[#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/,
wiki at http://wiki.linuxcnc.org/"
annnd my lathe stops feeding
"Core Cell chip extreme you've ever needed!"
(caption under a transparent-capsule thing logo when this motherboard boots)
All this exciting stuff and I did nothing today except hang around with the grass-dogs http://filebin.ca/uwhdzk
looks like they want you to entertain them
"wonder what the two-legger does next"
I point out food to them so I'm an honorary sheep
so - does anyone know what the distance limit of single ended encoders is? and B) does anyone know a source for line drives?
I think we start losing postion at around 60ipm
we kinda thought we might have a problem with that. the distance is a good 15ft.
skunkworks: are you positive your 7i33 jumpers are set right?
yes - it is set for single ended. well - I will double check. but I am pretty sure.
when I was working on my (single ended) spindle encoder I scoped it at the 7i33. the roundoff is very apparent and it's easy to see when it'll all go wrong
Sometime 1K or so pullup resistors on ABI can help
15 ft is a long way for single ended...
well at 60ipm that is 60khz
did I do that right?
60960 counts per inch. *60ipm/60sec/min?
Often single ended encoders just have the opto detector comparators (open collector with a pullup)
driving the lines so the rise time of the pullup resistor driving your cable capacitance will limit your max rate
crosstalk could also be an issue depending on cable type (say 30 pF/ft = 450 pF = 1.5 usec glitch from one output driving low into a say 3K pullup on another output)
This could be mitigated by setting the encoder filter slower (You need to scope the signals to see whats going on)
I will have to play with it. the cable is a 4 conductor shielded with no twist
shield is hooked to the - on pin 1 iirc
of the 7i33
pcw_home: did you see this? http://www.youtube.com/watch?v=FgOqEz5Tk-Y
that's not really all that impressive until you notice it's a tenths indicator.
I was giggling when I saw how it was doing.
Yes, I did. Look really good especially since you are just starting to tune it up
cradek: I think I am getting what you were saying about tuning - if the ff1 is set to 1 and things are scaled right - the drive should do exactly what emc whats it to do without any exta tuning.
emc doesn't have to do much extra work for the axis to follow.
compared to a voltage mode setup.
Yes, do as much as you can with the feedforward terms
Then the PID loop (with its limitations) has less to correct
I leave I at zero - if I have offset when stopped, I fix it with the amp's knob
are there line drivers out there? I am not finding anything. Not a popular product?
us digital makes one
it plugs into the back of their encoders
ah - ok. I think I ran across that - I didn't know what I was looking at - thought it was part of the encoder.
For long single ended cables (which I do not recommend) you would be better off with a twisted ground for each signal
Line driver is Just a 26LS32+VCC bypass on a PCB
If you can make PCBs thats about as easy as they get
heh - I will look into it.
does the a(not) actually go negative then?
signals are like TTL levels just differential
(I meant 26LS31)
heh. well I must be dense tonight. ;) I don't see why it would be much better. - just that your referencing a - a(not) instead of a to ground?
it has been a long day :)
the receiver looks at the difference between the signals
so even a /1/2V differenec will be detected properly with common mode of -7 to +12 or so
induced noise will be induced to both A and 'A almost equally (since they are a twisted pair) so our little differential TTL signals
can be detected with 10 times as much common mode noise as signal
ah. thank you
whereas 1.5V of common mode noise will put your single ended interface out of operation
For a convenient differential driver (or receiver), I'd look at USDigital: http://www.usdigital.com/products/interfaces/encoder/cable-drivers/ea/
spec sheet at: http://www.usdigital.com/assets/general/104_ea_datasheet_0.pdf
The CUI AMT10x series encoders have a "cable lump" driver/rcvr too, hang on...
OK, here's the driver, still looking for the receiver... http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=102-1787-ND
Maybe they don't supply the receiver because it's "not their problem"?
don't need a reciever - got a mesa 7i33
Hey a dip 26LS31 some hotmelt and shrink tubing and youv;e got a cable lump driver
KimK: one more time.. ;) did you see http://www.youtube.com/watch?v=FgOqEz5Tk-Y
Yeah, that's all they are. But they did all the work, lol.
pcw_home: I would have no problem doing that :)
what do people use for cable then? seems shielded cat5 would work well.
3 pairs for abz and one for +/-
[03:42:59] <SWPadnos> http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=296-14971-1-ND
that's a transceiver with 2 ins and 2 outs
there are probably others that have just ins or outs
I like soldering...
Isn't cat 5 usually solid?
yes - but remember my servos don't move
shielded cat5 only has an overall shield I think, not per pair
oh - I need a shield per pair. that seem expensive ;)
I don't know that you do, but then again ...
Doubt you need a shield per pair
pcw_home, they do make stranded cat 5 for motion
Or use a bunch of runs (if you already have the CAT5)
OK I know they make solid and stranded plugs
pcw_home, do you know of a good isolated differential xcvr chip?
alex_chally: That's exactly what USDigital uses for their (finely stranded) encoder cable. It's even color-coded the same way, white-blue/blue-white, etc.
SWPadnos: Analog devices make some
the adum1100 series maybe?
(or were those the little A/D's?)
Dont know off hand look at their interface section
ok, thanks for the pointer
they make some with isolated power as well
yeah. I think I didn't find anything great, so I ended up with one of those little 1W DC:DC isolated power supply chips plus an isolated single ended buffer, plus a differential transceiver (powered by the isolated supply)
more complex than a single chip, that's for sure :)
skunkworks: No, I had not seen your video. So that's 0.100" of motion? Nice!
no - .05" and the minor divisions are .0001
The neat thing about Analogs isopower chips is the power supply has amazingly low capacitance across the isolation
the bad thin is that the power supplys are quite inefficient
OK, 0.050", then I stand corrected. I thought it was .010" per rev and 10 revs.
yeah, I like the integrated power supply
as cradek pointed out - it overshoots a half a tenth. I am still giggling
Now this is just on encoders, right? You haven't (yet) tried to bring back the Accupins, have you?
no - just encoders
accupins are way down the road
ok - I think it is time for bed. I am starting to see double
Thanks for all the help guys. It's moving!
Ha, OK, goodnight. Glad i got to hear the good news.
'nite again congrats on first controlled axis movement!
(with a Mesa 7i33!)
heh - the first 2 tries where not controlled... ;) I inverted the tach back to the emp. run away - run away!
I assume you did this with the belts unhooked!
* KimK guesses that skunkworks unhooked the belts for the *second* try, lol
Can't fool skunkworks more than once!
Ah, he's gone, lol
Be quite treat to see that really old iron doing useful work again
and with that I'm gone as well
root is now known as Guest96675
Guest96675 is now known as new2emc
Fox_M|afk is now known as Fox_Muldr
Hi, I have to build a new breakoutboard. Gecko uses a HCPL-2531 optocoupler, it is very fast (1 MHz), but I would have to buy some. What I have at hands is a Sharp PC123 coupler. Will it be good enough for steper driving?
how fast is the PC123, and what is the max step rate you are going to be using?
I read about Cuttoff frequency of 80 kHz
that doesn't sound very fast, if you are going to generate steps at 10 or 20kHz
but if the gecko already has an optoisolated input, why do you need isolation in the breakout board?
Gecko uses a breakout board in the G540 drive. The G540 includes four G250 modules. I use the G251 which is almost the same and also has no opto-isolators inside.
How fast can a parallel port be at all? Parallel ports never were really fast...
20khz if you are lucky
So why should I worry about 80 kHz to be "slow" ? ;)
it might work ok
Very good chances, or bad chances?
if the PC123 datasheet has a picture of a 80kHz square wave and it looks like a square wave, not a sine wave, then you are ok
Ah. Ok. Ill look on that.
Are Geckos very sensible in the signal?
I don't think they are very picky
So, as I am surely below 20 kHz (I use a VIA C7 CPU with 1000 or 1500 MHz), the signal will surely be better as at 80 kHz.
Hi all from sunny germany Thunderstorm ahad
i made a final and upload the new grid generator pleas feel free for testing -> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Simple_EMC_G-Code_Generators#Grid_Rectangular_Circular_SpiderNet_G_code_Generator
and here is the video of testing http://www.youtube.com/watch?v=QtCabDDm_EI
IchGuckLive: is it only for positioning holes for drilling? or for actually milling the grid?
milling the grid offcause
you can ad a drill at all pints by adding G81
so thunder starts now by i got to safe the satelite dish the solarsystem and so on !!!!!
whats up ?
lazy day today
well yeah sundays
just the usual round on ebay here
where are u from ?
i am looking for some proper bearings and ballscrews right now
ballscrew - as a leadscrew ?
gotta start somewhere
i am thinking of adding encoders to my machine
ah already got a working mill ?
nice, buildt yourself ?
do you want to build a machine from scratch ?
yeah ill give it a try
pretty sure it will take time
a year or something
hoping to be able to make it in less time than that
last project took 3 months
its a matter of getting cash in fast enough so the build doesent stop:P
got some plans from rockcliff
pretty straight forward stuff
not to invent the wheel a second time
i will make another later in aluminum
but the costs was huge for that one
what kind of machine are you building?
awallin: rockcliff for now
mine is a proxxon mf70
ok so it's a router made of MDF...
wait ill fetch a photo
neat little thing that proxxon
its big enough for my purposes
so its a manual mill converted to cnc ?
have you done any estimates on the total ?
for the whole thing
not bad :)
i figured out it was best to build something basic first before i take on the task with the metal milling process
wouldnt like to crash that one because of lack of experience
ofcourse im not going 100% with the plans
i thought in getting a manual mf70 and a cnc kit separetly
but i deciced to get a the ready made one
so it was more expensive then because it was readymade
but i was worth it because i lack the experience
maybe next time
yeah i buildt a reprap mendel before so i am familiar with some of it
cool stuff a reprap
i use my machine for doing pcb engraving
just for prototypes
havent etched pcbs in 15 years, i buildt the electronics for the reprap myself but i bought PCB readymade
will probably get a chinese board for the mill
servos or steppers ?
nema 23's i think
have to keep the cost down for this "preview" build
mostly to learn the basics of wood milling and how to set up the electronics
do you do the build for yourself ?
no one else here have a clue what it is and doesent care about knowing it it seems
my friends are just saying "huh whats that CNwhat?"
sounds somewhat lonely
here its close the same
its first when the stuff is finished ppl realized what its all about
I enjoy the build/think process more than the use ;)
you are a real inventor
i need to go
im back after a heavy thunderstorm here in Germany
here is the grid -> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Simple_EMC_G-Code_Generators#Grid_Rectangular_Circular_SpiderNet_G_code_Generator
can you check if this works for you
The link works.
[16:24:07] <IchGuckLive> http://www.youtube.com/watch?v=QtCabDDm_EI
here is the test video
Jymmm: python file in axis? for you to
IchGuckLive: I don't have a emc setup here at the moment.
i only got it run on lucid in simulation not RTI mode
anyone here than can check this script in Realtimekernel on axis emc?
so i go to bed its late here
Is there some purpose made preprocessor for EMC2 G-code.
ichudov, there is one that works, but I can't for the life of me remember
I *think* it was either the fanuc or fidal one
What I mean is, say, read G code, and in addition do some macros, or call truetype-tracer to get its output etc. I am thinking about tasks like creating a panel, with cutouts and numbers engraved.
ichudov, have you looked at he g-code tools on the emc wiki?
alex_chally: thanks, I will look. What I was looking is a preprocessor, from which I could invoke other generators.
gotta go, talk to you later
has anyone built something like this? http://www.youtube.com/watch?v=a2rM5Zy5T5Y
and used emc2 to control it?
could it be simulated with vismach?
I think fenn had a hexapod vismach model
maybe it's even in git - not sure
I need to learn how to do that then, with some custom kinematics...
[20:12:20] <awallin> http://fennetic.net/bfi/hextatic.jpg
[20:13:23] <awallin> http://fennetic.net/bfi/hexagui.jpg
the second image is vismach, I guess?
well this is fun: http://www.youtube.com/watch?v=Du2f-EUDqio
KimK: thanks for the link of the CUI AMT10x series encoders have driver/rcvr. I think that is what we are going to try
cheap and easy :)
skunkworks: You bet. We're going to use at least one here to test a CUI AMT10x against a USDigital of exactly the same resolution. At least that's on our list of things to do, lol. But right now we do seem to have bigger fish to fry.
Good to see the K&T moving. X & Y both, or just one? I think you said Z was a different situation?
Fox_Muldr is now known as Fox_M|afk
KimK: z is pretty close to moving
y still need a bit of enginering but not a big deal
that is x and z
That's some big iron.
:) the only way to go?
you where standing on that end when you were down
(they were not mounted yet)