good night all
hi guys, everything going pretty good here, i cut a few pieces today, and everything good except the circles have flat spots parallel to the x and y axis's, is this cause by my backlash, i think it was between .03 and .04. Also has anyone used emc to drive laser scanner servos? Just wondering how hard it would be to rig up a laser above the table and have it project an image from the g-code onto the table?
you could probably set up some non-trivial kinematics module to do the scan, but it would only happen once
emc isn't meant for replaying G-code files rapidly, like you'd need for a laser light show ;)
yeah, everything i've search on the net has taken me to laser light shows, and that stuff is expensive. since all the other tools these days have a laser, i thought it would be a nice addition to the cnc router. I may put some mirrors on some servo motors and play with it one day when i get back around to playing with servos again.
you could use "O- while" to loop in the g-code. but a bigger issue might be that each segment of a laser show takes a really short time---hardly long enough for EMC to do closed loop control
if you want to get what looks like a line instead of a moving dot, you need to retrace the pattern many times per second
thats hundreds to thousands of times faster than normal machining
that was my point - you need rapid repeat of the pattern, ideally with some laser poer control since the speed at which the dot moves will likely be incinsistent
if onl;y I could type better
anybody ever dealt with http://www.kelinginc.net/SMotorstock.html
or have any other recommendations for buying strong NEMA 34 steppers?
steves_logging is now known as steve_stallings
got those two links from your pages ;-)
I have dealt with both Kelinging and Homeshopcnc to get samples. Good service from both. Big NEMA-34 have strong detent torque. Keling seems to not fully understand driver electronics.
drivers will be gecko
neither one shows speed torque curves
Then jumper Gecko as if NEMA-42 if motor is 1200 oz-in NEMA-34 in order to get anti-resonance to work correctly.
didn't know that - good info
Torque curves are always for a specific driver and supply voltage. These guys, and the Chinese vendors that they buy from, aren't prepared to furnish that data.
I'm planning on running a shoptask, so I probably don't need outrageous torque
but I want good high speed torque
Expect MUCH faster torque drop-off with NEMA-34 than with NEMA-23. I would guess ratio tracks rotor inertia.
nema 23 is too small though - the machine has mounts for 34
I think shoptask sells 300 and 500 oz-in systems, and the concensus is the 300 is lame
Homeshopcnc sent me a 400 oz-inch motor in a NEMA-23 oversize that seemed good, though I would have recommended a higher current, lower inductance winding.
I was looking at the 1160/1200 oz-in motors
6A, ~3V, ~6mH
I have a 1200 oz-in that you are welcome to play with, cannot remember current/voltage/inductance but it is probably in that range. It came from Keling if I remember.
I appreciate the offer, but I kinda have to make the decision soon
I'm selling my south bend 13" lathe in a complicated three way "barter" deal - one of the parties will be paying me about half the total price in cash - the other can't pay cash for complex reasons, but is willing to order me some motors, or other stuff totalling around $500
Then I would recommend going for the torque you know you need, not overkill. That extra rotor mass is going to resonate and reduce your top speed.
measuring how much I need is the trick
I can measure zero speed breakaway torque at the handle
and I can even assume that high speed torque requirement will be similar
For this kind of situation, I think cut and try is the only method that really works.
but how do I know what high speed torque I will get?
looks like the choices are:
640 oz-in, 2.3V 5.5A 3.5mH 1400 g-cm^2 inertia
Geckos are a good choice, and plan on about 20X supply voltage to motor voltage. Motors will run warm due to circulating current at rest, but so what.
990 oz-in 2.1V 6.1A 3.3mH 1400 g-cm^2
1200 oz-in 2.7V 6A 5.1mH 2700 gm-cm^2
Ouch, inertia on 1200 oz-in hurts!!!
looks like the 990 might be the winner - same inertia as the 640, lowest voltage and inductance of all three
Agree, assuming rating data is real.
they're that bad when it comes to meeting specs?
[02:45:57] <jmkasunich> http://www.kelinginc.net/KL34H295-43-8B.pdf
is the datasheet
My guess is that original manufacturer does a good job, but data gets jumbled or misquoted along the way. In my limited exchanges with msmotor of China the engineers seemed quite knowledgable.
I take it that non-china motors are much more expensive?
.... and are actually made in China but branded and sold as USA product.
Truth hurts sometimes...
I'm glad I talked to you
I was actually seriously considering the 1800 oz-in ones, thinking "overkill can't hurt"
USA companies do the research, quality control, and good documentation, but the actual hardware is often the same as cheap Chinese.
I wonder if kelinginc.net accepts returns if the motor doesn't match the specs?
I can certainly measure DC resistance and inductance upon delivery and send them back if they don't match the datasheet
(it will be some time before I'm ready to actually test them under power)
I would guess so. I bought a stepper driver that was supposed microstepping and it turned out not to be. When I mentioned it he offered to take it back, but I never got around to sending it.
must not have been very expensive...
(I'm looking at $350 worth of motors - I want to get what I ordered)
you said "he" - is this a one man or very small company?
Driver was less than $100. Company seems to be 2 or 3 people doing importing. This is the guy that was at the CNC Workshop in the back room of the school building.
are those the motors you had in a box when we were packing up?
If you are remembering a NEMA-43 about 8 inches long, then yes.
you had several nema 34s, different sizes, and you mentioned an importer
but at the time, I read what you said as "I got these from an importer and brought them here with me", not "I got these here and am taking them home"
My biggest complaint is having 4 loose wires instead of a jacketed cable, no sealing of opening.
yeah, that isn't nice
but its also pretty much the norm
I did get the from Keling before the show. I was partly responsible for him coming to the show.
shame I didn't meet him
here's a table of specs for some NEMA 34 units: http://www.servosystems.com/amp_step34ht.htm
I take it he seems like a decent guy (or you wouldn't be dealing with him)
could use as a sanity check
HT34-476 seems similar to the 990 oz-in
Yes he was decent and tries to work with hobbyists. The guy from HomeShopCNC, name slips me at the moment, is also plesant work with and actually seems to know quite a bit about motors. Seems to have a mechanical engineering background.
oops, I take back what I said about the HT34-476 - it has much higher inertia
I'm liking the 990 - I'll have to give him a call and ask questions
Have a go at it, but my imperssion is that he is a businessman, not an engineer.
well, the two questions that come to mind are:
1) the drawing seems to show a round cable, not loose wires - am I that lucky, or was the drafter just lazy
2) if it doesn't match the drawing electrical specs can I return it
I think I already know the answer to #1
the drawing says UL1430 leads - that is plain old wire
I would expect it to match electrial specs. Can you test rotor inertia easily?
can't test inertia
but if the inductance and voltage are as spec'ed, I'll be reasonably convinced
Yea, if figured. Might be impossible in assembled motor due to detent torque.
the physical size of the motor limits the inertia - it is significantly smaller than the 1200 oz-in ones that have 2700 gm-cm^2
on the hobbycnc site, length and inertia pair up like this:
the 990 is 96mm
seems logical enough
it claims 1400gm-cm, which might be optimistic, but I doubt its 2700
logical comment refered to first 3 pairs, didn't see 990 before sending
the 990 does kinda seem too good to be true
its weight and inertia numbers match an 80mm long 640 oz-in motor
even if the inertia number is wrong, the 990 still seems a good compromise in your case
as long as the inertia isn't 2700 - if its that high, I might as well get the 1200 and have more torque
cut and try, cut and try......
oops, scratch taht last comment, the 1200 has 5mH instead of 3.3
that gets expensive!
resell on eBay to moderate cost
(of course, I could ebay the 990's if they don't work out)
the other thing I've wondered about is direct drive vs. belted
the machine is set up for 2:1 reduction
but I _could_ direct drive
dunno whether torque increase from the pulleys is more or less than the the high speed torque falloff at doubled speed
I like belts. Choice of ratios, no critical alignment, elastic damping in belts.....
ok, let me rephrase that
2:1 belts or 1:1 belts ;-)
the current leadscrews are 10 tpi acme, may someday replace with ballscrews, probably 5tpi
with big motors you usually see low ratios to avoid needing high RPMs
2:1 gives me 40000 steps/inch with the acme, 20000 steps/inch with ballscrews
I'm gonna have hardware step generation, so that isn't a factor
referring to torque fall off vs. speed not pulse generation
thats why I'm bummed about the lack of speed torque curves
20000 steps at 10 step microstepping, right?
2000 full steps/inch
I'll be running either 48 or 60ish volts
try having a look at curves from an American vendor like Pacific Scientific, they should be similar as that is the kind of thing the Chinese are trying to copy
48 if I use a 48V 10A supply that I already have around, or 55-60ish if I get a 40V toroid transformer
not a lot of difference between 48 and 55 volts....
the biggest improvement in speed will come from ballscrews, 80% efficient vs. 15-20% for Acme
on the Chinese mills I have played with, we typically almost doubled the rapid speed when changed over to ballscrews
these are mill-drills, or knee mills?
small stuff like Jon Elson's demo minimill
some day I will try putting steppers on the Bridgeport that I brought home from the CNC Workshop, I got it primarily as a test bed
PMDX needs to address the larger end of the driver range
hard to compete with gecko
no kidding, I would never go for a direct competitor to the G200 series
there seems to be a real need in the 4 to 5 ampere range
tough spot to work, to much current for monolithic chips and discrete works out too expensive
IOW, discrete ain't gonna be much cheaper than the 7A gecko
so why bother?
as soon as you go discrete, I _think_ the cost of 5A vs 7A becomes almost indistinguishable
interesting - according to PacSci, one stack motors have much less torque falloff at high speed then longer ones
there is some room to work if you do creative packaging and also target higher function units with pulse multiplication and overcurrent protection
lets see, motor behaves like spring-mass in resonance, now does that have a square term related to mass?
here is a feature you can offer: quadrature input
same two wires/optos that you use for step/dir, but more noise immune, and you can double the step rate for any given CPU interrupt rate
I like quadrature technically, but the market is going to ignore it.
(jumper selectable of course, for folks who use the old approach)
if you go FPGA for the logic, quadrature conversion is pretty cheap
input commands would typically go directly into a micro, we do pulse mult in software in our units
uP instead of FPGA?
come to think of it, we use a D-flop to capture direction signal in real time, that will be more difficult with quadrature
do you poll the inputs (after latching direction)? or do they go to a hardware counter in the uP?
yes uP lets us add lots of features like self test, demo modes, flash codes for errors, reduce current, lots of stuff that Gecko typically avoids
step causes interrupt, dir get read from D-flop
the newest gecko is using FPGA - I don't think they have quite as many things as you just mentioned, but they have definitely advanced from the 4000 series days
the uP has hardware PWM generation I assume?
Gecko added overcurrent protect, but that has nothing to do with FPGA
actually uP has on board DACs
so analog PWM generation?
analog switching current regulation
what do you use for current feedback? sense resistors in the fet source?
hesitate to call it PWM because people tend to think pre-computed constants
I guess that depends on the people
standard current sense resistors, sensing FETS are too expensive
triangle wave, comparator, analog reference = PWM
sort of, but stepper drives use dead times, multiple recirculate modes, etc.
3lbs2oz of water in a milk carton hanging from a crank handle with a radius of 1.6" is enough to move my axis when the gib is tightened moderately tight
that explains the long silence 8-)
only 6 mins
you are hired....
so, 80 oz-in is what it takes to break friction and start motion at speed = zero
with 50-60V, PacSci's midrange NEMA34 one stack motors retain ~20% of stall torque at 1200 RPM, and about 7% at 2400 rpm
now guess at friction while at speed, figure acceleration of mass, get headach, then cut and try.... 8-)
friction at speed is lower
for the slide, yes, but what about the Acme screw?
accel of table is probably small (if not negligable) compared to rotary inertia
(rotary inertia of the motor that is)
especially with 2:1 ratio - table and screw inertia are reduced 4:1 when reflected back to the motor
rotary inertial of screw might contribute, though not as much as rotor of motor
with 2:1, 1200 RPM = 60 ipm
and the 990 oz-in motor should still have close to 200 oz-in at that speed - twice the breakaway torque
Bridgeport Boss machines were typically around 100 ipm from factory. I should think 60 should be acceptable on a Shoptask.
anything above there is gravy
(I will certainly test and see what I can get)
cut and try is of course what I'll do when figuring out max vel and max accel - I just don't want to be swapping motors
if I only get 45 ipm, so be it
truth is that what is needed is reliability, low speeds are far less of a penalty than lost steps
I'll test to see what causes lost steps, then reduce by a significant margin
I can get my little demo machine to do 50 ipm rapids if well adjusted and lubed. Dependable speed is about half that.
what motors are you using on that?
Surplus 200 oz-inch NEMA-23 driving 20 tpi Acme screws with 2:1 reduction.
Remember, really small machine. Seig X1.
have you been following Mariss's "unstallable stepper" project?
I think of it as the "50 pole AC vector drive" project
if you wanna get radical with your drive design, include encoder inputs on it
two phase motor with lots of poles, but otherwise similar to "brushless" DC
stepper drive: constant current, varying angle between rotor and stator flux (varies with torque load)
detent torque will still sap power and limit speed
vector drive: variable current, constant angle between rotor and stator
the only thing I see that is good about idea is the fact that stepper motors are commonly available, otherwise quit playing around and use a servo
maybe you should build an AC servo drive instead of a stepper drive?
I have considered a "brushless" DC servo, but fear the support costs. People will try to drive almost anything they can find, and expect you to solve their problems.
I wonder if there are servo equivalents of the chinese steppers?
import them, and sell matched sets only
Selling matched pair of motor and drive is the way to go, but I really don't want to sell motors. Inventory and shipping costs are too great.
Keling sells brushed DC motors and claims to be able to source brushless.
cooperative effort? he sells motors, you sell drives, instructions tell the user what motor to buy, no support for non-standard motors?
even better, I sell OEM drives to him like Gecko does
either way, the margin has to be better than competing with gecko
a guy with a 5A stepper can choose between you and gecko - you gotta be cheaper than gecko, cause most (IMO) won't pay much extra for the features you described
currently the other hole in the market (besides 4-5 ampere stepper drives) appeares to be brush type step/dir input servo drives big enough to handle old American iron
you mean like the mazak?
though that is the upper extereme
I wish I could remember who's servo drives we wound up using
probably Servo Dynamics
that rings a bell
they aren't big
but they are analog input
I dunno what they cost - more than a gecko, but I think well under $500
right, forgot about that
currently people are trying to use the Rutex drives rated for 40 amperes and 200 volts, but they do not survive in the real world
highly optomistic ratings
they using rectified line? or still have transformer based power supplies?
still needs transformer, Rutex step/dir inputs share power ground, not even opto-isolated!!!
definitely not real world
setup uses SPI interface and most users report such severe noise problems that the communications are not reliable enough to tune the drives
they could either isolate the control from the power, or isolate the step/dir/comms from the control
have you seen the new analog devices isolators?
the more I think about it the more I like your high power drives approach of isolating gate drives, shame it is expensive on smaller drives
TLP250, $1.88 each from digikey
common power supply for the low side, bootstrap the high side
we use the AD isolators in our PMDX-103, great device, really like the fact that they do not need an initial transition to sync
I just used one on an analog isolator circuit - very happy with it
TLO250 is fine, isolated power source is the cost item
sub-1% accuracy and sub-millisecond response time on the analog output
you only need one supply, referenced to the negative rail
???, perhaps I better look again
you know the bootstrapping approach used by the various internation rectifier driver chips?
same thing can be used with TLP-250's
discrete diode from +12V to phase A floating +12, electrolytic and ceramic caps from floating +12 to phase A FET source
IIRC (its been a while) the TLP has undervoltage lockout
all three low side FETs and TLPs can use the same +12
another option might be the international rectifier drivers (they have 6 channel drivers in one package), with analog devices isolators on the logic level signals
yes, I understand bootstrapping, just didn't realize it would work with TLP250
can't say I've personally done it (yet), but I see no reason it wouldn't work
aarrrrggghhhhhh!!! Acrobat is hosed again, most unreliable software of all I work with!!!
as long as the TLP has undervoltage lockout, the caps and diodes are sized to deliver the TLPs quiescent current as well as the gate charge, and the frequency/duty cycle isn't too weird
the real problem isn't gate drive, its current feedback
especially if you are talking three wire/three phase motors, sense resistors in the source lead really don't cut it
(plus, even those need isolation)
oh yea, I forgot about isolating the feedback 8-)
we use hall effect sensors
you need 2 (assuming no grounds in the motor, you can calculate the 3rd phase)
the cheap halls are not fast enough for chopping at 20KHz
[04:31:23] <jmkasunich> http://www.allegromicro.com/sf/0750/
those cheap halls?
$6.20 in small qty
need response time of around 1 uS to prevent overcurrent
thats for short circuit protection only? or for the current loop?
short circuit mostly, though you might need 10 uS to accurately microstep
a single resistor in the negative bus could be used for fault sensing
we're talking about AC servo (brushless DC) now, not steppers
most deadly fault is phase to ground
I am still taking steppers in this case
few steppers want 180V dc bus ;-)
to me, the gain from isolating the power circuit is getting rid of expensive and heavy transformers
and 180VDC with 20KHz chop would make a dandy plasma cutter
we have made drives where the uP and other control circuits are referenced to the negative bus
user inputs and control signals are isolated
I don't know if I am brave enough to sell a driver that is not isolated from the line.
if you interface is step/dir (or quadrature A/B) and a couple comm lines, thats one 4 channal ADUM isolator
it would need to be enclosed
but I understand
also a pain in the rear when doing development
when running rectified line, neither positive nor negative bus is grounded, so even a sensor in the plus bus doesn't protect you against ground faults
we've used small cheap CTs with both bus conductors (or all three output conductors) passing thru them for ground fault detection
multiphase GFI 8-)
now if you could only use the same CT for both ground fault and IOC....
instantaneous over current
IOC and ground fault are the primary overcurrent protection faults
then there are various slower faults for sustained overloads and such - most of those can be done in software
software is wonderful, as long as you know when to stop adding features... 8-(
fault protection isn't really a feature
tell that to Rutex
what I mean is that its an expectation
probably the only reason they can get away without it is lack of competition
well thats annoying
and a relatively naive customer base
found a cheap fast 150:1 CT, but they don't specify the max current
probalby sub 10A
finally got a look at TLP250 datasheet in Acrobat, looks like no undervoltage lockout...
excellent isolation properties
thats not excellent
it is when you are used to 80 volt bus
I wonder what HCPL-3150 or HCPL-3120 costs?
(thats what we've moved to from the TLPs)
I _know_ they have UV lockout
$1.62 in hundreds
will they sell to normal people?
(I usually use digikey and the like as an indicator of "normal people" prices, didn't find the 3150/20 there)
make sure you include the hyphen when searching
Digikey's search engine is fast and excellent, but it considers a hypen to be a significant character unlike most other search engines
3150/20 has UVLO, 5mA maximum operating current (not counting gate current pulses) so the bootstrap cap doesn't need to be huge, 15KV/us dv/dt immunity
the 315J is interesting - two drivers in one pacakge
$3.30, so cost competitive and less space
another Digikey search hint, if you get lots of hits on your full part number as if it were a sub-string, just add -ND to the end
holdover from the days when Digikey had discounts for total order value and some parts were excluded with a -ND for non-discountable
now they pretty much all have the same thing?
anyway... back to our hypothetical 120V offline powered AC servodrive ;-)
weird package on the 310J, but nice part
HCPL-315J = high and low side drive for one phase for $3.30
add another $2 (or less) for bootstrap parts
I meant 315J, oops
and $5 for a 15V supply referenced to the negative rail
total cost for three phase isolated bridge drive: $20
current sensors for two phases $13 total
looks like the 2 channel device (315J) is kind of wimpy on drive, 500mA peak
Its been so long since I worked with small devices, I don't know what you'd even need
for say 20A FETs
as long as the gate resistor is 30 ohms or more the driver is fine
Well I was thinking more like 50A FETS for a 20A drive
I still don't know what the drive requirement is ;-)
and I don't know if a 20KHz chop is needed in such a drive
whats the cheapest small transistors you get (sot-23 or such)?
I guess that doesn't matter
2n2222 equivalent, but Vce is only 50 volts
two of the 2A drivers is cost competitive with the 2 channel one
I was thinking of boosters for the gate drive
npn + pnp complementary emitter follower
we do that on a larger scale for peak gate currents in the 30 amp range
but I really think the 2A driver is a better and simpler choice
so do I
so, $20 or so for gate drive
$13 for the current loop feedback, but that doesn't provide IOC or gnd fault protection
actually 500mA may be enough with modern FETS
thats a detail
how to do IOC and ground fault on the cheap......
* jmkasunich ponders
yeah, most likely
mosfets have some amazing short term current withstand compared to IGBTs tho
I tried a 2 amp fast blow fuse on a 10 amp FET and it did not protect against shorts
IGBT SOA is about 2x rated current, or transconductance limited (comes out of saturation) at maybe 6x for 10uS
experimental results are a good thing ;-)
fets can often do 5-10x, until they fail thermally
of course a 10 Amp FET is still a rather small die
yeah, but if you use a 50A fet for a 20A drive, the fuse needs to be 20A
problem is short circuit may have produced 100 amp spike
I want a cheap ferrite toroid about 1/2" OD, with 500 turns of wire and an ID that will pass two strands of 10AWG
run the DC bus thru there - the net current is zero unless you have a ground fault, so it won't saturate
go to Home Depot and buy a GFI outlet
not cheap enough for production
especially if you count the labor to tear it apart and get the transformer out ;-)
so try to figure out where they buy them...... 8-)
china, by the hundred thousand
how did it get to be after midnight?
and can you imagine the pleasure cradek, jelper, and SWP are going to have wading through all this.... 8-)
well, we didn't do it in the -devel channel
there have been many discussions that were farther off topic here
well, I did accomplish something tonight - measured the friction torque on my machine, and pretty well narrowed down the motor choice
and gave me plenty of food for thought on drivers
hope it proves usefull
for a stepper drive, probably not - if you need a 50-80V bus, you're gonna be isolated from the line anyway
ground the negative bus
maybe, if I ever have enought time to try any of it
isolate the control and comm signals with an ADUM1403 or something like that, and call it done
actually it might be useful for a high power brushed driver even if not direct line powered
time for bed I think, thanks for the ideas
steve_stallings is now known as steves_logging
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2006-12-02.txt
ack, not a good sign
anonimasu: you there?
any idea of where to get semi-strong steppers
[11:30:21] <anonimasu> http://www.farnell.com
something like 240oz-in
it's your best bet
aratron has, but for larger ones they end up expensive
I just woke up
im at dreamhack
how large a stepper would you estimate one needs for 1:1 drive of a mill this size? http://lerneaenhydra.shacknet.nu/images/images_other/random_other/img_6280.jpg
2.5mm/rev, acme screws
11 000 sek new
that's more then ballscrews off skf costs
oh, 11 000 for the entire mill
it has acme screws
ok price I guess
how much torque would you estimate is needed to drive it 1:1?
I drive 1:1 with 2mm screws
and pretty heavy table..
how much do you have?
torque that is
just a sec..
im going to gear a bit, and look for bigger servos on ebay
oh, that's not much
no, but it's plenty
[11:42:36] <Lerneaen_Hydra> http://se.farnell.com/jsp/endecaSearch/partDetail.jsp?sku=8425876
9nm holding torque
though im going to gear a bit more..
my machine's too fast..
I ened 20k/rpm to machine ta the speed I can feed :)
I can push it to 12m/min..
but there really is no point
im going to set my rapids at 6m/min
so for a 1:1 something like 3nm or greater should be enough?
it makes a big difference what kind of pitch you have on your screw
but keep in mind that ballscrews are 90% efficient..
and acme are much worse..
im in jönköping today
acme is 70%-ish?
I was there once too, the subcontractor
where did you get your steppers?
it seems that farnel misreported ncm as nm :/
the ones I have are bought from aratron
the other motors are servos..
the ones on the mill
detent torque, is that when powered off?
why are all their steppers so weak? they're all <0.8nm or so
their strongest one seemed to be 1.6nm
peak torque is the max continous for steppers
so something around 0.3 like you had would probably work then
hm, larger is always better :/
keep in mind that it depends on what speed you want :)
imo 4000mm/min would be enough
maybe even 2000
Lerneaen_Hydra: depends for what ;)
* alex_joni just ordered an external axis for a robot
for the mini-mill I'm looking into
1m/sec speed :)
something like that
[12:03:59] <Lerneaen_Hydra> http://lerneaenhydra.shacknet.nu/images/images_other/random_other/img_6280.jpg
Lerneaen_Hydra: the one I'm getting is a "bit" bigger
1,43kW drive power
that's a bit more
20kN = 2000kg max payload
that mill's main spindle is 850W
oh, seen a nice turn/tilt table this week
that was a big one, even by my standards :D
100.000N payload (aka 10tons parts)
and that was an articulated robot?
just a turn/tilt table
so 2 degrees of freedom
scary fast or not so scary?
ie faster or slower than 1 degree/sec?
anonimasu: do you think it's easier to have gearing instead of 1:1?
its really a speed/torque issue :)
yeah, that's what I was thinking
which is cheaper?
Lerneaen_Hydra: 8-9 degs/sec
it's more a issue of making parts..
the smaller ones go up to 140degs/second ;)
alex_joni: ok, that's almost good :p
and of course the robot axes go up to 400 degs/sec
anonimasu: making parts?
imo belt drives and gears arn't made easily
yeah, but bearing mounts and stuff..
belts/gears are easily bought
but making mounts can be a pain
you also get more backlash...
that's negliable almost..
there are backslash free belt-gears also
but I dont know prices
normal ones do fine for 99% of all stuff :)
better accuracy would be nice...
so total cost is similar?
Lerneaen_Hydra: well, I think it's stupid..
Lerneaen_Hydra: why would you put very expensive gears on a machine with normal screws where you do have backslash..
oh, I was planning to preload the nuts, and get a low amount of backlash. the only thing is that with the belts I've used before there's quite a lot of backlash
0.05mm or so
Lerneaen_Hydra: that's lots..
but it's very old and worn
probably from the mid 1980s
and I'm too cheap to buy a new one
well, dont start huting for 0.01mm's then :D
call aratron and ask them at monday
but that's backlash, not "glapp"
they are really helpful..
what's glapp called in english?
I can't recall
wouldnt that be the same thing as backslash?
backslash free (glappfri)
hmm, afaik backlash is "stiff", but glapp is "loose"
backslash is glapp..
so even if you have backlash the table doesn't need to be movable
if your screws have backslash they have "glapp"
say if the slides are too tight
how do you differentiate between the two?
stiff and loose backlash
what you are talking about are different things..
well, I'd try to bend the table..
oh, what are they called?
it's 2 different things..
I'd push my table and measure it..
hmm, I'm not sure I'm explaing well enough... one form would be if you can push the table and move it with some force, the other would be how much the axle needs to be rotated before the table starts moving
I'd put a lever and push my table..
if there was no friction then there wouldn't be any difference
there always is..
yes, but hypothetically
if there's no mass..
isnt that right?
Lerneaen_Hydra: that's why im going to gear my machine a bit..
Lerneaen_Hydra: so I can preload my table a bit more
Lerneaen_Hydra: bend on the table and see if it moves.. and rotate the screw and see how much you have to move..
before the machines start to move..
the second is much more benign
nice/less bad/easier to deal with
if you have decent motors you can preload the table enough :)
but, that adds friction also .)
jepler: I got it. Yesterday. Did you get the links for the software in a email or did they come in the box? All I got was the fpga, ribbon cable and some sort of male female plug for the ribbon cable.
Cool little board.
skunkworks: the URL for the "startupkit" came in one of the e-mails from them
skunkworks: you have to go to altera.com to download the development environment
* alex_joni just got ~8 of his favorite magazines
alex_joni: what magazine is that?
it's a german IT magazine
I think I saw it once
german magazines aren't much use to me in general
I can figure
I'm a bit surprised there's any computer magazine worth reading anymore, compared to what you can read on the internet
these are really worth reading
usually 3EUR / piece, no CD attached
so you don't get the usual crap
what kinds of articles does it have?
all kinds: HW, SW, processor news, linux news, mac news, etc
they have pretty good hardware compares (kinda like tomshardware, but with stuff available in the eu)
some articles about telephony and voip, and software patents stuff.. all kinds actually
do you remember Imperator? a guy from germany
* robin_sz does
we talked and submited an article about emc to this magazine ;)
alex_joni just got ~8 of his favorite magazines
unfortunately they said it's not general enough
[14:26:02] <alex_joni> http://www.heise.de/ct/
not that close ...
do you think ... if I took:
1) thyristor controlled 500A MIG power source
2) sqaure wave generator
I could make a pulse-mig PSU?
pulsing might involve some more things like filtering
I suyspect just pulsing the normal control line would not work
as there is probably some R/C network int here
but, with a bit of hacking ...
I presume pulse mode uses no additional inductance in the output?
yes and no
on the very old models you had some to determine the pulse form
the newer ones are adjustable
and normal pulses wuold be? 10hz?
our last welding psu can be tuned in a LOT of ways
pulses are 50-300Hz
what wire diameter?
1.0 to 1.2?
* alex_joni can look it up
3 and 5mm
different characteristica based on wire diameter
I meant material type
alu is lower pulse rate from what I remember on a Fronius welder
alu is both :D
lower and higher pulse rate
there's an additional pulse added on top of the normal pulses
so you pulse with normal frequency, then with high, then normal again, etc
so it looks like a TIG (from the result)
it's called AluPlus or something like that
Fronius are the pulse people, right?
ok, so Fe, 1.2mm, 82-18 gas?
robin_sz: all have pulsed welding ;)
cloos introduced it back in 1970
I was thinking of just "playing" with one of my old 500A murex PSUs and seeing if I could do it ..
I have 92-8 .. is that ok?
alex_joni: are you good at gdb?
anonimasu: not at all
cradek & jepler are :D
alex_joni, im guessing, but aroun 30A background, 320A peak?
I meant what medium current are you looking for
oh, for 3mm, 130A maybe
for 6.0mm base material 280A
150A for 3mm, right
ok, so for 150A
OK, that seems fair
6m/min wire, 130Hz pulsing, 35A background current, 36.5V pulse voltage, 1.9msec pulse time
that's for Fe, 1.2mm wire, 92-8 MSG pulse U-I
U-I means you control the pulse voltage, and the background current
there's also I-I, but I never liked that
it's for thin sheets, and very crappy ;)
36.5, thats quite ... high
not at all ;)
well, for *normal* mig it is
yeah, but it's pulse voltage
so you'll get about 300Amps pulse current or so
and hopefully, zero spatter
I've been welding with 50A base and 500A pulse
that's an intense setting
and can you really get zero spatter like they claim?
you can also do the same with 90A, 300A pulse.. less strain on the machine ;)
that would be impressive
you can also get few spatter from spray welding
but spray on 3mm?
you know I hate freenode..
Lerneaen_Hydra: because it's full of stupid zealots.
welcome to irc :p
Lerneaen_Hydra: I asked a question in c and got told I werent qualified.. because I've avoided gdb.
better install linux and start learning
ah, isn't rtfm lovely
yeah kind of ovbious, when you say "I really have no clue about gdb as I have not used it.."
Lerneaen_Hydra: oh, well, the manual dosent really help much..
that's never stopped them from saying that :p
I were temped to tell him to fuck off.
I hate the notation that linux rocks at everything.
it's a blind statement.
anonimasu: it drives my washing machine, therefor it must rock
alex_joni: yeah.. that's the kind..
[14:50:23] <Lerneaen_Hydra> http://antonyms.istheshit.net/
[14:54:24] <Lerneaen_Hydra> http://killerseal.istheshit.net/
Got a quick question
[14:55:38] <alex_joni> http://pick_a_car.istheshit.net/
I have setup emc, but i am having trouble with the direction. the feed goes the opposite way I want. I am sure there is a setting, but not sure where
make input scale negative
okay. I'll try that.
one more thing
I am use to give information on how many steps and lead screw information. I am trying to get the scale right.
maybe if you describe your scenario, we can help
If i tell it 1 inch, it dosen't move one inch
it moves more
ok, what kind of motors?
steppers 1.8 degrees? (aka 200 steps/rotation) ?
then the scale number (or its absolute value) is too large
yes. 1.8 degrees.
ok, what drives?
full steps, half steps or microsteps ?
brand name and model might help too
just 110 oz-in 1.8 degre
so full steps
Do i need to just adjust my scale?
I meant the drive, not the motor now (the box with the power transistors inside)
Guest951: you surely need to
what do the motors drive?
a screw? rack-pinion? is there a reduction?
they direct drive a 1/2 lead
I'm not familiar with the term 1/2 lead
A one half inch lead screw?
that just means that it is a lead screw, with 2 turns per inch
is that a screw with 1/2" / turn ?
so you have 400 steps / inch
[15:01:32] <Lerneaen_Hydra> http://smileys.istheshit.net/
how much does it move if you tell it to go 1" ?
it should be exactly two motor turns
what is your INPUT_SCALE currently? It should be 400 (or -400) based on what you've told us so far.
I don't remember off the top of my head
I think it is 4000
whatever the default is
then it's wrong..
and it would move 10" for every 1" commanded
I'm not sure, I just know it was a lot. Lot more than it should have
do you know how to change it?
I will do that and go from there
this is emc2.. right?
copy the config to ~/emc2/configs/ and so on
I appreciate your help
it's described on the wiki http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?CustomizingConfigsOnUbuntu
Guest951: we are around if you need more help
no problems, bye
[15:09:50] <Lerneaen_Hydra> http://american_views_on_authority.istheshit.net/
[15:21:40] <Lerneaen_Hydra> http://dividebyzero.istheshit.net/
not much happening here today
robin_sz: still there?
was _supposed_ to be in Geneva this evening
so.. regarding that pulsing
be sure to keep the pulse width constant
and change the frequency (aka base time)
that way you have a chance of setting up a good weld ;)
otherwise it's gonna be a bit of a tuning problem
usually I set pulse voltage, pulse time, base current is less important
then wire speed, then frequency
Im not sure I will be able to do it with this welder ... its 3 phase txmr, then 3 scrs or triacs ..
start welding, and adjust frequency till it stops spattering
yeah, probably not
so its just phase control at 50hz * 3
going to need a proper chopper ...
once an thryistor is on, it stays on until the current drops to zero on the crossover point, right?
I think you can turn it off with negative voltage though?
it's been a while since I looked at thyristors
[16:44:27] <SWPadnos> http://video.google.com/videoplay?docid=5125780462773187994
steves_logging is now known as steve_stallings
JMK has got to see that!!!!
I was thinking that ;)
too bad he has no audio on most of his computers :)
.... and not even a hint that it was a parody
nope. they also used all the real logos, which is weird
(at least it looked that way to me)
either it was an in-house skit or the legal beagals have not seen it yet on google
heh. maybe I shouldn't have posted the link ;)
Hello peoples =)
Anyone here have any time to help with an emc problem?
I should be around for a few minutes, and there are probably others lurking, ready to help :)
Well then, I have installed emc 2.0.3, and setup the ini file to include loadrt probe_parport and emc works fine right after I reboot my machine.
Once I close emc and try to restart it, emc can no longer send any thing out to the parallel port.
So I have to reboot the machine each time I want to have emc actually run. =)
hmmm. I'm not sure I'll be able to help with that problem ;)
I haven't found anything in the mailing list archives about this problem
what type of machine (brand, motherboard maker, chipset ...)
Machine is an HP Pavillion, motherboard is TUW-LA (made by ASUS I think)
ok. and the parport doesn't work at all without loading probe_parport, and only works once with probe_parport?
I have tried loading emc with the probe_parport line commented out, and it still doesn't work. I am trying a reboot and running emc without the probe right now.
just to confirm, you can actually run emc a second time but it doesn't seem to write to the parport, correct?
ok, little bit of weirdness
emc works without the probe_parport first time. I'm am going to restart and try it again. one sec
hehe, ok now it is working after restarting emc. I guess the probe_parport was doing something funky.
it shouldn't, but if it's not needed, I'd not use it
hmm... well this was the easiest fix ever. =)
glad to be of service
(phew - it worked ;) )
hmph, I can't get it to break now. I guess that was the only thing screwing up my configuration. Thanks SW!
err - no problem
haha, great film clip SWP
steve_stallings is now known as steves_logging
Quite the video.. Very odd. :)
I picked up an interesting linear axis this week
a linear rail, a motor and a umm threadless ballscrew
weird thing it is ...
its basically a shaft, about 1m long driven by a DC servo
and when you turn it, the carriage moves 20mm for each rev
its gripped by 3 cam roller bearings, at about a 10degree angle
suprisingly, it needs a good 500N of force before it slips
well, 250 maybe
Our accuguage machine (sort of a optical computerized comparator.) has a drive like that . It has a shaft - 3 bearings mounted at angles to create the movement. then there is an external glass scale to do the postision feed bac.
cool, sort of like those stepless-transmissions that have movable rollers for different ratios
[19:42:18] <Lerneaen_Hydra> http://en.wikipedia.org/wiki/Continuously_variable_transmission#Roller-based_CVT
SWPadnos: regarding http://video.google.com/videoplay?docid=5125780462773187994
I've seen it - it was an in house thing done by the marketing folks
have two of my drives hook up to steppers
and they buzz a little when I try and move either axis
but no real rotation
drives are getting 24v from a ~18A supply
5v logic tests good from the comps psu
they're bipolar motors so it's hard to wire them wrong
what drives are they?
iirc if you swap one of the windings it will hop between two locations
oh pooh, hes gone
I know the answer to that one :)
he has the step and direction lines swapped over on the drives
the buzzing is the step pulses hitting the direction pi
if there are no pulses hitting the step pin, you should be able to send anything you want to dir with no effect whatsoever on the drive or motor
well, as you like ... but from past experience, if you get them crossed, thats what you get
or one phase not connected
misconnection on the motor side makes more sense to me
I suspect then he is just driving one of the two windings
jmk, were you the one working on backlash and 2x maxvel/accel?
Lerneaen_Hydra: currently no-one is
multiplier on the current set resitors was _brown_ and not orange
hmm, I thought it said that jmk was on sourceforge
the most recent attempt to fix that code was by someone whose name escapes me at the moment
ejholmgren: too less current?
put the right resitors in and it works fine
jmkasunich, well, there went my theories ;)
[21:44:45] <Lerneaen_Hydra> http://sourceforge.net/tracker/index.php?func=detail&aid=1523126&group_id=6744&atid=106744
<-- it says that you're assigned to it there
assigned to it maybe, but not working on it at the moment ;-)
jmkasunich, did you see the earlier msg about the linear axis I got with a completely smooth "leadscrew"??
lines 2217-2240 of src/emc/motion/control.c are a block of comments from Bas that pretty well sums up the issues
there is a linear stage ... with this leadscrew, completely smooth shaft and when you turn it, the carriage moves 20mm/rev
theres a "nut" on the shaft, but its totally smooth, no grooves at all
confused the hell out of me it did
the "nut" has 3 cam roller bearings, 60 degrees apart set at about 10 degrees to the shaft
gripping it tight
jmk, sounds nasty
robin_sz: that reminds me of the latest wire feeders
2 bearings tilted at 45 degrees rotating around the wire
this was on a picka nd place machine, so it didnt need a lot of force
what happens when that speed tries to go faster than the TP speed?
and there was a rennishaw encoder to keep track of wher eit really was
or faster than tp_thread
alex_joni, who does wire feeds like thatr? kloos again?
bah, g'night everone
robin_sz: most robotic equipments are like that lately
but only for the push-pull
so, some sort of spinning nut with the bearings in?
ok, question ...
why does it not just twist the wire ?
not enough space, and they press from the oposite parts
can't find a damn picture
it's called a planetary wire feeder
[22:05:43] <alex_joni> http://www.weldingdesign.com/Content/Site323/Articles/GWD/09_01_2005/11636Aluminum03_00000005220.jpg
can't google for any better pic
so the bearings press agaisnt each other from each side of the wire
they are tilted 45 in opposite directions
but the wire could just twist up, no?
twist? I doubt that
there are some other bearings before and after the feeder to amke sure ?
I mean, if you put say, 100mm of wire in it .. it woudl not feed, it would just spin the wire , right?
cia is sleeping again
* alex_joni kicks CIA-8
cradek: see.. not sleeping :D
just lazy as usual
maybe we should get NSA or FBI
03cradek 07HEAD * 10emc2/nc_files/g76.ngc: optional tapered start and end for the threading canned cycle
03cradek 07HEAD * 10emc2/src/emc/rs274ngc/ (4 files): optional tapered start and end for the threading canned cycle
cradek has changed the topic to: Welcome! EMC (Enhanced Machine Controller) is a linux-based CNC control. | Latest release: EMC 2.0.5 | http://www.linuxcnc.org
EMC 2.0.5 is released
03cradek 07v2_0_branch * 10emc2/debian/changelog: 205 release
cradek: I'll do the SF and linuxcnc.org stuff in a bit
alex_joni: thank you
alex_joni: would you check the wiki too?
03cradek 07v2_0_branch * 10emc2/VERSION: release
wow, cia is way behind