command must generate a negative tacho signal, otherwise servo runs away at full speed. The servo is working correctly but I cannot get the tuning right
[09:07:59] <false> The servos are fanuc 5 DC servos
[09:08:46] <pcw_home> So you are simulating the tachometer with a 7I33 channel?
[09:08:55] <false> Yes that is correct
[09:09:34] <pcw_home> and the 7I33 velocity out scaling is correct?
[09:10:20] <false> Yes, I measured the original signals, and compared them to what I am getting now, they are only .0005 volts of
[09:11:48] <false> When using the above tuning the static error goes away when moving the axis but it returns when I stop moving
[09:12:22] <false> So the PID is correcting the error
[09:14:21] <pcw_home> what is the PID output voltage when you have this static error?
[09:17:09] <false> 0.0775Volts
[09:17:31] <false> thats strange?
[09:18:47] <false> shouldn't that be zero?
[09:18:50] <pcw_home> well 77 mv should make it move
[09:19:41] <pcw_home> there will be some offset but it should be maybe 1/10 or or less
[09:20:27] <false> It isnt moving, I watched hm_5i20.0.encoder.[z].count and its dead quiet
[09:21:10] <false> I tuned the drive for 0 drift with the analog and tacho shorted to ground.
[09:21:43] <pcw_home> tune for 0 drift when the drives are enabled and connected to the 7I33
[09:22:34] <false> funny story, when the 7i33 is connected the offset potmeter does not have any effect, could the 0.0775V be to much to correct with the offset potmeter?
[09:23:33] <pcw_home> when the 7I33 is connected you must disable the PID loop to make this adjustment
[09:24:54] <false> So I must enable the 7i33 outputs + drive, and disable the pid loop?
[09:25:53] <false> That makes sense, the loop is just correcting the offset adjustment I am making, correct?
[09:26:03] <pcw_home> Yes (it will not drift if the PID loop in working)
[09:26:46] <false> Ok I'll go and do that, thanks for your help Peter!
[09:30:19] <false> I can disable the pid loop by just setting everything to 0 right?
[09:31:59] <cradek> yes
[09:32:23] <false> Thanks
[09:42:56] <false> Again no response from the offset potmeter, the tacho out is responding to the moving axis so the drive is getting a tacho feedback, with the PID disabled the 7i33 output is at .0244V
[09:44:04] <false> I should offset this to as close to 0 as I can get right, but still very strange the offset potmeter is not doing anything. The drive does respond when analog an tacho are shorted to ground
[09:44:23] <pcw_home> wiring problem?
[09:45:58] <pcw_home> does the drive have differential inputs?
[09:46:26] <false> All shielded twisted pair separate for each signal, shield terminated at the 7i33. Connecting a 1.5V battery to the drive's analog in the drive responds as it should.
[09:46:51] <pcw_home> but does the drive have differential inputs?
[09:47:10] <false> Unfortunatly, no
[09:47:19] <pcw_home> Yuck
[09:47:21] <false> common ground
[09:47:25] <false> yep
[09:48:15] <false> sorry didnt catch your question the first time
[09:48:26] <pcw_home> sure you have the PID /velocity feedback disabled?
[09:48:52] <false> all set to zero
[09:49:24] <false> i can comment out the following to be sure: net x-enable => pid.x.enable??
[09:50:14] <pcw_home> Yes
[09:50:22] <false> Ill do that
[09:50:29] <pcw_home> how is the velocity feedback wired in hal?
[09:52:06] <pcw_home> it should be something like: mumble.encoder.n.velocity --> mumble.pwm.value
[09:53:10] <false> net y-tacho-out <= scale.tacho-y.out => hm2_5i20.0.pwmgen.04.value
[09:54:09] <false> net y-vel-fb => scale.tacho-y.in
[09:54:31] <false> net y-vel-fb <= hm2_5i20.0.encoder.00.velocity
[09:54:57] <pcw_home> OK so a scale comp in between (though PWM scale should do)
[09:55:23] <false> yes i tried the pwm scale but it didnt work, probably me doing it wrong
[09:56:03] <pcw_home> it should (and you can use a negative sale to reverse the output)
[09:56:13] <pcw_home> negative scale
[09:56:36] <false> setp hm2_5i20.0.pwmgen.04.scale [AXIS_1]OUTPUT_TACHO_SCALE
[09:56:36] <false> setp scale.tacho-y.gain [AXIS_1]TACHO_SCALE_GAIN did not work for me
[09:56:57] <false> copy paste error over here
[09:57:13] <false> setp hm2_5i20.0.pwmgen.04.output-type 3
[09:57:13] <false> setp hm2_5i20.0.pwmgen.04.scale [AXIS_1]OUTPUT_TACHO_SCALE
[09:58:13] <false> where the scale is set negative
[09:59:23] <false> correction the scale is not set negative, but the encoder is
[10:02:08] <pcw_home> well the encoder scale needs to be set to get the correct axis readout
[10:02:10] <pcw_home> (though its probably better to swap encoder leads)
[10:03:16] <false> I will do that when everything works, the problems was the encoder connector was repaired and connected the wrong way around. So that caught me by suprise (runaway)
[10:03:50] <pcw_home> well runaways are a 50% chance when setting up
[10:04:47] <pcw_home> at least if you get the drive enables and encoder working first, linuxcnc can catch them
[10:05:19] <false> Yes indeed
[10:05:52] <JT-Shop> unless you turn the ballscrew by hand first and verify the direction is correct
[10:06:47] <false> I knew the polarity of the orignals so i checked them first, but assumed after 2 encoders the 3rd would be the same, it was kind of hard to reach
[10:07:38] <false> I should have been the same but the repairman had switched the A and B channels
[10:08:00] <pcw_home> Yes you can also disable the drive and see the that the analog output voltage changes in such a way as to resist the motor shaft motion
[10:08:37] <false> I will go and try some more things, brb
[10:54:49] <PetefromTn> Morning Folks..
[10:56:27] <tjtr33> #join #zeromq
[10:56:39] <tjtr33> damn
[10:56:54] <jdh> at least it was just #zeromq this time.
[10:57:00] <jdh> we still remember the last one.
[10:58:10] <tjtr33> hokay, i must have offloaded that mem to the cloud :0 sorry i dont remember myself
[11:01:37] <archivist> where cloud=nsa these days :)
[12:08:34] <IchGuckLive> hi all
[13:03:48] <IchGuckLive> hi
[13:07:52] <andypugh> AHoy!
[13:09:38] <archivist> so I went hunting for the maths, found some, stuck it in some php... http://www.archivist.info/gear/designbevel.php
[13:10:42] <andypugh> That's a bit over-zealous :-)
[13:11:13] <archivist> there is still some missing though
[13:11:52] <andypugh> Do you think it is possible to generate bevels with spiral hobs and CNC?
[13:12:08] <archivist> no
[13:12:59] <skunkworks> heh - are you sure? ;)
[13:13:02] <archivist> the faster machines use two spindles on a taper path
[13:14:18] <archivist> reasonable sure for straight bevels, I am ignoring hypoid
[13:14:34] <andypugh> There is a simplified type that uses formed tools and two cuts to get the taper. I guess the problem is that a spiral hob doesn't generate the correct form?
[13:15:21] <andypugh> It would be interesting to see what yoy _do_ get :-)
[13:15:28] <archivist> the form is right but no taper on a spiral hob
[13:16:01] <archivist> there are some tangential hobs
[13:16:04] <andypugh> And I guess that the tooth gap is fixed.
[13:17:06] <andypugh> Have you seen http://www.youtube.com/watch?v=NtKH_fKOod0
[13:17:07] <Tecan> (NtKH_fKOod0) "Bevel gear solutions through up-Gear Technology" by "sandvikcoromant" is "Tech" - Length: 0:06:45
[13:17:55] <andypugh> That looks like quite a stiff machine tool!
[13:19:45] <archivist> andypugh, just had a look in the barber colman book there is a taper tooth method with unsymetrical tooth form, used on anti backlash gear, used an oblique feed
[13:23:36] <andypugh> Sound like fun
[13:24:41] <archivist> there is a coordinated move in that vid that could be done with geared axes
[13:25:45] <archivist> this needs to be adjustable in gcode though
[13:31:42] <archivist> over zealous? , that is a mod of an earlier program, later it will output the gcode now it has most of the numbers
[13:31:43] <FinboySlick> I just learned that Tormach ships linuxCNC with their new lathe.
[13:31:48] <FinboySlick> That's pretty cool
[13:32:01] <jmasseo> i thought tormach had historically supported mach
[13:32:20] <archivist> yes but they want the lathe to work :)
[13:33:06] <archivist> they need better software for screw cutting
[13:33:43] <FinboySlick> It looks like a pretty nice lathe.
[13:35:30] <andypugh> Tormach came to Wichita
[13:37:36] <FinboySlick> andypugh: What was in Wichita?
[13:38:03] <archivist> linuxcnc meetup
[13:38:14] <FinboySlick> Ooh.
[13:40:22] <andypugh> Mainly a developer meetup.
[13:40:57] <FinboySlick> andypugh: Do they plan to contribute to the software too?
[13:41:28] <andypugh> They already have.
[13:42:05] <andypugh> http://git.linuxcnc.org/gitweb?p=linuxcnc.git&a=search&h=HEAD&st=commit&s=rogge
[13:42:05] <FinboySlick> That's very nice.
[13:42:53] <FinboySlick> That's a great case for mutual benefit of open source and hardware.
[13:50:43] <false> PCW: Still there?
[14:11:43] <andypugh> It looks like Tormach need a LinuxCNC module to control their tooolchanger, it is currently listed as "Not compatible with LinuxCNC"
[14:12:58] <archivist> cannot be too difficult to make it work
[14:13:38] <archivist> has to be easier then effin bevels
[14:14:42] <andypugh> Nobody made you decide to start making bevels!
[14:15:25] <archivist> someone asked if I could
[14:15:36] <andypugh> Ah
[14:15:45] <archivist> so I can apportion blame
[14:16:48] <archivist> I knew there would be some difficulty, but it was an itch that needs scratching
[14:18:34] <automata> hello
[14:19:06] <archivist> olleh
[14:19:51] <automata> I guess network order is big endian
[14:22:44] <t12> yes
[14:38:37] <andypugh> Quick survey: Who knows where the phrase "Big Endian" originated?
[14:40:10] <FinboySlick> I'm not even sure if it means most significant first.
[14:40:30] <andypugh> It does mean most significant first
[14:41:36] <false> it comes from the story of gullivers travels, no im not kidding
[14:41:47] <FinboySlick> andypugh: Is that what you meant by 'originated' though?
[14:42:42] <AR__> got some new mic6 to make a nice fixture plate today
[14:42:44] <andypugh> false: Indeed, along with the L in the LES lamp base :-)
[14:43:08] <AR__> made the bolt holes to hold it down, now to drill and tap a shitload of holes in a grid...
[14:43:38] <andypugh> AR__: Do you have rigid tapping?
[14:43:52] <AR__> nope :D
[14:44:09] <AR__> not even a tapping head
[14:44:23] <AR__> my wrist is gunna be aching
[14:45:17] <andypugh> AR__: I often drive taps with a cordless drill. It works well.
[14:45:22] <AR__> guh
[14:45:26] <AR__> in aluminum?
[14:45:42] <andypugh> Mainly steel.
[14:45:58] <cradek> in thin stuff that works (it's forgiving about uprightness)
[14:46:00] <andypugh> But I can't see it working less well in aluminium
[14:46:35] <andypugh> I wouldn't necessarily use it for blind holes.
[14:46:50] <cradek> if your fixture plate has blind holes you're doing it wrong
[14:47:13] <AR__> lul
[14:47:20] <AR__> it will all be through holes
[14:47:22] <AR__> for sure
[14:47:30] <andypugh> Actually, in your situation I think I would become gradually more enthusiastic hole by hole, then get the drill out at about hole number 6
[14:47:43] <AR__> lol
[14:47:54] <cradek> in your situation I'd buy a tapping head and stick it on the drill press
[14:47:57] <JT-Shop> yep with a sprial point tap and a cordless you can go to town
[14:48:19] <JT-Shop> tapping head = $500
[14:48:36] <cradek> only if you're crazy enough to buy it new
[14:48:48] <JT-Shop> well yea, just call me crazy
[14:48:52] <false> Hey all, I have a question, earlier I was having trouble tuning my servos. I could not ajust my drives for 0 drift with drives enabled an PID disabled, turns out when I connect a multi meter to measure the output on the VCMD pins I can. I measured the resistance of the drive input and it is about 300ohms. I am using a 7i33, the manual states the load should be at least 5KOhms, could this be
[14:48:52] <false> the problem?
[14:49:36] <AR__> i've heard you can make a holder to lightly hold the tap, with a square hole in the back for the square shank
[14:49:44] <AR__> then run it in a drill press or mill under power
[14:49:53] <AR__> and when you stop feeding it just pulls out of the holder
[14:50:11] <AR__> but if it galls up and sticks it would definitely break
[14:50:29] <AR__> need a slip clutch or something lol
[14:50:41] <JT-Shop> if it galls up your doing it wrong
[14:51:00] <AR__> i mean like
[14:51:06] <JT-Shop> what size tap?
[14:51:08] <AR__> when you hand tap, you break the chip by backing out a bit
[14:51:15] <AR__> under power you dont do that
[14:51:24] <AR__> 1/4-20
[14:51:33] <cradek> you have the wrong tap if you have to back up
[14:51:43] <cradek> get good spiral point taps
[14:51:44] <JT-Shop> nope and I tape thousands of 1/4-20 and never back up except to come out of the hole
[14:51:50] <AR__> hm
[14:51:51] <cradek> even if you're hand tapping get spiral point
[14:51:54] <JT-Shop> yep
[14:51:58] <JT-Shop> yep
[14:52:13] <andypugh> false: Is the 7i33 enabled?
[14:52:14] <cradek> department store "hand taps" are for fools
[14:52:15] <JT-Shop> OSG electralube is the best I've found
[14:52:32] <false> yes
[14:52:38] <JT-Shop> AR__: are you in a big city?
[14:53:17] <false> I was talking to Peter earlier, he gave me some tips I was trying out when I discovered this
[14:53:24] <andypugh> I would expect the multimeter to be a few megs, not enough conductance for the 7i33 to notice.
[14:53:53] <andypugh> With the drive disabled and the mutlimeter on the 7i33 output, can you make a voltage appear?
[14:54:07] <archivist> I used a hand tap and battery drill and 3/8 al plate and tapped 20 odd holes for a test fixture yesterday
[14:54:30] <AR__> i just have normal taper taps
[14:54:50] <JT-Shop> AR__: http://www.mscdirect.com/product/64949001
[14:55:07] <false> Yes, everything is working fine, and I can repeat this on every axis. As soon as I connect the DMM the problem goes away
[14:55:24] <andypugh> So, invest in 2 more DMMs :-)
[14:55:36] <false> I can move the axis around, no problem, but I got stuck tuning
[14:55:47] <false> Hehe, if that solves the problem, gladly
[14:55:57] <andypugh> I only have hand (straight-flute taps) and they don't cause me any problems with the cordless drill.
[14:56:51] <AR__> http://www.ebay.com/itm/HSS-SPIRAL-POINT-TAPS-6-32-through-1-2-13-Individual-Free-Shipping-/261237838408?pt=LH_DefaultDomain_0&var=&hash=item3cd2fd1248
[14:56:53] <AR__> how about these
[14:57:05] <AR__> probably would get here in <2 days
[14:57:42] <AR__> they dont look like spiral points though o_0
[14:58:33] <andypugh> They also look nasty.
[14:59:15] <cradek> yeah those don't look right at all
[14:59:28] <false> I could connect a 1MOhm resistor across the input to test maybe?
[14:59:29] <cradek> order hss sprial point from enco or msc
[14:59:31] <JT-Shop> AR__: what is your location?
[14:59:40] <AR__> pennsylvania
[15:00:12] <AR__> there's a lot of 1/4-28 ones for cheap on ebay that look good
[15:00:20] <andypugh> Spiral Point is not the same as Spiral Flute
[15:00:30] <JT-Shop> got a MSC near by?
[15:00:31] <AR__> but i dont have enough 1/4-28 bolts
[15:00:31] <AR__> lol
[15:01:08] <JT-Shop> if you have a bunch of holes to tap $10 is cheap for a good tap
[15:01:17] <AR__> no msc around
[15:01:20] <AR__> http://www.ebay.com/itm/1-4-28-TAP-USA-MADE-NEW-FROM-BOX-/220735305495?pt=LH_DefaultDomain_0&hash=item3364d97717
[15:01:24] <AR__> that is spiral point right
[15:01:47] <cradek> yes
[15:01:50] <AR__> but i want 1/4-20
[15:01:53] <AR__> bleh
[15:03:16] <JT-Shop> there are 6 MSC's in PA
[15:03:24] <andypugh> You sure you don't want 1/4-28: http://www.ebay.com/itm/DORMER-Spiral-Point-Tap-1-4-28-H3-3FL-HSS-PM-S-O-UNF-E036-Qty-5-786-/121049381612
[15:03:33] <JT-Shop> http://www.mscdirect.com/corporate/locations-branches#Pennsylvania
[15:03:57] <AR__> well, i could just buy 1/4-28 bolts
[15:04:56] <andypugh> (There are other good brands than Dormer, but Dormer are certainly excellent)
[15:05:01] <AR__> JT-Shop, closest one is about 2 hours away
[15:05:55] <JT-Shop> darn
[15:06:17] <andypugh> Even better: http://www.ebay.com/itm/DORMER-HSS-FLUTELESS-ROLL-TAP-UNC-1-4-20-E523-/221100042171?pt=UK_BOI_Metalworking_Milling_Welding_Metalworking_Supplies_ET&hash=item337a96e7bb But wrong country
[15:09:51] <JT-Shop> enco has them at a better price http://www.use-enco.com/1/1/49604-1230002-osg-h3-hss-2-flute-nc-thread-plug-spiral-point-taps.html
[15:09:52] <AR__> how in the fuck does that tap work
[15:10:19] <JT-Shop> thread forming tap, not cutting
[15:11:02] <archivist> brute force, moves the metal
[15:11:17] <JT-Shop> I'll have to start buying my taps from enco they are more cost effective
[15:12:02] <Loetmichel> thread forming is great
[15:12:03] <AR__> o
[15:12:56] <Loetmichel> no swarf, harder and smoother thread surface, much more rigid tap tool (breaks not that easily).
[15:13:29] <Loetmichel> only drawback: taps are expensive and makes a "volcano" on the surface of the material
[15:13:44] <Loetmichel> i.e the holes should be generously countersunk
[15:14:11] <cradek> and best if reamed to size
[15:14:23] <AR__> https://www.youtube.com/watch?v=WRAXB545MVA
[15:15:07] <cradek> I've only form-tapped 4-40, but it worked great (blind holes)
[15:16:07] <Loetmichel> the company which builds our computer cases does it all day
[15:16:15] <Loetmichel> m2,5 to m8
[15:16:22] <Loetmichel> in aluminium and mild steel
[15:16:32] <Loetmichel> best threaded holes ever
[15:17:19] <Loetmichel> especially in zinc plated sheet steel
[15:17:56] <Loetmichel> whre his cnc punch makes little "volcanos" on poupose to sink the form-tap in
[15:18:31] <AR__> lol
[15:18:39] <Loetmichel> (i.e uses a 2,8mm punch and a 4.8mm die to make the pre-holes for the treads
[15:18:58] <archivist> how many holes per tap?
[15:19:10] <Loetmichel> because a m3 thread in 1mm steel doesent hold much
[15:19:22] <Loetmichel> so it has at least 2,5mm of thread
[15:19:31] <Loetmichel> about a 1000 or so
[15:19:38] <Loetmichel> then the tool is worn
[15:19:52] <archivist> lubricated or dry
[15:19:58] <Loetmichel> lubricated
[15:20:15] <AR__> ok at my machine
[15:20:20] <AR__> guess i'll spot all the holes
[15:21:20] <Loetmichel> and it is a Trumpf like this one: https://www.youtube.com/watch?v=0R62U0jY5Yo
[15:22:37] <AR__> think 1" hold pattern is good for standard fixture plate?
[15:23:10] <archivist> depends....
[15:23:12] <JT-Shop> for a desktop mini mill?
[15:23:17] <AR__> yeh
[15:23:38] <JT-Shop> unless your bored I'd just make them as needed
[15:24:03] <AR__> maybe i'll just do 2" for now and i can add more later
[15:24:05] <JT-Shop> but most of them I've seen are on 1" centers
[15:24:30] <Loetmichel> ... nust without the laser head
[15:24:32] <Loetmichel> just
[15:24:37] <archivist> this little plate I did has them at 20mm
[15:25:22] <Tom_itx> i did a 1" pattern on my little mill
[15:25:28] <JT-Shop> this type of tap handle makes tapping with the aid of the mill for position much easier to do http://www.use-enco.com/CGI/INLMKD?SIITEM001=325-4922&SIQTY001=1&SICOUNT=1
[15:25:43] <JT-Shop> http://www.use-enco.com/CGI/INSRIT?PARTPG=INLMKD&PMPXNO=5810007&PMAKA=325-4922
[15:25:55] <JT-Shop> if you have enough Z
[15:26:39] <Loetmichel> its an experiens seeing and feeling this trumatic punching 6mm mild steel sheets...
[15:26:45] <Tom_itx> We're sorry, use-enco.com is temporarily unavailable
[15:26:51] <archivist> I had to double up on this plate later that is under the A http://www.collection.archivist.info/archive/DJCPD/PD/2013/2013_08_11_bevel/IMG_1633.JPG not enough holes error
[15:26:52] <Loetmichel> thes grumbling through the whole company ;-)
[15:26:57] <Loetmichel> AND the parking lot ;-)
[15:28:17] <JT-Shop> Tom_itx: must be you, works for me
[15:28:46] * JT-Shop hands Tom_itx a Bit-O-Honey to tide him over
[15:52:55] <AR__> 45 holes spotted
[15:52:58] <AR__> guh
[15:53:10] <AR__> yeah i think i'll not do all of them right away
[15:53:49] <false> Is it possible to use a mesa 7i64 over usb to control some relays, this doesnt need to be realtime
[16:03:30] <andypugh> No. The 7i64 does not talk USB
[16:03:52] <andypugh> Hang on, I am actualy incorrect...
[16:04:23] <andypugh> It would be interesting to try.
[16:05:15] <andypugh> false: What other hardware do you have?
[16:06:01] <andypugh> (Also, what voltage and current do you need to switch? The 7i64 has quite a high voltage and current rating)
[16:06:09] <false> I have a Mesa 5i20 + 2x 7i33 + 1x 7i37ta
[16:06:42] <false> I need to switch some relays for an ATC and some extras like auto lube, coolant, etc
[16:07:39] <false> So the 5i20 is full, and I was thinking to interface via usb
[16:07:50] <andypugh> You have run out of outputs on the 7i37TA?
[16:08:26] <false> Yes almost, so I thought this might be a nice solution for non real time critical stuff
[16:09:02] <andypugh> It ought to be possible, but I don't know if the software components exist
[16:09:35] <false> I couldn't find anything on it so that's why I asked here
[16:10:16] <false> What would we need for this? An rt compatible usb driver (rt-usb maybe??) and a way to interface this to LinuxCNC?
[16:10:22] <andypugh> The 7i33 encoder connections are RS422 and the comms protocol that LinuxCNC supports the 7i64 on is Rs422. It is just about feasible that a special bitfile could allow you to connect the 7i64 to 4 terminals of a 7i33
[16:10:55] <false> Sounds very interesting, I should ask Peter about this right?
[16:11:11] <andypugh> I could try plugging a 7i64 into a USB port and seeing if hal_input can see it...
[16:11:33] <false> I'll try that, how would I go about doing that?
[16:11:54] <FinboySlick> Not that I know what I'm talking about, but if LinuxCNC can drive gpio directly, perhaps there's a usb gpio interface.
[16:12:26] <andypugh> false: Do you have a 7i64?
[16:12:30] <false> It happens I have one lying around yes
[16:12:35] <andypugh> Aha!
[16:13:02] <false> It was for another project but they canceled on me :(
[16:14:34] <andypugh> Can you get to linuxcnc.org?
[16:14:39] <false> Yes
[16:14:44] <andypugh> Odd, I can't
[16:14:58] <false> *hang on F5ing
[16:15:22] <andypugh> Look on the Wiki at "simple remote pendant". That has instructions for setting up hal_input and seeing what the device name is (etc)
[16:15:55] <false> Thanks I'll read up on that, I can get to http://linuxcnc.org/docs/html/man/man1/hal_input.1.html, but not to linuxcnc.org?
[16:17:13] <andypugh> Try the instriuctions here: http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Simple_Remote_Pendant
[16:17:33] <andypugh> They won't actually work for a 7i64, but you might get a feel for how hard it is going to be :-)
[16:21:54] <false> Hehe, I like a challenge, but its not visible in /proc , probably need the drivers for the ftdi chip
[16:34:19] <andypugh> USB _should_ declare some basic info at the very least.
[16:35:56] <false> It's not showing up
[16:41:16] <JT-Shop> grave digging detail is done...
[16:41:40] <skunkworks> uh oh...
[16:41:55] <JT-Shop> neighbors dog
[16:42:04] <JT-Shop> no match for a John Deere 310A
[16:43:16] <JT-Shop> it didn't look too good yesterday...
[16:43:53] <skunkworks> aw
[16:44:42] <skunkworks> when I was learning to drive - the neighbors dog ran out in front and I hit it. what a way to start driving
[16:44:56] <JT-Shop> yuck
[16:45:32] <JT-Shop> Fideau was an old dog, so it was her time. George is down to 5 dogs and 1 cat...
[16:47:46] <false> andypugh: I just thought of something, the FPGA on the card has no firmware on it, so I think it would be impossible to control the gpio on the card
[16:48:08] <andypugh> Which card?
[16:48:18] <false> the 7i64
[16:48:30] <andypugh> None at all?
[16:48:44] <andypugh> How did that happen?
[16:49:24] <false> I don't think so, doesn't it get pushed by the hostmot firmware?
[16:49:38] <andypugh> Not to the 7i64, only to the 5i20
[16:49:56] <false> Ah ok, I didn't know that
[16:50:14] <false> I thought it was programmed by the hostmot driver on init
[16:50:42] <DJ9DJ> gn8
[16:50:44] <andypugh> Not as far as I know. And I did write the driver, so I would expect to know :-)
[16:53:43] <false> Hehe, so I am talking to the right man for this quest into the unknown :P
[17:00:46] <andypugh> hal_input isn't a horribly complicated driver, but I think it doe rely on the USB device being detected.
[17:00:59] <andypugh> Have you configured the 7i64 jumpers to run of USB power?
[17:04:12] <false> Yes, and I found it in /dev/ttyUSB0
[17:05:04] <false> So I could in theory control the in and outputs via the protocol descibed in the manual
[17:06:20] <false> Ubuntu didn't detect it untill I jumpered it for 'always on' i.e. not after enumeration :$
[17:11:59] <andypugh> Do you need to add it to the plugdev group?
[17:12:13] <false> Just trying that out
[17:14:49] <false> already a member
[17:16:11] <false> but i can't add the device without the info that I would normally find in /proc/...... right>
[17:16:12] <false> ?\
[17:17:03] <false> I'm not a total Linux noob, but it's not my every day OS, so bare with me
[17:18:14] <andypugh> If that is the case, then you know more than me. I _am_ a Linux noob
[17:18:28] <false> :P cant be that bad
[17:18:57] <Tom_itx> andypugh, how is that possible? i thought you were a guru
[17:18:58] <andypugh> I can't even nfs mount my NAS (which is what I am sat here trying to do)
[17:19:13] <false> which nas?
[17:19:45] <andypugh> No, I had never even touched Linux until I started with LinuxCNC.
[17:19:54] <andypugh> WD Mybook Live
[17:20:04] <andypugh> It claims to be exporting nfs
[17:20:14] <andypugh> but attempts to mount time-out
[17:20:15] <Tom_itx> i set up a server (with lots of help) but that's about it other than lcnc
[17:20:42] <false> is the nas reachable by other means>
[17:20:43] <false> ?
[17:21:15] <false> a time-out usually means there is not server on the other side. If there was, it would respond with an error
[17:22:08] <andypugh> Yes, I can see it in the various GUIs, and my Mac is cheerfully making a backup to it. I can log into the web interface. It's clearly up and happy.
[17:22:58] <andypugh> There is, however, something odd with the Linux box. For example it can't be founf by name, only IP (for ssh purposes)
[17:23:37] <andypugh> ie ssh andypugh@dn2800.local doesn't work, but andypugh@192.168.0.12 does..
[17:24:02] <andypugh> But as I am trying to access the NAS by IP, there should be no problem.
[17:24:03] <false> check your /etc/resolv.conf
[17:24:19] <false> an also check /etc/hosts
[17:24:40] <andypugh> hmm, what does resolv.conf do?
[17:24:45] <false> resolv.conf should contain your dns
[17:24:57] <andypugh> It doesn't.
[17:25:07] <andypugh> 127.0.0.1
[17:25:16] <false> add a line containing your routers ip
[17:25:30] <andypugh> Actually, that might be right
[17:25:42] <false> save and quit
[17:25:45] <andypugh> 127.0.0.1 pings OK.
[17:25:52] <false> thats localhost
[17:25:59] <andypugh> Ah, yes.
[17:26:05] <false> it is the loopback adress for you box
[17:26:23] <false> add your router, than restart your networking service
[17:26:28] <false> ubuntu?
[17:27:02] <andypugh> Lubuntu
[17:27:58] <false> sudo /etc/init.d/networking restart
[17:28:30] <andypugh> I rebooted, I am curious to see if I also managed to correctly set the default kernel in grub
[17:29:40] <andypugh> ssh: Could not resolve hostname dn2800.local: nodename nor servname provided, or not known
[17:29:45] <andypugh> And the wrong kernel!
[17:31:24] <false> that's because you are not running a local dns, there is no way for your computer to know which adress belongs to that name
[17:31:57] <Tom_itx> did you set up your listen port?
[17:31:58] <false> have you verified that he nas is running nfs
[17:31:59] <false> ?
[17:32:23] <false> dpkg -s nfs-common
[17:33:00] <andypugh> I can happily ssh into all the other machines on the network by name. Including the NAS, the router, the Mill, the Lathe....
[17:33:19] <andypugh> And the NAS claims to be exporting shares on nfs
[17:33:48] <false> are you trying to ping the box you are working on?
[17:34:32] <false> how is dn2800?
[17:34:54] <andypugh> Hmm?
[17:35:03] <false> which box is dn2800?
[17:35:34] <andypugh> That's the Lubuntu box that a) Can't NFS the NAS and b) Can't be found by name by other machines.
[17:36:11] <andypugh> dn2800 is my Linux dev machine. It is sat in a corner running headless and with a bunch of mesa cards plugged in.
[17:36:28] <andypugh> I am sat at an Imac, ssh-ing into it.
[17:37:04] <andypugh> resolv.conf is back to 127.0.0.1
[17:37:10] <andypugh> (on dn2800)
[17:37:20] <false> apt-get install nss-mdns
[17:38:04] <andypugh> I can't help feeling that I shouldn't need anything other than a basic installation..
[17:38:31] <andypugh> It only has an 8GB flash disk, so I am resistant to blindly installing packages.
[17:38:43] <false> these are very small
[17:39:05] <false> it advertises your hostname to other computers, windows has such a component to
[17:40:54] <andypugh> I have never needed to install a package for that purpose previously.
[17:42:32] <false> maybe the component is not running, normally when smb networking is installed this package is installed to
[17:42:46] <false> service nmbd status
[17:43:13] <andypugh> nmbd start/running, process 1133
[17:43:24] <false> nmbd restart
[17:44:10] <andypugh> What should /etc/hosts say?
[17:44:26] <andypugh> 127.0.1.1 dn2800
[17:44:32] <andypugh> Sounds wrong
[17:45:32] <false> 127.0.0.1 dn2800.local dn2800
[17:46:47] <false> then do a sudo /etc/init.d/networking restart
[17:46:55] <false> nslookup dn2800
[17:48:10] <false> also add the following to resolv.conf: nameserver [routerip]
[17:48:16] <false> then do a sudo /etc/init.d/networking restart
[17:48:41] <false> it's not mandatory but good practice
[17:49:49] <andypugh> No improvement.
[17:49:58] <andypugh> I can carry on using IP addresses
[17:50:00] <false> whats the output of nslookup
[17:50:51] <false> of course you can, but something is very wrong here
[17:51:13] <andypugh> ** server can't find dn2800.local: NXDOMAIN
[17:51:25] <false> thats a nice trick
[17:51:43] <andypugh> andypugh@dn2800:~$ nslookup dn2800.local
[17:51:47] <andypugh> Was the command
[17:52:16] <andypugh> It can find localhost :-)
[17:53:30] <false> try looking up dn2800
[17:53:47] <andypugh> ** server can't find dn2800: NXDOMAIN
[17:54:08] <false> again a nice trick
[17:54:25] <false> well lets continue using ip then
[17:54:42] <false> we can try to resolve (no pun intended) this later if you want
[17:55:06] <false> what command are you using to mount the nfs share?
[17:55:22] <andypugh> mount?
[17:55:32] <false> the whole command :P
[17:56:25] <andypugh> sudo mount -t nfs 192.168.0.49:/nfs /mnt/mybook
[17:56:40] <andypugh> Bearing in mind this is _on_ dn2800
[17:58:14] <false> this just times out?
[17:59:13] <andypugh> Yes
[17:59:41] <false> sudo mount -t nfs -o nfsvers=3 192.168.0.49:/nfs /mnt/mybook
[17:59:56] <false> sometimes helps
[18:00:27] <false> is this share password protected bdw?
[18:00:37] <andypugh> No
[18:00:53] <andypugh> Though the uid is defaulted to 500 by the NAS software
[18:00:59] <false> ok
[18:01:14] <andypugh> mount.nfs: Connection timed out
[18:02:21] <false> is there a firewall of some kind running on the mybook?
[18:02:28] <false> service iptables status
[18:02:55] <andypugh> iptables: unrecognized service
[18:03:06] <andypugh> Or do you mean on the NAS?
[18:03:10] <false> yes
[18:03:29] <false> also run cat /etc/exports on ther
[18:03:30] <false> e
[18:03:40] <andypugh> -bash: service: command not found
[18:04:12] <andypugh> That gives /nfs *(rw,all_squash,sync,no_subtree_check,insecure,anonuid=500,anongid=1000)
[18:05:00] <andypugh> It does say" # Use guest user (uid 500) for nfs guest. This is restricted from private # shares by trustees."
[18:07:16] <false> try mount -t nfs 192.168.0.49:/nfs/Public /mnt/mybook
[18:09:26] <andypugh> Wasn't looking any better, and the time came to trigger a "make"
[18:09:37] <false> heheh
[18:10:10] <false> then open another terminal to do you work
[18:11:56] <false> did you enable anonymous logon for the nfs?
[18:12:21] <false> If you do that we can at least determine if this is a permissions problem
[18:14:03] <andypugh> I am not sure what you mean
[18:14:23] <false> you can enable anonymous logon in the nfs share for the nas
[18:14:44] <andypugh> I can?
[18:15:34] <false> If I remember correctly (my brother has one)
[18:17:03] <false> hang on let me look it up
[18:18:49] <false> cant find it, but just thought of something
[18:19:09] <false> on the nas go to /etc/host.allow
[18:19:20] <false> I mean edit
[18:19:49] <andypugh> Aha! http://www.smallnetbuilder.com/nas/nas-reviews/31314-wd-my-book-live-reviewed?showall=&start=1 says that only the default Public mount is NFS-able. (not that we have had any more luck with that)
[18:20:57] <false> correct, but you can add more
[18:21:07] <false> but first you need to get this one working
[18:21:56] <andypugh> says "portmap:ALL"
[18:22:13] <andypugh> Assuming you meant "hosts.all"
[18:22:19] <false> nope
[18:22:21] <andypugh> "hosts.allow" even
[18:22:26] <false> yes
[18:22:44] <cradek> nfs can be done over tcp or udp, and some servers require the mount request come from a reserved port
[18:22:54] <cradek> there is also nfsv3 and nfsv4
[18:23:07] <cradek> you might have devices that don't agree on this basic stuff
[18:23:25] <andypugh> I think you guessed that there is something wierd with udp on this box, cradek?
[18:23:36] <cradek> I don't recall guessing that
[18:23:49] <andypugh> Might have been seb
[18:24:28] <false> yep, already tried the -0 nfsvers=3
[18:24:37] <false> -o that is
[18:24:54] <false> the fact it's timing out bothers me
[18:25:12] <false> ps aux | grep portmap
[18:25:39] <false> on the nas of course
[18:26:29] <andypugh> root 15193 0.0 0.6 3648 1600 pts/0 S+ 00:11 0:00 grep portmap
[18:26:34] <false> also rpcinfo -p
[18:26:42] <false> nfs should be in the los
[18:26:44] <false> list
[18:26:50] <andypugh> rpcinfo: can't contact portmapper: RPC: Remote system error - Connection refused
[18:27:22] <false> sudo rpcinfo
[18:27:23] <andypugh> You that at this point I would blindly type rm -rf * ?
[18:27:39] <false> sudo rpcinfo -p
[18:27:57] <cradek> what os is the server?
[18:28:01] <andypugh> I am logged in as root. sudo didn't change anything
[18:28:15] <andypugh> 2.6.32.11-svn70860
[18:28:21] <false> that is a nice trick
[18:28:25] <false> again
[18:28:32] <cradek> ok linux
[18:28:41] <andypugh> Linux Backup 2.6.32.11-svn70860 #1 Thu May 17 13:32:51 PDT 2012 ppc GNU/Linux
[18:29:04] <cradek> is it running portmap and mountd?
[18:29:12] <andypugh> Maybe not?
[18:29:46] <cradek> it oughta be :-)
[18:30:00] <andypugh> How do I tell?
[18:30:14] <cradek> they are just processes you can see with ps
[18:30:29] <false> ps aux
[18:30:45] <false> ps aux | grep portmap
[18:30:54] <cradek> linux spells it rpc.mountd
[18:31:30] <andypugh> Yes, both are running
[18:31:39] <false> ps ax | grep portmap
[18:31:53] <false> what ip is listed?
[18:32:44] <cradek> sucks that you have both a client and server whose correct operation is unknown
[18:33:00] <andypugh> Where?
[18:33:03] <false> yup, that makes troubleshoot a real PITA
[18:33:15] <andypugh> I am reasonably happy with the NAS
[18:33:26] <false> cradek: that should list the listing ip for portmap right?
[18:33:53] <false> My english is getting worse by the minute, 1:16am over here :P
[18:33:54] <cradek> on the command line? not necessarily.
[18:34:04] <false> on debian it does?
[18:34:48] <false> andypugh: look at /etc/default/portmap
[18:36:54] <andypugh> OK, I can mount the NAS Public directory on my Mac with no problems at all
[18:37:31] <andypugh> false: Only 00:22 here :-)
[18:38:30] <false> Ah this is starting to get interesting
[18:38:50] <false> but thats not via nfs right?
[18:39:25] <andypugh> The command was sudo mount 192.168.0.49:/nfs/Public ./Resources
[18:39:49] <andypugh> If I can figure out how to unmount it I can try again with an explicit -t nfs
[18:40:35] <andypugh> On the NAS everything in default/portmap is commented
[18:40:50] <false> umount
[18:41:31] <false> umount /dir of course
[18:41:50] <false> Never got why they didnt name it unmount
[18:42:31] <andypugh> Yes, -t nfs still mounts nicely in the Mac.
[18:43:45] <false> ok then run the following on the dn2800 box
[18:44:26] <false> sudo apt-get install nfs-common
[18:44:43] <andypugh> Already did that hours ago
[18:45:08] <false> of course you did im sorry ;P
[18:45:16] <andypugh> nfs-common is already the newest version.
[18:46:06] <false> dpkg reconfigure portmap on dn2800
[18:46:54] <false> dpkg-econfigure portmap
[18:47:05] <false> dpkg-reconfigure portmap
[18:47:07] <false> damn
[18:47:16] <andypugh> Hmmm.. Package `portmap' is not installed and no info is available.
[18:47:51] <false> apt-get install portmap
[18:47:59] <andypugh> But: "Note, selecting 'rpcbind' instead of 'portmap' rpcbind is already the newest version."
[18:49:06] <false> rpcinfo -p
[18:49:42] <andypugh> Strangely, the Mac can see a share called "dn2800" though it does continually drop it.
[18:50:20] <false> there is something very strange going on on dn2800
[18:50:54] <andypugh> http://pastebin.com/zCxzBNRx
[18:56:27] <false> You are very sure there is no firewall the dn2800?
[18:56:32] <andypugh> No
[18:56:48] <false> service iptables status
[18:56:52] <andypugh> I have no idea if there is a firewall on dn2800
[18:57:15] <andypugh> iptables: unrecognized service
[18:57:22] <false> thats a no
[18:58:24] <false> rpcinfo -p [ip of nas]
[18:59:59] <andypugh> http://pastebin.com/5967Ugsx
[19:03:24] <false> last shot
[19:05:06] <false> mount -t nfs -o timeo=50 nfsvers=3 ip.of.nas/nfs/Public /mnt/mybook
[19:05:30] <false> don't know about the CAPS Public
[19:05:44] <false> use the same url as you did on the mac
[19:07:36] <false> if this doesn't work we can try to force tcp connections only
[19:07:47] <false> and after that I am out of ideas
[19:08:10] <andypugh> That worked
[19:08:16] <andypugh> :-)
[19:08:20] <andypugh> But why?
[19:08:41] <false> the problem lies with the server not responding fast enough, the default timeout for this is 0.7 per command
[19:08:49] <false> now its 5 seconds
[19:09:41] <andypugh> Is that a sticky setting?
[19:09:45] <Tom_itx> the woes of technology
[19:09:54] <andypugh> Because now it works without the -timeo
[19:09:55] <false> no you need to put it in your fstab
[19:10:20] <andypugh> But previously it was taking about a minute to timeout..
[19:10:58] <false> It times out trying to connect, the timeout we set just now is the timeout for the client to wait for a server response after a command
[19:11:44] <false> So if the client ask to connect it waits 0.7 seconds by default, if this doesnt work it tries again for 60 seconds or so
[19:12:17] <false> so if the server doesnt respond within that period the connection is never established
[19:13:03] <false> just put it in your fstab like this
[19:13:39] <andypugh> 192.168.0.49:/nfs/git /home/andypugh/git nfs rw,hard,intr,timeo=50 0 0
[19:13:45] <andypugh> ?
[19:14:01] <false> yes and add nfsvers=3 to be sure
[19:15:09] <false> Have fun nfsing :P I'm going to sleep, I need to get up in about 3 hours :(
[19:15:36] <false> you pulled me into it ;-)
[19:16:49] <andypugh> Ohh, sorry.
[19:17:06] <andypugh> I can still use flexitime to get a full night of sleep :-)
[19:17:27] <false> No problem, glad I could help
[19:17:38] <false> You lucky bastard
[19:17:39] <andypugh> Thanks for the help.
[20:15:46] <Tecan> i should fancy something up