That drive there in the video is a 6kW drive...
Endeavour: looked at turning off power savings stuff in bios? http://wiki.linuxcnc.org/emcinfo.pl?FixingSMIIssues
if this applies especially, don't know much about i7
andypugh: 6kw would be plenty to push Z and huge in comparison to X
What sort of power do you need? I suspect that a proper drive with current sensing has to be a better plan for a serious machine. The Mesa 8I20 is 2.2kW (400V, 24A) and might work.
andypugh: I'm running Ubuntu 10.04 with EMC2 installed.
[00:06:17] <andypugh> http://wiki.linuxcnc.org/emcinfo.pl?RealTime
Specifically the bit about isolcpus in grub, though that is old, and based on grub, not grub2, and uses the term "both cores"
athe 8i20 ought to work for what?
none of them are that small (2.2kw)
Mine is only 6kW on paper, I suspect, being 600V and 10A.
i had looked at it, honestly, i would love to mod our sharp to cnc, just because we could
The 8I20 uses a 30A 600V module but even Fairchilds data lists it as 2.2KW
Sharp, another very Bridgeport like kneemill
how does 18kw -> 2.2kw work? extra safe stats?
I guess 30A or 600V, but perhaps not at the same time, all the time.
[00:11:27] <andypugh> http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=6880751
Is the 16A version of what I am using, but it is a component, not a drive.
400V max for short circuit protection, 15A max at 100C Heastsink
That claims 2.2kW motor power too.
3X normal current needed for servo peaks so 5A RMS at 340VDC (240*sqrt2) = ~2 KW
i wonder what size servos our small mill has, when they rip out the messed up ballscrew(s), i'll check
Like a car, 140mph, 50mpg. Just not both at the same time.
pcw_home: i never asked what is delivery usually on 5i2X and 7i33?
you could compare it to a lathe motor, 127ftlb and 4500rpm, just not together :)
i realized i assumed it as stock and never asked
i have so much backlash (and play from somwhere) the OD chucking lips i milled today had 4 steps at the transistion points, i can't wait for them to rebuild X+Y (and will talk them into looking really good at Z and not forget about it)
I like the air conditioner picture in the IRAM page
I think thats why they are so cheap (millions used in ACs)
pcw_home: 8i20/8c20 are not on pricelist because you don't stock them/are still new/phase them out?
IR have a driver they reckon can do a 3.3kW motor (and then you are into seperate IGBTs, though they go up to 1000V 450A, which sounds like plenty.
[00:21:29] <andypugh> http://www.irf.com/product-info/datasheets/data/iram136-3063b.pdf
Still new and we had lots of trouble getting the firmware solid
And no EMC2 driver, as yet.
Fairchild has a 75A module (37A at 100C Heatsink)
That sounds pretty useful
i was just about to ask if you thought you could build it bigger
I can't believe Fairchild is still in business
Yes no EMC support, had to figure a way to hide the serial protocol
Actually, looking on the IR website the 2.5A module at $2 each looks useful too.
Sure for little stuff
pcw_home: so 5i20+7i33+8i20+8c20+propersizeservo and we would have the same as 5i20+7i33+yaskawa_drive+servo?
Fairchild has some really nice stuff
in axis sim after I run a program how do I rehome? I keep getting this message "Cannot home while shared home switch is closed"
more like 5I23/7I44 or 7I52 and 8I20
5i20 + 7i33 + 8i20 + EMC2 driver is possibly the way to go.
pcw_home I have their TTL Databook on my desk from 1978
(Actually, what _is_ the 7i33 in that chain I just typed?)
(no 7I33 needed), 8I20 is all digital
pcw_home: 8c20 needed?
pcw_home its a classic!
Fairchild made the first smart TTL chips
no 8C20 needed
did Gordon Moore come from Fairchild or was it the otehr way around?
did some leave Intel to found it
7I33 is analog servo interface (4 +-10V analog out + 4 encoder ins)
I suspect that there is a good chance of an EMC2 8I20 driver happening eventually. Probably quicker if someone like you actually needs it.
"The 8I20/8C20 combination adds the features of analog input velocity and torque modes, step and direction mode, quadrature input mode, smart position control mode, and four isolated inputs and outputs.
" That's why i thought 8c20 was needed, this is when using it outside of EMC then?
I'm going to get buried with this Databook
Probably from (wasn't Fairchild one of the Shockley pieces)
pcw_home I sort of remember some tie in between the two
Yes 8C20 is for standalone
could you give me an idea of what a 8i20 costs, not something officially quoted :)
You would move all the clever into EMC2/HAL
8I20 itself is just a 3 phase current mode amp (which is al EMC needs since EMC can close the loop)
andypugh how did your 3 phase motor driver ever go?
Did you not see my video?
andypugh can't on this machine
It on our web page (or maybe the pdf price list) I think its 239 Q1 maybe doe to 149 or so in large Qtys
Linux cnc1 2.6.24-16-rtai #1 Tue Sep 30 22:54:33 EEST 2008 i686 GNU/Linux
OK, well, it works pretty well now. At 300V the motor flies
/exec -o uname -a
this isn't exactly multimedia paradise I'm afraid
I with shit flat screen did higher resolution
whoops missed my homes
I wish this flat screen did higher resolution
I can't see it on the web price list
must only be on the pdf
pcw_home: would the performance of that combination be comparable to 5i22+7i33+manufacturer'sdrive+servomotor? or is there anything to be aware of
And you had the pricing right
never noticed the pdf link below it, sorry pcw_home
I jogged the tool all around why are my home switches in sim still "closed" ?
It might turn out to be better. Or it might not, it depends how good the manufacturers drive is, I would guess.
Should be, but since its a torque mode drive you may need to run EMCs PID loop at 2-4 KHz instead of the default 1 KHz
pfred1: I doubt that sim has home switches.
pcw_home: now comes my real question, how likely is it mesa would be willing to build a 8i20like one based on the large version you mentioned a few lines up?
andypugh well it won't rerun a program
Never used sim, so no real idea.
( though torque mode is nice because you have direct load monitoring)
andypugh when i open it it homes and touches off just fine but if I try to rerun it refuses with the Cannot home message
and when I search google its just noise returns
281,000 pages of nonsense
hello how can i make my own GUI
pcw_home: and sorry for that awkward grammar
Its scalable so we could make It bigger but we'd like to get a lot more experience with the 8I20 before we venture into bigger drives
morfic get a soldering iron
Does anyone here think it's worth my time trying to learn Inventor? I got a freebie copy.
It's veru nice software
I've heard I should jump right to SolidWorks.
I've also heard that SolidWorks is like going swimming in a computerized piranha tank.
It can't suck any worse than SketchUp.
I used Inventor 8 hours a day for 2 years and still like it.
I just learned image-to-gcode I think that is about as advanced as I'm going to get for a while
spasticteapot: i would say it is, guy at work was just good at figuring out, but now he has a good foot in our company
No comparison, Sketchup is horrible for engineering
I'm hoping to bribe Alex into doing a bit of machining for me, and I need to get the 3D models done lickety-split.
i am impressed to see him animate his designs and check functions, and to me solidworks got nothing on it
xie: Do you intend writing a GUI from scratch, or do you just want to add buttons to the existing one
spasticteapot rushing with the unfamiliar is a fair recipie for disaster
Is there a recommended book for it?
pfred1: I'm going to pay a machinist I know to fix the CAD information before I send it out.
I'm mostly trying to figure out how it will look.
Just go through the included tutorials. That's all I did
I'm rubbish at CAD, but much better at paying other people to do things for me. :P
you and me both
spasticteapot: i don't get to use it, so i have no idea, when i watch the puerto ricans model something i need, or boss needs and ask them to give me dimensions or have them check their model if my program and their model don't agree, that's when i interact with inventor
Just realise that you can't draw anything until you model it.
since I switched back to paper and pencil I've come up with much better solutions than I was coming up with trying to be a CAD
But for thinking in 3D it is way cheaper than aluminium :-)
pfred1: I have pen-and-paper pictures. Sadly, my grasp of perspective drawing is a bit limited when it comes to complex surfaces.
is there someone hellp me i am new user
pcw_home: we do a lot of learning on this emc build/retrofit already, was wondering if i could propose it to my boss, mainly because we would deal less with "oh i can't tell you how our magic works" from a big guy, and would get more of the "off the shelf, non specialized" parts my boss was focusing on when he proposed the project
Inventor lets you try out whole assemblies, spot the problems, fix them, try some more. Of course all the simialr packages do that too, but I have found the Inventor assembly constraints to be a lot more robust than some others (like the 8-ball-joint shaker I designed)
andypugh you can always think in styrpfoam
i want to make own GUI
axis is OK
It's a rather big project for a newbie
if i culd ever get it to home!
pcw_home: my thought is that 8i20 would be interesting if it "made us more free".....no replacement motor from company X in stock, we call up Y and Z and get theirs, without the "but you need our drives too"
If you just want to add buttons to Axis, then it is easy, but writing a whole replacement GUI is a huge task
andypugh: In this case, I'm modelling a lumpy block of Corian.
Not terribly exciting.
yeah a new interface isn't my idea of a beginner project
spasticteapot: Model the block, add on the lumps, job done.
Grrr! with this Cannot home message!
I have it right on the line it originall homed on with no problems are my simulated home switch springs broken or something?
andypugh: In this case, it's more along the lines of "create intersecting spheres."
We look at it that way as well, by putting most of the motor specific setup in EMC, almost any motor type can be driven
(synchronous AC servo motors are pretty much all the same) so especially for retrofits it good to have one type fits all on the drive
That's fairly easy. Make them in Black Chrome, that texture is a Dodge Viper reflection.
spasticteapot a sphere was the only thing I was ever able to create when I tried Blender once
pfred: They're funky spheres.
andypugh: sometimes it seems to be much harder for them to revolve a hole pattern than it is for me in Esprit, wish i could recall the kind of pattern, i think it's irregular stuff, where i start with a row of holes at 0deg, then copy n-1 4.5deg and then translate, while this sounds cumbersome, i beat them at modeling that every time
In this case, I'm trying to create a hemisphere of an oblate spheroid.
pcw_home: hm, how do you feel about putting a 8i20 on a fanuc motor, sounds like lemon squeezy? ;)
* spasticteapot is building ridiculously complicated speakers because he can.
And because, hopefully, they'll work well and look snazzy.
spasticteapot I've built a lot of speakers and one thing I learned is they all sound better in any enclosure and I actually prefer boominess
as opposed to "optimal"
pfred1: I'm still something of a newbie to speaker design, but I'm going kinda nuts on this design.
Active crossover, bi-amping, low-diffraction enclosure....
speakers with the right volume sound too flat to me
They sound better in ANY enclosure?
reminds me i wanted a new cabinet for my bamboos (fullrange)
oh no doubt you cancel out the back waves
You're referring to OB use?
morific: Tang-Band 4" units?
a speaker on a flat surface a raw speaker that is sounds like next to nothing
morfic: Well if you ignore their encoder I'm sure its possible
the main difference that make drive incompatible is Hall patterns
and or different schemes of determining startup rotor position
put it into any kind of a box and it is night and day
(all basically encoder issues) the motors are all much of a muchness
Not exactly a fan of open-baffle speakers, I take it?
I like ported speakers
get some MDF and make random speaker boxes
pcw_home: so we would want to have a way to have encoders outside the drives, some w/o undocumented magic?
spasticteapot: yes, tang band 4" bamboos, the speakers they are in are WAY small, i need a little more volume to get rid of that nasty boomyness
morfic life begins at 12"
Yes. I think most have standard encoders with other gizmos added (or resolvers)
I've had 8s and 10s there's nothing quite like real 12s
spasticteapot: i try to level out the low end with jack-rack eq plugin, which would work better in a case where i can let the speaker move
morific: Have you tried WinISD?
Also, there's a lot of transmission-line and horn designs that can use that particular driver - I'm not sure of the Qts, though.
what is it with people adding 'i's to my name (to explain: people do it to my last name, even after i spell out all 6 letters to them :/)
pfred1: Eh, it's a question of amplitude. 7" is just fine for a nearfield monitor.
There are only 48 hall sensor patterns, it is a manageable problem.
spasticteapot to get a "live" sound it was made with 12s so nothing but 12s are going to reproduce it
A really good 8" woofer can reproduce 20hz. (Admittedly, that's a pretty unusual woofer).
You can't really hear below 30hz.
So what's the point?
to feel the chest punch?
Yaskawa is interesting, they use a standard quadrature encoder but with a battery backed up CMOS counter
that remembers how far the encoder is from index, they transmit the absolute position to the controller serially at startup
pcw_home: I could see that causing problems if the battery failed.
i want some decent sound on my desk, those 4" produce some nice sound playing Rasputina's Tourniquet
one of my systems is a Sony 270 watt amp with an old Samsung 45 watt for the back fill
with 6 speakers
morific: Sealed or ported?
I can hear it when i go get my mail
pcw_home: you know a lot about yaskawa, probably more than the distributor we try to get our stuff from
thats the other thing if you don't have the room for it yo ucan never reproduce that either
just those tangband in a way too small enclosure
that system is in a 20x20x12 space
so it gives a nice small club sound
buddy of mine has a 1200 watt mixing board and 4 CV club monitors its awesome
pcw_home: This "serial". Is it on a serial protocol, or just a burst of quad pulses to get the counter to the right position (the latter is what my resolver-quadrature Arduino thing does)
morific: Which tangband?
andypugh you can also use Hall and sinusoidal (Hall state gives approximate reference angle, enough so that servo control works: 87% torque)
once you hit an edge you can lock in an accurate reference angle
Endeavour I thought this was for my axis sim home problem
morifc: The 4" one?
[01:06:57] <spasticteapot> http://www.parts-express.com/pe/showdetl.cfm?Partnumber=264-847
(truely serial, specs are available from Yaskawa)
pfred1: Hah, sorry.
pcw_home: are you aware of something that lets you drive a (large) AC motor like a servo and when needed position like a C axis?
Endeavour ah thats OK
You mean induction motor?
Endeavour yeah that is crazy I get 15,000 ns on this wreck of a 1GHz PIII
spasticteapot: yes that driver in http://www.overstock.com/Electronics/Teac-Book-Shelf-Wood-Trim-Speakers/2929384/product.html
that enclosure (tweeters not in use and black)
Endeavour: The mailing list is a lot better than the forum. There is only really one person actually answering on the forum, and that's JT (I think)
Yeah, that'll cause problems.
BTW don't try hardware accelerated GL on ATI cards it locks the system up
morfic: Are you using a sub?
I had it all working here soon as I fired up an rtai app like EMC it'd die
spasticteapot: and no existing cabinets work for me, they all get deeper and taller, i would prefer to just get taller or much more tall and little wider/deeper, brings them up to ear level at the same time, does not affect desk space
Outside perhaps of a big horn, you won't be getting any bass.
In a sealed box, pretty much none at all.
spasticteapot: no, i might be happy in a larger enclosure, it sounds ok, perhaps i don't hear as low as i should
pfred1: That stinks.
Endeavour yeah for a bit there it almost looked like I could have lowered my latency to 12,000 but then it spiked to about 18,000
I think its possible, at least I've seen papers on it (its tricky not to lose rotor flux at reversals)
I have no idea why my latency is as high as it is.
It's a bit of a kludge, but it might work.
Endeavour have you tried a live CD release to compare?
Or you could sell them (not hard) and build some proper speakers, like the Overnight Sensations.
6" wide, 9" high.
spasticteapot size is everything with speakers
mo biggerer is mo betterer
spasticteapot: see, too deep
Right, time to sleep. Tomorrow I think I will try the induction motor as servo idea, I am curious about how well it will work.
there is no upper limit to that truth either
[01:13:31] <spasticteapot> http://sites.google.com/site/undefinition/diy-overnightsensations
<pcw_home> (truely serial, specs are available from Yaskawa) <--- makes me feel a little better about yaskawa, not as secret society as Fanuc
pfred1: Won't install.
Endeavour if it was me I'd try just a plain EMC and see what numbers yo ugot might narrow it down between hardware or software
Endeavour why not?
* pfred1 runs HAL Latency Test ... 13309 so far
spasticteapot: rear ports also don't work too well for me, speakers are way back to wall, sealed is much more practical and there is no port that's wrong and starts blowing :)
morfic: Here's a taller, shorter version:
pfred1: I don't know. Fails with regularity.
[01:16:01] <spasticteapot> http://977414429758362792-a-1802744773732722657-s-sites.googlegroups.com/site/undefinition/WWWJDOS-frontfull.jpg?attachauth=ANoY7cpa1tnCKW9cbQRlVCWui8FJ3Qu31NsWQhlcNb_I10PendJqqEGQjsEPBd6G0t8-BGXJ4xGTN-C21v2gWNE8sH-v6VwccYQ--dPODzK6Cd0x3CqmDTYVE6fYYwUbtL3TgCmgqRVn37H9oOThOWJky-peV0jB8q_g8O7f03HWp6rTr4uBGLyZ4d8ySO4efvmuz5JpO5nEn0MQD14dBGzjVEKBeHfD-A%3D%3D&attredirects=0
Won't install and always throws me to initramfs
Endeavour has to be something going on with the hardware then
because the EMC live CD works
morfic: If 8" is too deep, you could make them shallower and taller.
Ubuntu 10.04 works well, I can't get any other Ubuntu to install.
even though mine claimed it had "One Errors" (sic) I installed it anyways
Endeavour maybe earlier kernels didn't have support for your mobo chipset or something?
It won't let me. Always goes to initramfs
Possibly, pfred1. I might get this board: http://www.newegg.com/Product/Product.aspx?Item=N82E16813121399
yeah mobo chipset is just my wild guess ATM
jitter is steady like a rock tonite
shot up to 13309 and hasn't spiked past it
I think i use 15500 in my config just to be safe
Endeavour what kind of a machine do you run with your EMC?
ha ha dragging a window around shot my Base Thread up to 14398 this gfx card sux
Endeavour everyone raves about the Atoms here
wow this is a combo with the CPU too?
wait a minute this thing doesn't even need a CPU cooler?
they should just dip it in epoxy with a power supply and call it a system or something
Also has a MiniPCIE slot
what do you do with those? is that the new gfx slot?
I heard about them then I forgot
spasticteapot: doesn't look bad, the closest i found was a TMM curved cabinet, so yeah i like em if footprint is ok (that was a prime example for why tiny url exists btw)
it'd be cool to boat build speaker cabinets foam gel coat then glass it
electronics just gets cheaper and cheaper
I can't believe this screen I picked up at a garage sale 2 weeks ago an Acer AL1916 for $5
morfic: 6" x 8" footprint for stock version.
You could downsize it to 6" x 6".
Best part is, the whole parts cost is about $85 - you could probably get about $75 for your fullrangers if sold secondhand.
Those TB drivers are really much better suited to being midranges.
I picked up a pair of junky old yamaha computer speakers about a month ago for a dollar
Yamaha made some surprisingly okay computer speakers.
when they were new I wanted a pair so bad good thing they were out of stock and I settled for the Jensens
well not these they have one driver in plastic boxes
I had to take them all apart to chande the power jack because for a buck I didn't get the wall wart with them
and in taking them all apart i saw just what crap hey were
pfred1: Picky picky.
yeah they're better than other cheap PC speakers I've seen I guess
not as nice as this pair of Jensens I have though
Like the old speakers from the 70s?
they at least have 2 drivers per box
One driver can work surprisingly well, providing you don't need much output.
I guess Yamaha thought so
but hey for a buck I got the speakers I always wanted
when I go garage saling I keep an open mind
people ask me what are you looking for? I always reply, a good deal!
I got a backup PC for EMC for free this season
it was in a pile and I asked whats up with that they said want it take it
nice corporate model
has little rubber booties over the unused power connectors inside it and stuff nice touches
Jymmm: APB for Mr Padnos?
morfic: Yeah, SHOOT ON SIGHT!
its always the damndest things that take forever
i have to mill out like 16 1x2x3 inch steel blocks all accurately to the same size
..seems so simple - but not with a chinese milling vice
buy 2x3 cold rolled bar, tollerance may be suitable for your application, saw 1" part and mill, 2x1 and cutting 3" would be easier i guess, i was more thinking EDM on the first bar
yeah i was thinking i should have done that in the first place - first mill two dimensions out of a long bar - that would have saved megatime
you cut them already?
madsci44: one dimension to mill :)
:P silly mistake.
yeah i had a bunch of pieces that were close - leftovers from other jobs
that is different
not much you can do about that
might have taken less time to melt them down into a bar hehe
left overs are as much a curse as that welded piece after someone has scrapped it, doesn't matter how much 316 likes to move, doesn't matter how often we saw it not work out, if it's friday and it could mean another shipment, we weld it, put it back on lathe, creep in, cut the weld, just to find it's not cleaning up or opening up some surprise elsewhere
well future practice for me - if its several of something forget the leftovers and start with bar :)
im going to try a torque wrench for tightening the vice consistantly
the flex is messing with my square
got what i paid for there
the moving jaw is rising up?
do you have round bar stock? 6" wide to cover the width if the vise?
yeah using a piece of ground stock for the rolling side - its the fixed thats moving
its tight but I suspect the vise bed flexes
getting about 0.0005 to 0.0015 out across the 2 inch dimension
madsci44 thats great!
hah hi fred! i wish.
madsci44 what do you wish?
i wish that was great :) but for my purpose it needs to be closer to 0.0005
smallest increment on surface grinders I've run is .0005
and 0.0015 is almost imperceptible to humans
its not imperceptible to machines tho :)
well i don't need to impress machines
depends what the part is for - where its goig what its doig
(keyboard also made in chia)
I don't expect accuracy any better than .003 out of a cutter machine
not to say it cannot be done just I don't expect it
well its good to design things not to require it where you dont need it
to me 0.001 in milling and turning is 'approximate'
I expect .0005 of raised bur on a cut surface
madsci44 why exactly do you think they make surface grinders/
because shops have floor space they need covered?
i expect to read my micrometer in the reflection :) (im joking somewhat)
nope - when you have to remove less than what is removable with a cutter - and/or what is too hard to remove with a cutter accurately - thats to me a surface grinder job
if it mattered we surface ground everything
yeah i should get me one of those
anything we wanted closer than .003
madsci44, so did you do it?
finishing another job before I clean it off (i want to recycle the chips from the bushing job)
madsci44 have yo uever worked in a commercial machine shop?
I guess production is a little different than doing one offs
and i have to check if my camera is coolant proff :-)
its alot different - specially if your always prototyping and the design is evolving
we'd throw out a half a dozen pieces doing some setups
our Production is, on a lathe, +/- 0.0003"
and also especially if job has to fit resources vs the other way around
mill, +/- 0.002"
elmo40 and how long will your lathe hold that tolerance/
we make airplane parts :)
our newest lathe is 4years old!
pfred1, .003 on cutting machines?
alex|lappy repeatable sure
they run ~18hrs/day 5days/wk
that is.... a little loose.
as cutters wear deflection changes
at least for cnc
cnc cutters are magical?
pfred1, no, but they are more reliable then us squishy fleshy bits
also predictable in a production environment
cut the same pocket 10,000 times I can guarantee you over the run the tolerance will change
after 10000 times if you track it you can measure and compensate in advance
we'd just blow up bits and change them
when they start counging up sand you know its coming soon
ah, so maybe instead of saying you only expect them to be within .003, you should say "within .003 over the entire life of the cutting tool"
alex|lappy its a lot easier to say .003 and leave it at that
its not the only variable
yes, but it is a silly statement without the qualification
bar stock is only so accurate etc
why leave out large amounts of relevant information?
instead of broad generalizations?
in a commercial production environment you're wasting money trying to get any closer
keep on operating like that and you won't operate for long
it all depends if your making doorknobs or your making aerospace components,
no it doesn't
the machine doesn't car if the knob is going on an outhouse or a space ship
what matters is cost per operation
i sure would not want a bearing journal to be out by 0.0005 - and its faster to get it on in the lathe than it is to remount for a 2nd operation in a grinder
that is coming out of your pocket regardless
unless its for a go cart or something
I'm not so sure I'd want a machine soft bearing journal I think I'd want it heat treated after roughing
madsci44 thats the other thing BTW stuff with tolerances that tight are often treated
we heat treated everything we made
well yeah perhaps - but they are often not as well - which goes back to the point - you worry about it where it needs to be, and depends on the job, the application, the part, etc
well we made tools so
they're better harder than softer
after I mill a piece i can knock it down with some wet or dry sand paper and it'll change a good .0005 pretty fast
as high spots get sanded off
we'd even do that with surface ground stuff polish it
though it changes a lot less
once you get up to a polished surface you can carry on about .0005
but I've had this conversation so many times it bores me
madsci44: I have a fix for the g92.2 problem you reported. jepler and I are working on testing it well enough to put into the stable 2.4 branch.
madsci44: if you want to try it: http://timeguy.com/cradek-files/emc/0002-Save-state-of-G92-correctly-across-restarts.patch
hey thats great!
cradek you wrote image-to-gcode right?
thank you I will try it out
pfred1: the one currently included with emc was written by jepler based loosely on an algorithm I had written earlier
so, yes and no
cradek well hey good job the both of you!
took me a bit but i finally made something like I was thinking about with it and it rocks
neat. what kind of thing did you make?
cradek you wrote axis too didn't you?
cradek oh just the gnu bull so far
it was mostly written by jepler based on a design I came up with (haha, this sounds familiar)
cradek in the sim after I ran one file I could never get it to home again any idea what I missed doing?
try jogging left
it keps erroring saying home switches were closed
oh i jogged every which way
sim/axis has simulated home switches and you're on one of them (you can'st see it so it's kind of baffling)
put it right back o nthe starting yellow line
you name it it drove me nuts!
I'd put it exactly where it started off still no dice
yeah I think jogging left is the key (or just home once right after you start up, and leave it)
ah OK home once
I was sitting there cutting my gnu beast and just wanted ot you know cut it twice
it was pretty neat though I think it'll be all i ever use to make stuff with
SWP_Away is now known as SWPLinux
cradek: applied patch and rebuilt - tested several combinations of offsets in effect restarting, reloading, running program, opening emc with and without offsets in effect - behavior is PERFECT.
that's terrific, thanks for the careful testing
that helps me feel better about putting it in 2.4
yw thanks for the fast fix
i guess in linux kernel 2.6.34 and beyond there is no more CONFIG_PCI_LEGACY and hence no more pci_find_device
(no wonder i couldnt find the prototype anywhere)
[07:53:24] <alex|lappy> http://www.google.ca/search?hl=en&q=define:+adun&btnG=Search&meta=
google is fucking amazing
err, wrong channel
although still cool!
if you are a starcraft nerd.
my parents had a tent trailer made by starcraft :)
madsci44, not quite what I was getting at
it was quite nice
micges: i figured out my compile problem with the drivers
madsci44: go on
the CONFIG_PCI_LEGACY is gone, so is pci_find_device and so on, as of the 2.6.34 kernel -
i guess thats supposed to "encourage" the driver authors to switch to hotplug api
i guess i must be only one crazy enough to have tried 34 :-) on the upside besides the drivers, and the version.h issue, it works great
it has been a while since I wrote any kernel drivers - like since kernel 2.4.xx - i thought if i could find the prototype for the function maybe i could come up with a solution - then i discovered there was no prototype in my kernel tree,
then i checked an older kernel tree and there it was
ah linux nerds, what would we do with out you guys?
reading MAN files so we don't have to
info is more fun to read
they should have named the man system Working Manual, then we could rather be reading woMan files
man touch is gay
madsci44: those are very important informations, thanks
im very glad if they are helpful
fortunately for the moment the only rtai patches for 2.6.34 are still in magma so I guess there is some time to upgrade the driver code
i need a cheaper hobby
my power supply arrived, but i am a 100 miles away
damn fathers day
i wanna play with toys
* micges away for breakfast and coffee
that face cutter is huge
slow reader tho
you could probably visually identify the tool quicker, or put fingers on the existing reader slide to feel the shape
it could feel it for sharpness and chips etc too :)
nice to see it working tho
Actually the reader reads at full track speed, look like it slows down to index (maybe too soon)
slows 2 before it needs to
but what is the speed of an actual tool change?
I don't think he has the arm going yet
has that been programmed yet?
prep time does not matter because you can prep ahead of time
but total tool change time matters
I'd be amazed if the arm takes more than 4 seconds or so
moving into position and starting/stopping the spindle probably take more time than the changer arm does
but I'm guessing - I've never seen his go
4 is a long time :P
ours is 1/2 second
Would be great to see that old machine doing real work. I wonder if thats the oldest NC-CNC conversion with EMC
oldest machine I mean
pcw_home: my NC lathe was '74 but his has that beat by many years
I wonder what they are asking: http://tinyurl.com/COBRA-42
they do too
wow, SO over priced http://tinyurl.com/2wyjhjq
Samco's machine is late 50s/early 60s ISTR
damn! found a good score... minor repairs http://tinyurl.com/37hnhuj
looks a little on the big side
should fit in the bedroom no problem... I can sleep on the couch :P
doesn't seem to have live tooling, though :(
what are we talking about?
I thought so... but I wanted to make sure...
actually - I think it would index correctly at high speed.
but it is a little less scary when it slows down for the last few
skunkworks: have any progress on the arm and drawbar yet?
no - that will be the next thing probably
it is pretty basic
I think just a bit less complicated than the chain
a good portion of the motion is hydralic - I think there are maybe 2 limit switches
* skunkworks wonders where in the world is cradek
hmkm best to use smaller or larger tools when milling alu a bit too fast?
small = more heat right?
skunkworks: sounds easy - you just turn it on and then wait until it hits the switch and turn it off?
frallzor: what do you mean a bit too fast? feed or speed?
i think of it like making fire with a stick, small = more heat at a point
cradek: more like - turn on this - wait for that - turn on that - wait for this
smaller tool reduces your surface speed
kinda makes it hard to fail
so far my small tools have been killed =)
works fine to a certain degreen, then they die when it starts clogging up
what size of tool are you talking about?
last test = 2mm
ehm red spirit I think its called in imaginary english
unless your spindle is amazingly fast I don't see how you could spin a 2mm tool too fast
so its a bit too fast :P
that's only 300 sfm which is fine for aluminum, even with HSS
but it gooks up
so it cant be fine
and i allmost drowned it in T-red
you have to keep it cool and let the chips get out
no idea what T-red is, sorry
t-red i ethanol
what kind of aluminum?
free piece I got
you're not cutting 4mm in one pass with a 2mm tool are you?
0.25mm this time
cradek: do you see any reason why iocontrol.0.tool-prepare would go high at estop?
well, I didn't look, but that seems weird
like I say - it is an internal signal from emc.. I don't see anything that would activate it.
im going to try a 4.5mm tool with same settings
frallzor: I'm not very experienced with stuff that small, but when I do tiny cuts, the trick is spin fast, feed slow, and keep it underwater (soluble oil)
slower feed, right Ill try then
are these 2 flute or 4?
ill try lower speed, more coolant and new tool
maybe 5-10 ipm? is your spindle good or is it a wobbly dremel?
its a good one
definitely keep it underwater if you can
10ipm? :P i was thinking like 100 ipm =)
wd40 works ok too
100 ipm is like .002/tooth - I doubt you can do that for long with a 2mm tool
well my last test was like that actually
how long did it last?
it lasted the whole work until the end when it stuck
but the result was cloggy
but the cuts were very nice before that
hmm, maybe it just got dry at the end then
4.5mm test is done now and im about to try it out
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2010-06-20.txt
it was cool to see the numbers flash by in ladder :) (that is how we found out what tool where in the chain already ;)
you can see at the end of the video - the tool numbers flashing in the variable display
at 21k rpm and .001/flute you would get 40ipm (that's figuring at 1000sfm for a carbide endmill)
thanks alot cradek....
did what you suggested...
broke another one? :-)
nah perfect result
besides a spot where I missed coolant =)
amazing quality of the cuts, im impressed
get a oil mister to follow the spindle to avoid missing :)
think is, MDF table
so have to cool with care
pic otw =D
[15:26:20] <frallzor> http://pici.se/p/nwaDPOsuH/
top right of 2nd and 3rd
there i missed some =)
but other than that Im satiesfied
wrong spelling but hey, im too happy to fix =)
but some sort of miniature flood coolant would be nice
SWPadnos_ is now known as SWPadnos
(10:49:39) cradek: you have to keep it cool and let the chips get out
don't think fluids are the only solution
compressed air with one or two of these http://qurl.org/D11
will do just fine for 4mm thick sheet Al
true, I hear air also works if you have it (but the pedant in me says air is also a fluid)
and you need lube to stop ally from sticking to a tool
need the proper chip clearance and grind angle of the flutes
you could mist lube with the air
flooding is so 1990's :P
I've been tempted to try air/lube mist but it seems like you'd end up breathing a lot of it, no?
btw - the tool chain should pre-fetch the tool before it is needed. So speed really shouldn't be an issue.
[15:43:57] <frallzor> http://pici.se/p/large/DDoOOMxyN/
decent cuts too =)
cradek: if you don't use a purpose built nozzle sich as those Exair ones ( http://qurl.org/D11
) then yes, you will breath a lot of it in. Those nozzles don't 'shear' the air when it comes out. Plus, they don't need high pressures to do the same job as an open pipe.
frallzor: nice cutting :)
frallzor: that looks great
I see a slight ripple effect on the sides. is your tool only 2-flute?
as long as I cant feel it It doesnt matter, allmost as smooth as a babies butt :P
medium helix 4-flute would make a very smooth egde for you, if needed, that is ;)
you should be able to feel it.
that would not pass at my shop, just so you know.
thats why I said allmost like a babies butt _Å
what is it for?
heatsink for leds
it will be placed in some wood
cradek: how was hgr?
skunkworks: fun but I didn't find anything I couldn't live without
don't let that stop you ;)
I need to mill a 2nd pice :P
this was fun
been running the latency test all morning on the Zotac ($59 with cpu) motherboard and so far it is 3350 for the Base thread and 5250 for the servo thread :)
that's the same motherboard that I have in the Hardinge
2nd milling done =) im happy!
* fragalot is waiting for a reply wrt warranty on his controller still
btw - whoever made the nifty little tool editor in 2.4.1 - that is pretty neat
Is there a way to script EMC's startup? I'm still working on my ini file and having to restart EMC everytime and home each axis then set zero then call up my program is aggravating.
madsci44: hey, I just found some old books on that regulator I asked you about.. want me to email some pictures of the pages related to it? I'm afraid one of them is in german, and I believe the other one to be gibbrishanistanisch. :/
StonedSlacker: you could try something like this:
/home/dan/emc_trunk122408/scripts/emc /home/dan/emc_trunk122408/configs/sim/axis.ini /home/dan/CAM/drawings/testme2.ngc
it won't home for you but it will get you closer
you need to change the directory syntax of course
just milled something that could pass as panel for something, im getting good at this =)
fragalot: Gppd, now make me a cheeseburger
all you fr's look the same
lots of j's here too
Thanks Daniel, I'll have a look at that
fragalot: how many jy<tabs> do you find?
one too many.
andypugh: Fsck you and the lathe you rode in on!
* fragalot makes Jymmm cake to make up for itt
That's the best offer I am likely to get this weekend, I fear.
about the double-characters i send: My keyboard doesn't seem to be debounced propperrly
* Jymmm buys andypugh a lapdance by placing two one dollar bills in fragalot's g-string!
Don't be daft, fragalot isn't a Lapp.
andypugh: Hey, getting a lap dance from fragalot might be the only action you get all year!
LOL, oh you poor b@stard
damn, the drawers look like a great deal, just a lil too long for my needs.. http://sfbay.craigslist.org/sby/tls/1801580026.html
What are they for? Fishing-rod salesmen?
put in your van or turck bed floor. holds parts like pipe fittings, electrical parts, etc
And they slide out 6'?
well, at least 4'
MattyMatt could convert them into a gantry miller.
attempt numero 2 http://pici.se/p/large/qnrNrlMGz/ http://pici.se/?p_large/VTrZvNTLk
Made on the Mechmate?
Better than my so-called milling machine would do, I suspect
mockup panel just for shows =)
so ppl wanting panels can see I can make them :P
frallzor: don't forget the counter sunk mounting holes
frallzor: and the engraving
im pretty happy, first made 2 simple heatsinks now this =)
Have you considered engraving captions under the holes? Something like "Gigavolts" and "Interociter Phase Angle"
engraving, good idea, would be fun to test that tool
andypugh: I just said that =)
next try will have some text
I was reminding myself of what the Interociter was called.
all out of sheets =)
frallzor: 600 count Egyptian Cotton?
haha Item #3 http://sfbay.craigslist.org/sby/clt/1801624405.html
Hmm, nice size. Or perhaps a little small. http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=300437304572
It's not even assembled. I wonder why he didn't finish doing that.
There are some odd folk about. Perhaps he was given it and never needed it.
[18:12:14] <andypugh> http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=110546343634
is about the same distance from me and probably a better size.
he said he took it apart, but eh
Perhaps he likes cleaning and fettling, but not making stuff?
Or, it works but he had an issue, tried to fix it, and couldn't.
What could not work on it?
andypugh: nice Harrison. ISO30 spindle, too
Yeah, the 30 spindle is what sells that to me. And the swing-table is ideal for gear hobbing in the horizontal mode.
now, how to motorize the head unit to go from vertical to horizontal...
I think horizontal to vertical involves fitting the over-arm, so is probably a job for an industrial robot.
looks like is rotates from one to the other
oh, nevermind... I thought the head had a 45deg swivel in it
andypugh: more on the bldc comp please. is the sinusoidal version aimed at the 7i39? connected to what? 5i20? ? ?
that would have made more sense ;)
It does that too, but that is a removable vertical head. http://www.lathes.co.uk/harrisonmiller/index.html
is this a score or what? http://qurl.org/F11
Roguish: The sinusoidal driver certainly works with a 7i39, but with the right BIT file will work with any power module or even discrete transistors, IGBTs etc.
if you had the space... ;)
slow rapids, though: 720 IPM rapids
I have been using it with a 7i43 but that was a special config, the 5i20 BIT file is probably in the current EMC2 release
I work on machines that rapid at 3500 IPM !
I built a machine with 50mm travel and 264000 IPM (50mph)
It was a compression rig for crash-test simulaiton.
zero to 50mph in 2 inches? wow
No, zero to 50 in 0.4" and then a steady speed the rest of the way ;-)
what did you use to accel it? gunpowder?
No, that is a little too fast (I discussed it with someone who did use explosives to excite his Hopkinson Bars). This was awkwardly between the conventional speed ranges. I used a 100kg flywheel at 3000rpm, a pneumatic clutch and a cam.
did I do the math right? I get 63000G accel
Sounds about right. The first-cut sliding platform left the bearing carriers behind. The carriage was a different shape after the test.
uh, you did the math wrong - 50 mph is 264000 inches/hour, not minute :-)
It was the most frightening mouse-click ever to release that clutch.
andypugh: where would i find your source in gitweb?
I can imagine
corrected calcs still say over 1000G
I think.... (calcs some more)
It was a while back. I built the machine out of the photocopying budget of the project, the _original_ proposal was well inside what the existing servo-hydraulic machines could manage.
Roguish: It isn't there yet. (I don't think) unless someone has committed it.
units 50mph/0.4sec gravity
er, 0.4", forget it
andypugh: thank, but i think i messed up the transfer. i got 0 bits
could you do it again? please.
Roguish: I am not sure it actually works.
(The file transfer that is)
Roguish: The correct sample HAL file is still queued
i'm mechanical. not electrical, and have no way or desire to make a drive.
which one is that?
I renamed it for safe keeping while doing a different driver in the same config
get error in the transfer. seems to be a time limit or somethingg
thanks, there it is.
i think i'll visit mesanet one day next week (1/2 hour from work) and pick up a 7i39 and check if it will work with my available motors
company uses lots of 23 and 34 bldcs, ones on vibbration tested systems get scrapped out. i dumpster dive.
i can get all the details on the motors: halls, etc.
anyone in need of a 'wee' lathe? http://qurl.org/G11
gee and I was looking for a50" x 150" lathe the other day
would be a shoe-in at my house... just move the couch into the washroom, tv into the kitchen and knock out 3 walls :/
one wall being an external wall?
hmm glxgears is not on 10.04 by default
I just need to find a couple of CC-19 by 5/8 or 3/4 holders
heh it says local pickup only for that lathe, they should deliver
jthornton, I found another faster way to get bad latency, open largish file in gedit and scroll with the mouse wheel
was much worse than glxgears on a particular box, didnt investigate why
heh I don't have any files on that computer lol
the latency is just below 6000 which is about what the same mother board did on the lathe
open a log file
[19:29:14] <pcw_home> http://www.govliquidation.com/auction/view?id=3406913&convertTo=USD
probably a little too big for home shop as well
hehe BUYER LOADS with no assistance
ok I give up looking where are the log files?
23000 lbs and no help...
Just hit the Home Depot Employment Office
they stand back and laugh at you
I looked everywhere but var lol
NAS == Naval Air Station, Jackson FL ???
better buy two then =)
archivist, no change in latency while opening and scrolling while surfing while getting a package installed and playing a silly game
Zotac GeForce 6100-Value same one that is in the Hardinge
Built in video?
with an Athlon II X2 245 AM3 2.9G RT
it was on special at newegg for $59 with the processor
I guess the curse is off built in video
sounds as good as the goal3 motherboards.
Is that with a SMP kernel
I'd have to say this built in video has shaken the curse for sure
i need to build emc on my new box, gfx are built in, but it has sideport ram you can use w/o dipping into shared system ram
I'm running 10.04 with mozmck's emc rtai
curious about latency in that case
pcw_home, what is a SMP kernel?
jthornton: Multiprocessor/Multicore enabled
Buzzword for Symmetric multiprocessing
it must be as I see both processors busy and what ever they do in the system monitor
Kernel Linux 2.6.32-22-rtai
jthornton using his packages?
they work fine?
yes and yes
installed emc via them too?
and for some reason I didn't have to select the rtai kernel on boot up either
yes the whole shebang - dev
tried machining with it?
I am pretty sure the package that mozmck is working on is smp for lucid
not with the one with 10.04 but plenty of cuts with the 8.04 install
same packs with hardy?
I don't think so
* jthornton wanders out in the 97F heat looking for a garden hose to finish flushing out the coolant tank :/
in that heat you should be flushing beer bottles into self
did that for breakfast just to prepare for this afternoon
morfic: You there? I tried using an AC motor as a servo.
Starts. stops and reversals are problematic
I think that systems that get good performance from an induction motor as servo have to model/control the rotor flux
Indeed. For indexing for probing, though, I thought it was quite promising.
there seems to be a little oscillation/overshoot at stop
Yes, if you do not expect good dynamic performance it will work
I guess that a vector drive would perform a lot better, but I think even mine would be usable with an inverse deadzone and some more careful tuning.
We a feedforward model if tuned to the motor should be pretty good
I'm pretty dense I just realized why Tesla automobile have that name
(they use induction motors )
Considering that it isn't meant to work at all. I wouldn't want to try it with a system that uses relays for the Fwd-Rev buttons though. (bear in mind that that is how this VFD works, except the buttons terminals are wired through a ULN2003)
Oh I though you were driving it with your IRAM
No, thats sort of the point of the excercise.
I would be disappointed with that performance with something I had built for the job, that is with a 1988 Fenner VFD I got from eBay.
pcw_home: got a question about the PLX9054-based cards (specifically the 5i23)
I think the overshoot is because you lose rotor flux and it takes a goodly amount of reversed drive dir to recover
at least on the 5i23, the firmware doesn't load correctly, and leaves the FPGA (or PLX9054) in a state that requires a power-off to recover from
you mean with a bad bitfile?
looking at the code, I see that the PLX9030 programming routine reverses the bits in the bit file but the PLX9054 routine doesn't
no, with the same bitfile, 2.3 works and 2.4 doesn't
bot should reverse bitfiles (bot not promfiles)
actually it's swp who noticed that there was a reverse missing from 9054..
I'm still half an hour away from finding the problem using 'git bisect' :-/
we could calculate how many more iterations it would take, now that we know which commit is at fault
but this brings up another question: whatever state the non-reversed programming leaves the card in, emc can't program the card again until a hard power cycle
yeah, "reset" doesn't seem to work, and the only way it checks to see if it failed is by checking /DONE (which isn't set in the case where programming fails, of course)
is there supposed to be a 'reset fpga to power-on-equivalent state that always works' method and the driver is buggy and doesn't reset it right?
or is it a limitation of the hardware
or at least a way of detecting that the chip hasn't reset, without the assumption that /DONE will accurately tell us that
That doesn't make sense.
um. I don't think I want to know what you were looking for ...
SWPLinux: I hit the free/barter/garden/tools/materials sections of CL on a regular basis
well OK Then
that was in barter
there's not a joke in the world where the punchline is "for barter, same as in town".
It should always be possible to reset the FPGA, but its possible to craft a bitfile that wil not allow reset or crash the computer
it seems unlikely that a bit reversed firmware would pass the fpga's crc or whatever it uses..
Yes thats strange, I would expect it to just fail to load
Ill try it here and see what state the hardware is in
Without a bit reversal the sync word would be wrong so loading would not even start
pcw_home: I guess it could be down to motor properties, but I think it might also be that the VFD does nothing at all until the PID output = 4 and with a gain of 20 on the PID and a 0-1 angular range that is a heck of a deadzone.
Maybe a little of both, least anti-deadzone woudl be a good thing to try
pcw_home: you can easily see this problem with a 5i23 and v2.4.0 or v2.4.1.
if you have a v2.3 and a v2.4 git checkout you can run the v2.4 first, then run v2.3 and see that it can't load the firmware
I put in anti-deadzone and it helped a fair bit, but then I tended to get oscillations.. Then I got bored.
I can just try it by hacking my loader so it doesnt do the bit-reversal
emc's error is: Jun 20 14:51:50 azazel kernel: [ 112.412154] hm2_5i23.0: Error: Not /DONE; programming not completed.
you get that when you run with the version that has a bad firmware loader; then after that you get it with any version
at least that's the current guess; I have tested that the revision right before reversing was changed is not broke
I'll look forward to hearing the results
If you load bad firmware, all bets are off, but bit reversed firmware should be a no-op. will check Moday Morning
jepler that symptom re mesa card resetting is the same also with the 7i43. first firmware load at power on is perfect, but restart EMC and it fails saying the 7i43 is a 400k gate ver when in fact firmware file specifies 200k gate. Fix is to power cycle the 7i43 and restart EMC
pjm: which firmware file? (Or all of them ...)
ah well i only run it with one, firmware=hm2/7i43/SVST4_6S.BIT num_encoders=3 num_pwmgens=0 num_stepgens=5
ok. which version(s) of emc?
That would be a different issue (Driver fails to reset 7I43 before attempting reload)
SWPadnos its the 2.4.1 from the ubunto update server
pjm: birdie on the desktop?
um i cannot honestly remember, but i think yes
8.04 of ubuntu etc
right , that
ok. do you remember if it happened on any older versions?
it used to not happen, but i cannot recall when that was
perhaps before this last update, i remember tweaking stuff in the ini and restarting the EMC without having to reboot the 7i43
i've put a NC 'reset' button inside my electrical cab now, so i can simply press to reset the card
that is pretty wicked! one of your conversions?
The driver is supposed to EPP write 0x5A to 0x7F7F to reset the 7I43 before re-programming maybe that gotten busted
its not really too much of a problem currently since my machine is running nicely, so I dont have to do multiple restarts of EMC, but the other week I was fiddling with Z axis speed/acceleration etc
to get a decant speed peck on drilling ops
pcw_home: the 7i43 driver does do that write
in the reset function
I wonder what changed?
it actually does that write first, then tries something else to reset in case the CPLD is in charge of the port
pjm: do you have a dmesg capture from a failed 7i43 load?
i will do one
Maybe when it restarts something is out of sync with the EPP interface
skunkworks: I have a feeling that every link in that toolchain cost more than my entire machine.
so top of file is normal start
bottom of file is EMC restart
ok i have to sleep, need to be out of the house at 5am ;-( I'll check in here tommorow
we have a 7i43 here, so we'll do some more testing
except that we actually have a 400k-gate one, so we probably won't see the problen
andypugh: we bought the machine pretty much for the price of scrap metal
Yeah, that would be 10x the value of my machine :-)
hmm - wonder if it is as simple as noise getting into the keyboard.
pcw_home: should the /DONE line read 0 after resetting the FPGA?
(on the 7i43)
oh, and do you recall if there needs to be any pause after writing 0x5A to 0x7F7F?
7i43-400 is working as expected
\done should be high after reset (I wonder if the 400K version really reset)
well, the interesting thing is that by default (or at least here), when you read from address 0 of the FPGA, you get FF
jepler: you are probably right about the put function - I am way behind on my kernel and catching up to the "new" stuff, do you need put when its called from a while tho?
which has the bit set that tells us it's a 400k gate chip
so if the FPGA doesn't really reset, it looks like a 400k regardless of what it is
madsci44: I think it's reference counting, so the number of gets has to equal the number of puts
yes it is
SWPLinux: but the FPGA is really resetting
I put a call to reset in cleanup, and it goes to "DONE" lit on exit
oh, right. the CPLD just doesn't seem to be getting control back then (???)
but i read its supposed to automatically dereference in the case that its called from a while loop? so maybe check counts - in hal_vti and opto_ac5 there is no while so im certain its required there
SWPLinux: Hmm, the FPGA registers pretty much always return FF when read.
Well you should get a change in \DONE so should always fail if you dont reset the FPGA
(not the readable ones)
Only explicitly readable registers are readable. I guess that is none of them without a valid program installed?
ok - jepler found that it is resetting
though we have a 400kgate chip, so it's hard to tell what may be wrong with the 200k
Without a program installed you are talking to the CPLD
maybe seb will bring a 200k to the workshop
Endeavour_: Did you blow your PC up by disabling SMI yet?
madsci44: oh, are you saying that in the construct dev = pci_find_device(..., dev, ...) that the count on dev is decremented in pci_find_device for non-null dev?
andypugh: I haven't figured out the problem. Haven't gotten too many suggestions on it.
I bothered to read some docs this time and it sounds like you're right
Tried disabling a few NVidia drivers from the Synaptic manager, no changes whatsoever.
jepler: yeah that :) except pci_get_device()
Thinking about removing the card and just hooking up to the motherboard on-board video and seeing what happens.
i really have to study up on this before I mess around any further
Endeavour_: if I remember correctly, you had delays in the 500k or more range
though that could be video related, it probably isn't IMO
Endeavour_: It might make sense to keep your nice computer for non-EMC use, and stick an Atom board in the CNC Machine. It might save a fair bit of messing about and people report good results with the Atom 330s. That does sound like giving up though.
Endeavour_ is now known as Endeavour
but only apparantly? if called from within a while loop
it seems put is automatic in that case
There may be some more robust way to do the 7I43 reset
(epp data read first to reset address hilo toggle, epp address write 7f epp address write 7f epp data write 5A)
andypugh: I'm probably going to do that in the end. I never really planned to use it for EMC to start with, I just didn't have another machine handy and wanted to get my machine running.
epp reset in case CPLD-FPGA handoff mungled something (shouldn't but paranoia)
I was probably going to get an Atom 510, andypugh.
The 330 is out of production and ridiculously expensive for old tech.
Have you seen latency numbers for the 510?
jeff added a hm2_7i43_reset call to the end of the module cleanup, and the DONE light now comes on when the module is unloaded
so the FPGA seems to be resetting correctly
we don't have the problem pjm has, since we have a 400k board here
so even if there is some error, this FPGA can be re-programmed
* frallzor has an atom 330 and it rocks
(I don't think we're seeing the problem, so it's hard to debug)
Endeavour: You have seen the latency test table at http://wiki.linuxcnc.org/emcinfo.pl?Latency-Test
andypugh: Yeah, don't think the 510 is on there.
So you might end up with something else that doesn't work.
Though I guess you could probably sell it back on on eBay
andypugh: I very well might. Only one way to find out. I'd rather use something that's easily accessible though,
I can always return the motherboard if it doesn't work right though.
I am not sure how many vendors would accept "poor RTAI" performance as a reason for return.
Ive seen some d510 latency figures and they looked similar to the 330 based MBs
Ah, well, in that case, go for it.
actually it probably is necessary
just because something is out of production doesnt mean its crap :P
its just more fun to have 10 good things than make 10000 different ones =)
frallzor: Of course not. It just means it's more difficult and more expensive to get.
cant see why, was very easy for me to get mine =)
and I got it like a week ago
same price as the rest too, slightly cheaper
How much did you pay?
$150 vs. $80; I prefer $80.
around 800 sek
Well, the only US source I found is $150.
ok its got nothing to do so much with the while loop - its where its called until it returns null that it drops reference
ok i think i understand it now - if its called from within a loop and it returns a device and the loop is exited, then you need to do the put, but if you keep calling within the loop until it returns NULL then the put isnt needed
so I got my PSU today. must weigh 50 lbs
what does it output?
1500Watts at 80V
also has a +5v and a +12v for logic
madsci44, probably a little overkill, but I figure if I ever want to do a 4th axis I don't want to have to buy another PSU
well its worth having headroom
only cost $160 too
hilarious, it came in a usps flat rate box
must weigh 45-50lbs
must not be a switch mode?
does anyone have josh lubell's current whereabouts?
the transformer is 9" in diameter, 3.5" high
madsci44, nope, all I have is a coil, a couple caps and some bridge rectifiers
its definitely good to have the reserve then - they can droop sometimes at heavy load
yeah, each of the motors is rated at 666W
The Servo of the Beast?
so in theory, I would not have enough if I were running all three motors at full load
but that seems incredibly unlikely
is that peak watts or continuous?
those sound hefty
You know, I am not entirely sure
they are.. 7.6A @80V
Mine are 7.5A @ 365V and are rated 500W...
yeah, it says continuous on them
so I am assuming peak is higher
should be tons enough for the mill anyway
madsci44, yeah, and it was only like $20 more then the 1000W one
where did you pick them up from?
[23:38:26] <alex|lappy> http://www.antekinc.com/
no, it was a lot more
[23:39:39] <alex|lappy> http://www.antekinc.com/details.php?p=339
I think this is the exact one I got
i should start selling my supplies
madsci44, are you making a lot of power supplies?
im doing way too many things at the same time
yeah I have been designing them for years for my own stuff
mine are all switchmode, very rugged, overload protected, small, lightweight, etc
madsci44, did you see the estop switch i was planning?
i make my own transformers, vacuum potting and everything
i think i saw an early version of it
I need to draw up a revised diagram, becuase I am going to be making a couple changes to it
madsci44, did you see the problem that was pointed out with the NC momentary switch?
i saw some typing about that
something regarding both switches closed create a short or something? I'd have to see the diagram again
yeah, if the NO switch got stuck in, it would put the load across the relay coil
or at least, enough of it to cause a fire
what do you think about just throwing a fuse in front of the switch, so that if it did get stuck ,the fuse would blow and I would have to fix it before i started the machine again
What voltage is the relay coil
well isnt that what the fuse is in that diagram?
If it is a 240V coli then no matter what load you put on the far side of it, it won't blow the fuse.
i never run my switches off the line - i always run 24 volt contactor coils with a circuit supplied from dedicated 24 transformer for the loop
So, you can run the machine through the coil, but the machine will only get as much current as the coil passes steady-state. It probably isn't an issue.
Well, I haven't seen the wiring diagram.
(lately i seem to be that alot)
whats the problem with the estop again?
madsci44, well, what happens if the NO momentary switch that passes an initial current across the relay gets stuck close
There is no auxiliary contact then?
well yeah then you are stuck on
but usually the estop switch is a latch isnt it?
you press the big red button and it mechanically holds open until you reset it
madsci44, I think the issue is that if the motors are under load while that is happening, that is a lot of current going through that little momentary switch
That little mometary fuse, you mean?
its going through the relay contacts - i guess it coudl go through the switch if the relay contacts dont close properly