awallin; hey there, I just saw the steel bulb video on your page, any reason to make it from steel rather than lead?
lead being easier to machine and all
We've made lead bulbs by casting. In aluminium moulds. Don't know where we could find a big enough chunk of lead for milling
oh, why can't you cast ingots of lead in roughly the right shape?
and then machine them to the right shape
you can even recycle the chips
we could try that. but perhaps we'll do the prototyping in steel and when we find a good shape we make a mould in aluminium and can cast lots of that shape bulb in lead. ofcourse the volume's aren't going to be the same if we want the same weight bubls
awallin; hmm I see, how does the mill cope with steel? is it still based on the BF20 (the milling column and XY table)?
just the column and the base are original. new table, spindle, linear rails, ballscrews
oh. that's not a whole lot left. how stable is the base? I've found that when doing relatively heavy work that it deflects a bit too easily
I don't know it deflects. we usually work in aluminum
pcnc is now known as SWPadnos_
hi folks - got a question about estop enables, charge pumps, and stepconf
maybe 2 questions
was that Romanian?
nope.. english/IRC :P
so, I set up a machine with stepconf, with charge pump on a pin
well.. actually DATAJA.. but too lazy to push Shift
but for some reason, the charge pump never gets enabled
there's an ESTOP input, which appears to be correctly connected
unless estop-ext==TRUE means "estopped"
but the machine is in the on state, so that seems to bo OK
I'm looking at estop-ext (connected to iocontrol.0.emc-enable-in)
It may well be that true is estopped. Some board makers don't know the truth.
iocontrol.0.user-enable-out connects to charge-pump-enable, which I think may be wrong
isn't that iocontrol output only "on" when the user is pressing the F/machine on button?
so the status right now is that axis shows that the machine is on (and I can jog around), but iocontrol.0.estop-out is false, as is ...user-enable-out
I've not looked very close at the output of stepconf for this. I do know that a fellow in portland, OR got his going with a pmdx131 using stepconf.
and nothing changes when I hit F2 (at least none of the signals change), but it does change whether I can jog or not
F2 is machine on in axis?
I need to check this once more, but here goes
(note that the charge pump is not relevant to the external enable input, it only disables the spindle/coolant)
hit external estop, but don' treset it - that turns off the estop-ext signal (connected to iocontrol....emc-enable-in
reset the external estop, so the machine looks ready, no charge pump
charge pump is on a pin of it's own?
there's a "machine ready" input to EMC
Okay. What turns that off/on
there's a charge pump output to the machine
just an interlock: big red button and start button with a contactor
Okay. So that is an external estop.
I'm trying to get the charge pump to turn on when I have machine on
for some reason estop-out isn't reliably turning on
which is iocontrol.0.user-enable-out
f2 has no effect on iocontrol.0.user-enable-out
that I can see here.
it almost looks like iocontrol.0.user-enable-out is inverted
With a loopback it does reliably show the status of gui estop
That's with mini. Let me start it with axis.
with the external estop activated (emc-enable-in false), iocontrol.0.user-enable-out is on
I wanted to show off my almost finiched cnc machine :)
[15:32:30] <JanVanGilsen> http://imagebin.org/17793
[15:32:32] <JanVanGilsen> http://imagebin.org/17794
ok, here's the pattern
WOW that is great work JanVanGilsen
with the external system "ok" - which puts EMC in "estop reset" state, if I ever hit F1 to stop EMC, I can never get the charge pump back
with external estop "not OK", I can hit F1 to turn the charge pump on, but never off
well, at least there's a pattern :)
SWPadnos, That third estop pin user-request-enable might be what you are looking for.
that never activates when the external estop is "ok"
Look at the way it's used in the demo step cl
I'll do that
It's a very short reset.
I'm not sure this machine is so different from others though
In fact the fourth row in run 1 shows it.
there's a machine ready input, and a charge pump
it seems that this setup shouldn't need CL, but maybe it does
there doesn't seem to be a machine on HAL bit
No it wouldn't need cl just an or between enable-in and user-request-enable
Is there somebody who made/converted a tilting rotation table machine with EMC? Else I'll just hope that my kinematic calculations will work.
I believe you are the first, JanVanGilsen
arrg.. but its also nice to be first
There are folk who can help with that kinematics if you find issues.
I've though about that exact configuration quite a bit over the years.
thats nice, I'm hoping to test it this (or next) week
Fantastic. What are you using for motor drives?
SWPadnos, halui has machine on and off and status.
SMD microstepper drives (8 microsteps)
Got a link for those?
[15:51:06] <JanVanGilsen> http://www.jvl.dk/files/pdf/lb043gb.pdf
machine on is available in emc.status, but apparently not in HAL
SWP you can find it in halui.
yes, true :)
You could also use halui for the estop stuff if you wanted to.
it does seem strange that I can only set user-enable-out when emc-enable-in is off, and I can only reset user-enable-out when emc-enable-in is on
I started halui-halvcp
and see that when the machine is out of estop and on
halui.estop.is-activated shows false
but iocontrol.o.user-enable-out is true
and if you toggle what would be the external estop (connected to iocontrol.0.emc-enable-in), then iocontrol.0.user-enable-out will go false
and not come back to true?
which you can turn on with F1 (but can't turn off again until user-enable-in goes true)
I think I'd edit a copy of demo-step-cl to match your estop pinouts and ins.
yep, that may be the trick
plus add the charge pump
or just add halui :)
Then start it up and select the fourth row in the estop rung
and try the external estop.
That momentary signal from user-request-enable might do the trick.
it doesn't seem to activate when emc-enable-in is true
SWPadnos_: that's the intended behaviour
yeah, explain! :)
alex_joni: whats the reasoning behind it?
JymmmEMC: about 2 days of work
jmkasunich and myself worked a lot of doing this right, but it was 1-2 years ago, so I'm not up to speed about details
it's documented there though
Mind if I ask a OT question?
Mind if I ask a OT (programming) question?
oh, an ot computer programming question
not a g-code thing
alex_joni: I remember the conversations, nut also don't re,e,ber the details
I guess the motor-enable outputs could be used in place of machine-on, since they more or less reflect that
I'm writing a function to give duration of time, but only using a single UOM (130 minutes), instead of multiple UOM (2 hours, 10 minutes). But I'm not sure where the breakpoints should be between each UOM? Example:
what's an UOM?
what's the input?
unit of measure
UOM == Unit of measurement
so you always want only one?
so the input might be the string "90 days"?
that would be the output
so you need to find the highest UOM where it still fits?
switch and a series of MOD's ..
now - timestamp == single UOM
that will almost always end up as seconds, due to floating point rounding error
now - timestamp == duration
duration will rarely be an exact multiple of seconds, minutes, hours ...
so without some fuzz factor, you'll end up using the smallest unit all the time
well, it'll be epoch, but convert to single UOM for readability
Do you think 3x the value would be about right?
I guess what I'm saying is WHEN does the significance begin to dwindle? Saying something happened 95 minutes ago provides enough accuracy, but saying it occured 1.5 hours ago doesn't (assuming you round off to 1/10's of an hour)
I'd rather not use decimals if at all possible
just whole UOM
silly question: if I press "home", will esc abort that?
I believe so
but if you break the machine trying it, I will deny saying that
damn. I'm not logging ;)
SWPadnos google is
yes, it does :)
logger_emc is logging
1) '179 seconds ago', add one but display '3 minutes ago' instead of '180 seconds ago', sound about right?
but it sounds strange
what timespans do you expect?
seconds, minutes, hours, days, weeks, months, and years.
(I would always go to a higher UOM if it goes over some treshold)
so instead of 210 minutes, maybe display 3 hours
alex_joni: Exactly, but I'm trying to determine what that threshold is/should be
or something like that
it's UOM dependent
I'd make an array that holds the relationship between each unit (ie, a minute is 60 seconds, an hour is 60 minutes ...)
the threshold depends on what you are using the information for
alex_joni: Right, but whole units only, no decimal
but generally something like 2 times the UOM
120 minutes turns to 2h
you build another array with the number of base units (maybe seconds) for each larger unit
48h turns to 2 days
the issue is when you have 150 minutes
7xx days turns to 2 years
do you want to round that down to 2 hours?
jmkasunich: That's why I was thinking 3x
for each input, start at the largest unit, and search until you find one that's less than the input value (in seeconds or whatever)
JymmmEMC: yeah, but 1214 days might be a bit hard to parse by people
once that's done, find the largest unit where input mod (unit size) == 0
alex_joni: that would be months or weeks
or where input mod (unit size) is < unit_size/something - ie 183 minutes is close to 3 hours
2) 72 hours ago + 1 == 3 days ago
I was thinking the other way - convert to the finest UOM (seconds) - that will be a big number. then start converting to larger units until the number is less than your threshold
but 6000 seconds mod 1 == 0
you'll always get a perfect fit for the smallest unit
so you want to find the largest unit for which the fit is close enough to perfect
SWPadnos who are you talking to? I'm not talking about doing any mod stuff
Days get tricky as it's between weeks and months I think.
actually. I'd probably have a different threshold for each UOM
"if more than 20 days, convert to weeks" "if more than 7 weeks, convert to months", etc
jmkasunich: Yeah, I dont think a single therhold will do at all
ok, what about this.... the higher UOM will never be less than 3?
3 seems kind of low, but again, this depends on YOUR application
do you need to be within 50%, 20%, 10%?
the problem is NOT periods that convert exactly (or near exactly)
21 days can expressed as 3 weeks with no loss of accuracy
but what about 18 days? calling 18 days "3 weeks" is an error of 15%
Lets look at it.... 90sec = 3 minutes, 180 min = 3 hrs, 72 hrs = 3 days, 21 days = 3 wks, 90 days = 3 months, 36 months = 3 years.
ooops, I forget weeks in there... 21 days = 3 weeks
weeks is in there
the tricky part is that you go from days to weeks, and from days to months, instead of days->weeks->months
also, you got the first one wong - 180 secs = 3 mins
oley, 180 sec is fine though.
4.2857 weeks in a month, isn't exactly a natural conversion
but for this purpose it works fine
12.857 weeks = 3 months
lets round it off to 13 weeks
13 wks = 3 months
do the computations with full precision, and round off only the final answer
the algorithm is:
convert time to seconds
then compute minutes (div by 60)
if less than 3, use seconds
else, compute hours (div minutes by 60)
if less than 3, use minutes
else, compute days (div by 24)
if less than 3 use hours
since you need to store the division ratios for each UOM, you could also store independent thresholds for each UMO
it's the weeks threshold that I'm still not sure about (not the math) limit at which it goes to months.
if months < 3, use weeks
but is 3 months "good enough"
so make it 4
jmkasunich: I think I found the 'whitelist' request page at my isp for the mailing list
thats my point - each UOM can have its own threashold
we'll see how it goes
btw, I don't have data on the number of outgoing messages, so I can't easily answer that
I think the archives page shows mail volume by day tho
thanks for looking at it for me. I'll let you know how it turns out.
jmkasunich: I get that, it's WHAT the threshold should be for each UOM. Just not sure about the weeks to months aspect, be it 3, 4, or 5 for that matter. If you wanted to see how old something was, does '11 weeks ago' give you enough information compared to '3 months ago', would you rather see a higher 'weeks' value?
and I keep telling you, it depends entirely on the context, which only YOU know
jmkasunich: ticket/job/odering system.
pregnant women count in weeks, because they care about +-/ 1 week even after nine months
right, but is that the only exception you can think of by chance?
DanielFalck: a dozen verizon.net users got their subscriptions disabled today. not sure if it's verizon's or sf's fault but something "new" is happening
I'd probably use 3 as the threshold for every UOM except weeks to months - switch to months after 4
cradek: I told dan that by email earlier today
ok, more "interesting" behavior :)
I'm sure this is known
(but not to me, because hey - I don't have a CNC yet :) )
star ta program with radius comp, and some concave corner somewhere later on
dialog pops up more or less immediately telling you about the problem
then the machine runs until it gets to the segment that errored, and stops there
thats probably the motion buffer
makes sense, but still a little surprising for an operator
the interp runs 200 segments ahead
yep, queued motion
yes - it stopped on line 192 :)
actually I'm surprised you don't get the message when you load the program, while axis is doing the preview
I think axis does something to cutter comp
when it's generating the preview
I'm surprised by that too
I have a G61 "remarkableness"
woot.com temperature monitor Voice alert tells you "Your stepper is almost toast" and "your stepper is an ex-parrot"
they sell a different item each day.
yeh, since you showed me i check once i a while ( you a bot? thats exactly the same reply!)
sometimes I feel like it, latelt feel like it a lot. Got Oilcan?
If i use g61 i got 1-time an axis following error. after that g61 looks enabled but dont work
The config is from stepconf-wizard
DanielFalck: still there?
I noticed a "Bounce action notification" from 11.05.2008 (dfalck AT verizon..) Subscription disabled, Reason: Excessive or fatal bounces.
JymmmEMC: i think you should just switch to a metric day - 100ksec is close enough
then you only have to keep the last three digits (unless you want to use scientific notation)
first three digits*
JymmmEMC: i even made a clock for you: watch -n 1 'date +%s; echo "G M k s"'
good for the next 275 years at least
alex_joni: thanks. I figured that was going to happen
I am trying to get the mailing list added to a verizon whitelist for my account
fenn: we'll just use swatch time instead.
BUT WHATS A SECOND AFTER ALL
naw will never sell
why do VTLs look like they're so much fun to use
because they are
Dracula's Castle is for sale.... 3.2 Million USD
JymmmEMC: not the real one
it's a fake :P
alex_joni: It is?
where's the real one?
toast... look up the cl 84 by canadair... and it was easy to use
JymmmEMC: I'm not telling :P
that would be a steal....
Egor... "heh heh heh yeeeees master"
JymmmEMC: actually there are a couple which are accounted for as "dracula's castle"
toast it was vstol
alex_joni: did he have multiple castles?
dmess: vertical turret lathe, not vertical takeoff and landing
JymmmEMC: depends what you undestand by "he"
you do know that dracula is a fiction character.. right?
i want multiple castles
that would be so excessive.
oh hell they are ALOT of fun to run too.. especially the toshiba MPF.. that milled too
alex_joni: I suspect it's based upon RL
dracula came from vlad the impaler
alex_joni: a tyrant or something along those lines overembelished
toastydeath: most likely
RL real life
yeah, but no vampire really :)
hmmm - is freqgen deprecated, or just not meant to be used for velocity control of steppers?
alex_joni: metaphorically perhapse
[19:49:12] <alex_joni> http://en.wikipedia.org/wiki/Vlad_III_Dracula#The_vampire_legend_and_Romanian_attitudes
SWPadnos_: both I think
vlad the impler is real in Romania.... he masacred many
it doesn't look like ctrl-type=v really does the same thing as freqgen
some with good reason
(and the need for a modern version is also quite debateable ;P)
JymmmEMC: the most likely castle one can refer as Dracula's Castle is : http://en.wikipedia.org/wiki/Bran_Castle
and I really doubt that one is out for sale
and even if it would be.. it would be 10-20 times the price at least
"Michael Gardner, Chairman and Chief Executive Officer of Baytree Capital, the New York investment firm which has been retained to create a plan for the castle and to sell it, predicted it would sell for more than $135 million, .."
[19:53:48] <JymmmEMC> http://www.usatoday.com/travel/news/2007-07-02-draculas-castle-for-sale_N.htm
right.. $135 million is something more like it
for that dump....forget it
Dude if the mold in the dungeons is less than a few hundred years old than it isn't ripe yet.
still a steal.... pass 15 million ppl thru the doors as is at 25 e per head... you kick thit in the head fast...
you could make a viable 5 yr plan with a recoverabe return on investment in 7 yrs or less
"Some 450,000 people visit the castle every year, Gardner said."
its DRACULA's castle after all... it would be world wide visitors... ; )
dmess: don't forget royalty fees for filming & events
weddings and funerals would be worth a mint
im good for moving to Romania
any where but home.. ; (
havent heard about my Russian connection..
I had a friend buy me SolidWorks in Russia while he was there - only problem is the support contract requires me to send any help issues thru the russian distributer :~
cool... how much??
[20:07:11] <alex_joni> http://www.popgive.com/2008/05/fireman-on-marijuana.html
Was like $1200 USD
at the time
favorable exchange rates then too.
that video... is a blast... LOL
the installer just noted that the existing O/S was set for english so it loaded it in English - so I have tonnes of help files on the CD's which are in russian.
I'll see what i can do about getting ya english ones
Oh all the english ones are there too
its just that my set came with extra CD's that the US release doesn't bother to include.
as crazy as it is - 5 axis software is a controlled export item.
wait for it to be open source first :D
im cleared quit high... could i carry it...??
is that still true (I remember the incident that started the restriction, but that was years ago)
hmmm. isn't motion.spindle-on supposed to be on whenever either spindle-forward or spindle-reverse is on?
was still true in 2002 - but that was the last I had to officially deal with export controls
SWPadnos_: and speed > 0
yep, got that
I like AWallin's newest video, what was the original base machine?
[20:23:34] <SkullWorks_PGAB> http://video.google.com/googleplayer.swf?docid=916833244018390029&
very nicely done... drop an INTERNATIONAL MINI_CUT tool in there.. double ALL feedrates... leave the rest the same... and BARK
theres a litle hysterysis to be seen in the finished model... but after all thats why we program to +.005" and polish/blend to make it pretty... we do it all the time
or you feed it a smaller dia cutter with YET finer resolution..
he noted on his blog that for a good part of the front he had it at 150% FO.
ROCK & ROLL.... ;)
that mini-cut throws a rooster tail that is SWEET... and catchable... been there done that..
.006"-.010" per tooth.... she loved it... ; )
fenn_ is now known as fenn