but this...I have some dexterity. I would like to see mty oral surgeon hand solder 602 surface mount resistors haha
les more like a PITA due to the sharp edges till it wears down.
And I am out in the woods...but plenty of Vicoden and penicillin in the emergency kit
les : Ironically, mouth wash does a LOT better for killing tooth pain than ambesol.
antiseptic mouth wash
Oh I have ground down sharp edges with files haha
les There ya go.... sell your oral surgeon a CNC tooth extractor
no pain here...It's a root canal...it's dead
gutta percha core
darn it, I can't find that E-STOP PS circuit on Mariss's website
hmm my machine ids big enough to lay several people on it
problem with the 2200 lbn force capability
if something goes wrong
otherwise I would have massage gcode
people actually asked
Time for me to go - Think I picked up a cold whilst in the States.
les : do you remember that E-STOP circuit MAriss has? I can't find it now
uses two buttons and a relay
Ah, found it!
Hmmm, it's for servos. Wonder if it'll work for steppers too?
I don't know about mariss' estop
I have a circuit on my site
industrial ones cannot use the computer
have to be hard wired
agh missed it
[00:25:07] <Jymmm> http://groups.yahoo.com/group/Xylotex/files/
then select 'servo E-Stop Circuit.pdf'
oops I am not a member of that group
Later guys, time for dinner. Thanks for the help with torque mode stuff Les.
ow ow ow I am trying to pull that tooth shard out
Dang I did it again. 8-)
super glue it back together?
we own devcon.
I have to go...I'm bleedin and stuff
les plasma cutter does carterizing too
5 lbs of silver nitrate?
I need to pick some of those up too... silver nitrate sticks (sterlized)
bye for now
see what living in the woods is like?
aw...no vicoden...I'll save that for when on eof us cuts our hand off.
an aspirin and a stiff drink will do.
Now THAT's entertainment
03jmkasunich 07kbuild-0-1 * 10emc2/ (scripts/realtime src/rtapi/Makefile): changes to rtapi.conf and the realtime script to properly load and unload RTOS modules on kernel 2.6
03jmkasunich * 10emc2/src/libnml/posemath/_posemath.c: Defined _GNU_SOURCE before including math.h in posemath.c
sounds super exciting.
everyone must have tired out from this afternoon.
* Phydbleep has been cleaning the house, yard and shop.
i got a cheapy cheap 2x4 and dremel tool foam cutter running today
emc is pretty simple to use
my friend built this thing.. it had wires hanging off everywhere, twisted together contacts, no thrust bearings.. ugh
i used a whole roll of electrical tape just tidying it up :)
geez some of these "hobby" driver boards cost as much as a new lathe
Yeah, That's why I'm working on the $150 idea.
dremel for cutting EPS ?
Jymmm, yeah, mostly just for lost foam castings
why not hot wire?
hot wire sucks
it smells bad, leaves a crusty layer on the surface, gets stringy goo everywhere
Jymmm: Yeah, Especially for complex castings... Stacking layers is work. :)
and you can't do normal endmill stuff
you're doing it wrong then
if stringy == too cold
crusty layer == not fast enough
well anyway the machine was made that way
and still too cold
and i don't believe for a second that going faster will make it not have a melted surface
the problem is that once you have the melted layer you can't evenly sand underneath it
I've designed enough foam cutters to know better.
in various types of foam too.
Hot wire is ok for model airplane wings, but i would not try a complex 3d shape with it.
i'm trying to figure out how limit switches are implemented in emc
Tied in series for NC and in parallel for NO?
right but i dont know where to tell emc which pin to check
um, what's NC and NO mean?
i guess the pin numbers are hard coded
NO (Normally Open) are wired in parallel so any switch will complete the circuit, NC (Normally Closed) are wired so any switch can break the circuit.
the machine already has all limit swtiches wired into pin 10, but emc wants max limits on 12 and min limits on 13
fenn: So grab the soldering iron?
right. it's just so much easier to sit here and pretend to be doing something useful
like check my email for instance
anonimasu you awake yet?
fenn: are you using emc2?
Jymmm: yes, been at work a couple of hours..
building a box with a some PLC's..
42 wires into the plc.. from damn connectors.
anonimasu, no just emc1.. bdi 4.20
i think maybe i was wrong about it being hardcoded
hm, I think they are
cant remember I wired my motors the way it said atleast to not have to mess with it..
there's a parameter in emc.ini called min_limit_index or something like that, it's set for each axis
hm, perhaps I should hook up my limits someday soon :)
bah who needs limits
worst case scenario your machine tears itself apart and knocks down the house
* anonimasu might have a cnc lathe soon
well, I have the steppers, and I have the drives..
but I need to build a amp for them to work with the paralell port..
anonimasu: Got any spares? :)
Phydbleep: japan servo brand new ones :)
expensive steppers :)
"japan servo" brand steppers?
but I need to order ballscrews forst..
* Phydbleep mugs anonimasu
I think they were pretty expensive.
cant remember.. :)
right now i'm in the mood that any cnc is better than no cnc
* anonimasu nods
I have a mill :)
wait.. i'm always in that mood
but I need a lathe
anything that'll cut to your standards.
Hmmm... $0.57, A condom and a "White Brains on Toast" tank-top.. :\
i don't get it phydbleep
hm, maybe i should take a break and write down all the pins i've connected so far
Evidently no one here is familiar with 'Firesign Theatre'.
Phydbleep None of us were born in the 1800's
Jymmm: I'll have to dig some out and let you give it a listen.. Odd, old radio show format comedy.
Good for listening to in the shop just so everybody is too busy trying to figure what it is to ask stupid questions. :)
Phydbleep "But, But, But, I dont have a victrola to listen to them on"
Phydbleep : couldn't resist =)
* Phydbleep beats Jymmm silly with 100 mp3's..
how old are they actually?
the broadcast I mean
Everything from the '70's up.
A-L-P-H-A: do you have any program for online programming your lathe?
A-L-P-H-A: or do you make all parts on a computer or hand code them..
anonimasu. I code by hand... well.. I actually draw the outline profile I want, and then code used the outline, and the command "offset" to make my passes.
* Phydbleep has vinyl for 75% of his collection. :)
* anonimasu wishes emc had some online programming stuff
it's be wonderful for turning quick stuff..
for non cnc people..
anonimasu, I guess I could use mastercam. but I'm like why?
what do you mean "online"?
anonimasu: That's what I wanted to try with the on the fly g-code editor idea.
point and click..
I'll hack somthing togther when I get there..
Phydbleep: do you know gtk/programming for X
Phydbleep, turbocnc has on the fly gcode editor.
as well as variable g-code programming.
Jymmm, seeking a career in criminal justice? :)
Phydbleep: ok, maybe we could hack somthing togther..
get paid to drink beers and watch ladies dance naked
and people are pissed off about the identity theft part.. weird
anonimasu: I was thinking a tie from a panel on the machine with the jogwheel and other buttons back to the pc..
* anonimasu too
like on real controls..
a bit atleast
i just want the number pad to do x/y/z jogging
hit EDIT and start manipulating the wheel to inc//decrement the value ADVANCE to hit the next field.
Phydbleep: hm, I'll start working on somthing.. tonight
* Phydbleep wants to do x/y/z jogging as well..
althouch non graphical at first, but for testing..
x & y are no trouble, But as soon as I get any altitude on the z it hurts to re-zero. :)
anonimasu: I was thinking totally non-gui.
Phydbleep: I was thinking part/toolpath visualisation..
use a 4x40 lcd at the machine to show the code/position/options.
so you can veryfy your profile..
I have a screen there anyway..
it sounds like you guys are thinking of two different things
phyd wants basically a DRO with handwheels
anon wants to be able to move the part code around in emc somehow?
fenn: generate code to throw at emc..
I'm thinking of a machine that without the pc/emc has dro.. When you add the pc it becomes a slave controller head.
Decide that the part needs to lose another 0.010 and not have to go to the main kb to re-enter code or re-draw.. Just select the right line in the g-code and tell it to lose 0.010.
problem is most gcode files are thousands of lines long these days
do you mean scale the part or translate the part?
translate I think.. remove 0.105 at one point instead of 0.100.. Everything els stays the same.
it's past my bedtime
well, you are close to what I want..
except I am going to use the screen to enter my values..
well, going to start writing on somthing tonight
Jymmm: very funny
anonimasu Ah (been a rough day/night)
although I need to crank out some parts first
but well, the idea isnt too difficult..
the only tricky part is to use HAL to get access to pins..
* anonimasu yawns
for not opening till 9am
A-L-P-H-A: So go postal on them @ 9:01. :)
hard to do over the phone
they just hang up.
I feel like coding
but I have to build this damn box.
physically build? or install software?
I found a nice little case for ipod shuffle, and ipod normal. very interesting to me.
I think that will be my next little project.
FedEx may be cheaper sometimes, and have fewer "oops" than UPS, but they've only been a PITA to me
make a case for ipod shuffle.
I tried to set up an online acct with them
UPS is bad. FedEX is better, but not good. I still prefer USPS Global Priority EXpress... it arrives in 3-5 days, and brokerage fees are cheap.
and they don't even always charge me brokerage fees and shit. they just drop it off at my door.
You know that little "charge an credit acct, to verify" thingy?
I had two packages to go UPS, and one FedEx
asdfqwega, nope. I have an account with FedEx, Purolator, UPS, and Canada Post.
I've called them all to setup an account, when I was selling gecko 201 rev1&2s.
A-L-P-H-A: It's something that paypal, ups, and I've had some others do
paypal? I still don't have a verified account with them.
it's fucked up.
They start a transaction for a small amount, like $1, to check you credit source - but they don't actaully withdraw the amount
nono. paypal charges you a small amount.
Well, FedEx did a test charge of $10
and on the transaction account summary, they ask you for the exact amount that they charged, so they know that it really is you.
and when you verify, you're good. That amount is dumped into your account.
asdfqwega, did you call customer support?
And since it was more than I had available in that acct...
Yes I did
and what did they do?
was this a bank account? or a credit card?
So I was stuck for a week, with no Fedex access, and no money for my UPS shipping
asdfqwega, learn to apply for real credit.
credit is useful... especially in the business world.
I have credit cards, thank you very much
asdfqwega, so did you max them out or something??
No, I was using a debit account at the time
asdfqwega, as well, there are places like Mail Boxes ETC, Kinkos... etc.
I was trying to use my Paypal debit, so I could get the lower fee rate
It was frustrating...I was broke at the time, and it was just the last thing I'd expect to happen
asdfqwega: What are you doing awake?
I am soon done
2 pressure sensors.. left to wire
and 2 proximity ones
Snazzy, I was just going to throw together some plunger types for the lathe.
Phydbleep: What? I slept...like 4 hours.
asdfqwega: Then quit slacking, check your email and get back to work? <jk> :)
Shell with a plunger that runs past a microswitch... There's a tapered section to actuate the switch when the plunger slides in.. Spring load them for a self reset or springless for a safety.
oh this isnt for a cnc
it's for some other stuff
does anyone know if axis can handle g02/g03 Z-axis helical movements? I want to simulate something.
shame. axis is stopped development.
of course it can
cradek_ is now known as cradek
it properly handles any move emc can make
I really gotta make the switch over to linux completely. or almost completely.
A-L-P-H-A: axis is emc with another frontend..
so, well, YES!
just making a lot of paypal "buy now " buttons today
your wood working?
les, read www.paypalsucks.com first.
please read that first.
naw helping out with a relative's site
well this is just the merchant shopping cart service
customers use plastic
looks like this:
paypal has buttons.
[13:42:51] <les> http://broderickcrawfordart.com/giclee6.html
so I have to just paste a lot more of those buttons
What tools do you people use for writing gcode?. I'm currently using mastercam (I hate it), tinkering around with visualmill, www.cncsimulator.com (free win32 simulator), metacut (sucks too, but does have one thing I like, material removal rate)
that painint looks nice.
I mostly use millwrite
I do not like mastercam
tried cncsimulator but it is fixed resolution
(and not enough)
This morning I also read a little about RTAI
les, the resolution is now variable.
as of 4.44f
let me double check
i'll grab a fresh copy
I saw something else.
radius tolerances is variable.
Oh and camsoft ($5500 plus galil board) is running .001 point spacing at 3000 ips
they did a spiral test program similar to ours
spiral test program?
ours starts at .05 point spacing
constant angle...so as it spirals in the points get closer together
so they stole it?
or what are you trying to say?
no just a coincedence
theirs is a helix of constant radius
ours is actually a helix too
They claim 50,000 points per second
kinda hard to believe
ours could do 50.
les: We didn't actually time the throughput.
put it on the scope. :)
just a rough estimate from what I saw
just before the stutter
paul let me email you this thing I saw....
les: that sounds neat..
If the RT code is running on a DSP, then 62uSec servo loops are possible. Using buffers in shared memory would be the most practical way of keeping a 3,000 point queue full.
hm, I wish emc would do that :)
paul_c: how fast can you keep the queue filled on a pc?
paul_c: without the dsp?
I am doing some reading on the RTAI so that I can understand a little better
where is the limit is it in segmentqueue?
I thought you could pre-calculate everything with segmentqueue if you would like to..
unfortunately it cannot be made to work.
is the algorithm usless or the current implemenation
The math is right
the rest I cannot say...if someone knew it could be fixed
well, isnt there anyone that understands how it works?
* anonimasu finds it strange
it miscalculates then locks
that's harder then..
Seemingly not...and I have even had the author here
why dosent the same happen at lowe throughput?
or is ir just invisible?
hm, I need to run away for a meeting..
will be back in 3
It screws up at low throughput too
anyway....3000ipm with points in the .0002"-.02" range is more than just about anyone needs!
But that is what's out there.
It is about $7000 total though.
kills some potential products...
kills like what?
Oh can't say too much...but we are considering marketing a high performance cnc router
les... it's a lovely product, and I would like it. I'm already getting 80ipm, at something like 0.000125" resolution. (4tpi ball screws, with a G201).
"we"? your company?
light metal capable
good profit in that.
This is only possible with using emc.
what's the pulse rate on emc?
say, a P2 standard chipset.
on the parallel port.
pentium 2, standard chipset.
don't know...we only do servo
I've got a AMD athalon 1300 runnin my box.
about 2k servo updates/sec seems to be the limit
That is fast enough for most things
...buttons buttons buttons....
I should have someone else doing this
pay a monkey to do it.
I mean co-op student.
I just go to paypal and put in item discription, price, etc....It makes HTML....I copy and paste it as a code object in sitespinner.
Wheres that chicken?
he could pack keys and do this.
is she hot?
if she is... you know the reason why.
She pecked Paul...so why not a keyboard?
Paul spoiled it
pecked congo the cat too
congo slapped it
les, isn't it a bit early to be smoking weed?
this is all true!
Not when you are trying to forget about the tooth problems! 8-)
Oh...that's all ok.
actually I just had a stiff drink and an aspirin last night...no vicoden
all better today.
time to test the code I wrote. :)
lets see if any endmills will explode.
brb I am going to walk to the store and get a DR pepper.
Gee, it didn't take long for you to become a Southerner.
paul - saw your comments on Sun. about CAM systems doing very short moves to do surfaces
just wanted to let you know that Synergy (optionally) converts point sequences to circles whenever possible
and Synergy is not $$$
that would be a problem unfortunately
because emc does exact stop only on adjacent arcs
we've done a few machine that had a nurb G code but there isn't a standard
right...and none of my cam stuff has that
but it would be nice
Don't know if that's the right way to go since you're trading cpu time to calc the nurb
well it does not have to be done in real time
knot points etc can be calculated and stored
or the plynomial coeff of the curve
I suppose the additional info would allow it to calc a better path
If done at max feed overide it will work (sub time optimal) at lower feed over rides
usual smooth profile is to either use a cubic to s-curve smooth a trapezoidal profile (emc does this)
or to spline up multiple points
So the former is done at the servo rate
the latter at the trajectory rate
Many point splining was attempted
But We just cannot get it to work.
sorry - on phone with cust
stevestallings2 is now known as steve_stallings
was just looking a bit at your site Bob
Site is very out of date
no endmill explosions.
but, turbocnc crashed on me.
exploding endmills are not good
int he middle of the part.
A-L-P-H-A: that's why I dont use tcnc
I've hard them crack on me before... but my system has LOTS of backlash
I've gotta sit down, and install emc properly.
what's the newest BDI?
But it sounds much more exciting the way you have it now
I was in the middle of a g03 statement, and I tried to override the feedrate, and it crashed.
in the middle of a part.
that's shibby! very shibby.
websys: I still havent gotten around to playing with synergy yet :/
Bob you are near Fon Du Lac?
Live about 20 miles from there
work about 45
I used to live in Chicago.
flew to Fon Du Lac and Oshkosh often on weekends
In-laws live near Pallatine
I worked in Glenview
yeah going to the museum
Ever been to the fly-in?
yes many times...I am a pilot
Some of those things don't look like they could get in the air
And I certainly wouldn't want to be in them
Gyro-copters = go-karts with a propeller
I just flew plain old certified light aircraft
We would always be drooling at the new ships
Yeah but you didn't have to worry about falling out of the sky
Salesmen at the show would try to talk us into 30 year $800 loans
that's a grim dela
airplanes are expensive toys.
but "that" expensive..
Well I have most time in a Piper Archer
I need to make a spring loaded marker... put it in the mill, and let it draw out the tool path for me. heh.
It is about $240k
* anonimasu nods
well, the thought of paying off for more years then I've lived for is pretty scary..
it's almost forever :)
anonimasu, that is forever. :)
well perhaps I should go home from work..
Well I can't have those toys...unless business gets awfully good!
to make parts for work..
10% done with BDI-4.20
~100kpbs only. :/
I have a 17 hour workweek this week..
work-do stuff for work-code more stuff for work ;)
anonimasu? 17 hours... that's what? 4.25 hours a day?
We aren't in production right now
3.4hrs a day.
but will start for next year's products soon
at saturday and sunday it's time to finish the code, that'll run on the series of machines..
A-L-P-H-A: more like 17 hours a day
So I am just doing odd job stuff
anonimasu... you said 17hr work week.
ah, meant days..
do I smell overtime?
web page, some machine maintenance, etc
so complain not my young friend.
selling engineering consulting time
actually I don't even know if I'm older than you.
A-L-P-H-A: I work for my father's company...
A-L-P-H-A: that kind of overtime
I am 21 soon..
in 4 days
I'm 26 in 33 days.
I'm wondering if I should buy a 1/2" or so endmill, carbide indexable.
1/2", or 5/8" carbide indexable endmill.
and use the fine stuff with a normal carbide endmill.
I just use the spiral flute ones now that I can sharpen them
sharpen them true?
just got a call from camsoft.
kind of fast talking slick operators.
I do not like closed source stuff.
Even if it does do < .02 segment spacing at 3000ips
I'm not even sure that math works out...
that is 2500+ segments/sec
and servo update is.... 16 khz
7 servo cycles/segment
yeah I guess that works out.
les: what problem do you have with your CAM software ???
I have no cam software problems really
This is machine controller stuf
like i said, if you need testing programms with good quality, let me know
have maybe the best CAM software
I note that Galil's best card actually only does 250 microseconds with 3 axes
250 is not so fast
also not that much, or
So camsoft must be using something else
one moment, have a look in the Haidenhain (--> best controller for free surface milling) OEM katalog
250 might be in the range that emc could do
There 62 microsecond claims must be with some other card
forgot to ask
Interpolater 3ms, subinterpolater 0,1ms, velosity controller 0.1ms and curent controller 0,1ms
hm subinterpolater,velosity and current controller have all the same speed
and it can take a g-code sentence every 0,5ms
ahh, subinterpolater velosity controller runs at 0,2ms. 0,1ms are only possible without a direct measurement system
hmm and that's their best?
the possition controller (is that the right translation) runs also every 0,2ms
I think Camsoft might be written for this:
[18:23:44] <les> http://www.aerotech.com/products/controllers/u500specs.html
hm, plunging into aluminium is scary.
or well helical entry also, with too little spindle power..
How you doing?
making some parts
if you ask the manufactures of machine tools, then you get the answer that the Haidenhain TNC 530 makes at teh moment the best surfaces, better then the Siemens 840D
wot they don't tell you EMC makes the best!
waiting for them to finish so I can start hacking on some online programming tool..
about your last mail to the list: EMC2 runs fine with steppers and servos at the moment !!! Ok exept the problems that less trys to solve
Oh camsoft knows of emc
It does. That is great.
hm and the trouble that it dosent change mode MDI > AUTO > manual > AUTO
Does it give any error message with the change?
I've ran it with -d -v
it gives the same messages as when it actually switches
Next time look at the status of the interp. On tkemc it's down above the auto display on the right.
The choices there are running, paused, idle.
below the feed override?
Idle should let you change modes from auto to whatever.
let me look.
right above the step and verify buttons below mdi input widget.
I'll have a look at it in a bit..
I'll try to provoke it to show the error..
after this part need to let the stepper cool down a bit..
You can do it.
Just got a call and note from Joe Martin at Sherline...
The CNC mill won an award at a tool show in Belgrade.
I'll copy a few of the pics into the dropbox.
I have read a lot of his entrepreneur stuff
the error happened again..
it appears after loading a new .cnc file..
in the debug window it says the following:
task status: 2
lo status: 1
motion status: 2
That doesn't seem very helpful at all. At least the hangup isn't with the interp.
yep no error or anythnig at all
this is with a pretty fresh cvs..
Issuing EMC_TASK_SET_MODE -- (+504,+16, +119, +3,)
usrmotWriteEmcmotCommand() success (0)
emcTaskIssueCommand() returning: 0
Issuing EMC_TASK_SET_MODE -- (+504,+16, +119, +3,)
usrmotWriteEmcmotCommand() success (0)
emcTaskIssueCommand() returning: 0
err sorry, read that once..
that's what happens when I change mode..
Ray have you seen anyone run servo emc on a pretty fast box...like 2 GHz?
brb, going to the kiosk to buy somthing to drink
Since the real numbers for these multi thousand dollar dsp cards are in the 250 microsecond range
We pooped emc at somewhat under 400 microseconds
not that much difference
the quoted 62 microseconds was per axis
les: No I've not watched a fast pc run servos.
fastes is 800 via.
Sorry I didn't get back last evening. Phone never stopped ringing.
I just wonder where that brick wall is....
In a pc?
if it is at 250...that is as good as most anything out there with dsps
rayh : You finally got that 1-900-HAWT-SEX number installed huh?
yeah a pc daryl
As usualy, I'm jumping into th emiddle of a conversation... what are the time's you're looking at?
Wow... lots of typos in that one.
we did 400 microseconds update in a 500 mHz box
Darn. I got all those semaphores stuck after my last emc run on this box.
Slowing down to a crawl.
slightly better as I recall in an old K6-200
servo cycles yes
If the $$$ stuff actually does 250....
And in each cycle is there just a single PID iteration?
that might be reachable
ie: 3 multiplications and addtions and one subtraction?
Hmm in emc there is a bunch of stuff in the cycle
a whole bunch
But is it needed?
heh...I don't know
PIDFF is just very short of course
Jymmm: Yep that's taking all the time these days.<g>
* rayh is trying to ftp without gftp.
Still thinking of stuff for your 500inch/min thing?
If I can't do that I don't have a product
What do you need for that speed?
les: How would I test cycle time with this box. I don't have a servo setup right now?
money it seems
you mean what type of work needs that speed?
I meant what kind of cycle time you need.
ray: I don't know if simulation is valid, but it's worth a try
daryl: faster is better. 250 microseconds would do.
I'm up for trying as soon as the ftp is complete.
I think emc might be as fast as the dsp cards after all
I know an old guy who says not. He claims to be the first PC motion user.
He's still looking at dsp chips.
I think Camsoft made that claim too
I used the old tech80 cards in the mid 90's
okay. pic's are up www.linuxcnc.org/Dropbox bel#.jpg where # is 1-7
1 khz servo update per axis
the AB 7360 used a 100 update per second.
I can try a sim here and see what I get.
great what speed box?
Are you hooking things up to the parport?
That's a fairly significant bottleneck in a pc.
This is a 1.8 athalon showing 3588 bogomips
well the simulation is not real time...
but the step stuff runs a virtual servo
I could put a motenc in
so that with nothing hooked up would be the way to go I think
freqmod has a lot o rt overhead but it it would at least show worst case.
ray: I think it would not run without encoder data with the motenc
What ini numbers do you want for time?
in the axes section of course
perhaps .001 in trajectory?
going for broke...
Just tried those number on my box with freqmod, and it runs.
We really did not have a lockup below 400...just a gradual going nuts
what box daryl
Athlong xp 2GHz
speed? cpu or motors?
How fast is your box daryl
could you guys try it till it goes?
les: try it till it goes? You mean reduce times until it stops running, or run it with those times until it crashes?
Man, rayh must be racking in a lot of money on the 900 number of his... it never stops ringing!
reduce times till it goofs up (it usually does not crash)
Lotsa folk want me to talk dirty to em.
What kind of period makes sense for this?
um...down to .0001 axis.... .0004 trajectory?
half the time
The image �http://www.linuxcnc.org/Dropbox/bel1.jpg�
cannot be displayed, because it contains errors.
Do you need to keep PERIOD= set proportionally to the smallest cycle time?
let me check
Mine's at 0.000024
started right up here.
? that is more than the latency jitter paul expected
Seems to be running with period=0.000024, tcyc=.0004, axiscycs=0.0001
ok, I'm off to the surplus stores! cya laters
Is frqmd or whatever bypassing some stuff I wonder?
I'm running spiral with no obvious issues.
does it do a servo cycle reading virtual encoders, writing to virtual cards etc...
can you also send me that spiral you youse for testing !!
I've got a pen in the chuck and "milling" a pcb outline pattern at full speed... seems ok.
Still on top of the pattern I've drawn about 10 times now.
(not 10 times with these settings yet... machine's not that fast)
I got a bad character used error.
that's something we saw...could not find it
Hmm... well that killed it.
Reduced period down to 0.000010
It is the % at the end without % at the begining.
No worries on that one.
Ran fine here. want me to clock it and speed up feedrate?
Any idea what period setting you need?
Or if it matters?
Not sure what that means for a servo test
it is some overhead i'm sure
My impression is that a real servo hookup is doing a lot more communication
It does affect freqmod which is called by generic.run and generic.ini
I'm running freqmod
If pauls box doesn't die at about .0003 with freqmod that would tell me some
that is where it died with STG
I haven't looked into this is much detail in recent times, but I think parport reads/writes take 1-2us.
I think talking to the STG might be taking a lot of time
each encoder read requires three inb() calls
Yeah... so that's definitely a brick wall.
3 DAC writes and and 3 encoder reads
and a DAC write, one outw()
how long do you figure Paul?
display is getting fairly slow
feedrate 200 period 11
questimate ~15uSec for the read/writes
ok so that's 90
would need to take a closer look at the code to verify it though.
how fast will your 500 box run freqmod?
pretty sure encoder reads are done two axis at a time.
$2.08CDN for a stainless steel external retaining ring. And I need to buy like 10 of them. Where as I only need 1.
so 75 then
you need a well stocked hardware casbinet...that's how I justify buying more than I need
Well it ran through with 200 feedrate.
was it really 200 ipm?
My only local clock isn't gonna work very well.
or look like it?
I'll give a start in 10 and then post when it stops.
so something with full servo is really slowing things
a lot more than DAC and encoder read/writes
start in 10 sec
hmm what is the path length....
a little calculus?
OK... Taking the spiral test and splining 7540 point together, and then taking 50 way points produces a reasonable representation.
a 50 point cubic spline would look pretty good
even a quadratic would look good (since a spiral is almost second order)
$12.80 for 100 in black alloy, versus the $20.80 for 10 (TEN) 18-8 SS. Stupid retaining rings.
7540 points splined and 50 way points generated in a lot less than 1 second.
some cad system?
you wrote it?
just need to go the other way and see what 50 splined points look like when broken in to 7540 way points.
It could also look pretty good
there are lisp functions for autocad that turn splines to plines. and 3dplines.
I'm guessing that if arcs are passed down to the TP as splines (given sufficient knot points), they should be reproduced without any significant error.
right...because a second order arc is a subset of a third order spline
It's not quite second order... but close
selected portions of parabolas would wrap around it very nicely
circles are special case ellipses and wrong side out hyperpolas
conversely replacing general curves with joined arcs results in an infinite jerk profile
even if tangently joined
* anonimasu hides
Using 18 control points to generate the spiral....
and interpolating 200 way points...
The spiral looks reasonably smooth, but....
i think splines are a bit dangourous, because they can easyly begin to swing
how does it come out if you machine it?
hm, well time to draw up some stuff, before I forget it.
"You're just jelous because the voices are only talking to me."
hm, it depends how havy it swings
i think because of this it is not that easy to use splines for a internal data format
don't know it exactly, but i think Siemens does something like this, and they have problems with it. The machine moves somethimes out of the given tolerance
you can specify the tollerance in the g-code file
If you have a better idea that is computationally efficient, feel free to code it up.
Well the information content in a helix or spiral is very low
I just drew a spline over a circle....it's damn near perfect. 20 points
what's the secret to get nice looking machine tooling marks after milling?
Imperator__ is now known as Imperator_
climb milling alpha
if you have slop climb milling will kill endmills..
I've tried with & without
oh...yeah. I don't have slop.
if you have slop the cutter will pull/push the table along... and make horrid marks..
anonimasu: Drawfile, polish, then use a hard mop & grit to put back engine turning marks.
dont you just get machining marks after you've milled your part?
wouldnt filing kill tolerance?
good filing improves tolerance haha
just do it in the right place...
does all machineshops file their parts?
I deburr with one a lot
Any recommendations for generating text for milling?
well millwrite is wonderful
Free and runs in linux?
even for genraral cnc cam
fre demo...does not run in linux
Actually... I suppose I'd consider not free. Just not M$ os.
gosh linux cam...that narrows it a bit
*points at webersystems*
Running under linux, you have Synergy, Qcam, and a couple of others.
les, not the edges, but the surfacing.
les, mine's a manual mill converted... the spring holds the quill up.
websys in on right now here
daryl: my truetype g-code renderer
that is synergy
oh I forgot chris' program
damn that's a cute cat.
hp2xxx does g-code from what I hear..
he thinks so too
I have a perl script which converts postcript to g-code... only does outlines of tt fonts. No fill.
cradek: Does your fill?
with autocad, sure it does
(read down further)
daryl, cause TTF are vector graphics. So outlines are the only way they come.
fill = scanline conversion = trivial to do with freetype
or vectors with fills.
sure, depends what you want.
A-L-P-H-A: Yes.. but some software which is aware of mill bit size could generate fills
I definitely need to make some of those CNC work holders. Les, got any tips on making them? [or trade secret]
my program, which extracts the outlines (*correctly*) would be a good starting point for any more complex software.
The ones Chris has and mine are similar. They are very handy. I make mine out of stressproof prehard steel.
les, these things http://timeguy.com/cradek-files/cnc/tool-holders.jpg
les, so that's 'tool steel'?
if you make your own on the lathe it's best to do it all in one setup cantelevering the work
'cantelevering'? I was gonna do it like how I did the 1/8" tool holder.
when making them on the lathe. In other words I don't do the final outside cut between centers.
very important that the hole and outer surfaces are concentric
Axis looks pretty cool.
I was gonna take maybe a 5/8" or 3/4" dia rod, face, center drill, drill, ream, insert Drill Rod, chuck drill rod into tailstock, flick down to size to have a specific collar. Lock in place, and drill/tap set screw.
I just cantelever it.
I don't know what cantelever means.
part not held by tailstock
only for light cuts of course
the finish cuts
[20:54:55] <A-L-P-H-A> http://dictionary.reference.com/search?r=2&q=cantilever
that doesn't make sense.
I hog most of em out with collet and live center though
that's the definition, and what you're saying to me, that doesn't make sense.
if I held a dead centre in the tailstock, I'd make sure that the part is concentric.
that dictionary is not a real good engineering description it's true
with a dead center one end will be true
but near the chuck end?
With a cantelevered part you don't need the drill rod bit
I still don't know what cantelevered means, in terms of how to setup the part.
ok I use perhaps 0.75" ground rod in a 5 c collet.....
I center drill it...
A-L-P-H-A: Cantilevered means unsupported at the end opposite the chuck, No center/tailstock.
cradek: To get axis to run on bdi-4.20, am I correct in understanding that I need to download emc source from CVS and recompile it so that I can then build axis?
mount in live center in tailstock...
and hog most of the shape.
Then back the tailstock off, drill the hole, bore it to size, and take very light cuts to finish the outside
I get concentricity in tenths.
then part it off.
did the fast PC tests say anything to you les?
* rayh is all ears
les, so, what's the difference between doing it in reverse?
Reads, write, and other things in servo mode are taking a lot of time
how do you mount the endmills in thoose
bore it, ream it, dead center, hog it to shape, light finish cuts, part?
anonimasu, weldon shanks.
That will work...I just don't need the dowel bit.
I have no idea what that is
A-L-P-H-A: Weldond shanks?
What makes you think servo mode takes longer with read and write
well, I'm only gonna have like four sizes. 1/8, 1/4, 3/8, and 1/2.
anonimasu, the flats you see on some endmills.
A-L-P-H-A: : ah, setscrews..
Ray: because freqmod ran so fast....and servo did not
hm, that arrangement woild be somthing to build a toolchanger on..
but that is only a guess
and ditch the morse taper..
A full card plugged in and hooked up servo test would be needed to know
and and make a bearing as retainer.. to keep the / \ in the machine..
But that is what you already have there, kLes
A-L-P-H-A: what do you think about that?
I just don't have the fast machine
anonimasu, I'm just thinking of how normal CAT systems work. They have a draw bar thingie.
A-L-P-H-A: I dont have a pneumatic drawbar..
anonimasu? make/buy one?
no space to fit one..
I have the space, at the top of my pully/quill area.
anonimasu: Use a through-bar and put it to the left of the spindle.
but that's it.
still the trouble is that the taper sticks..
For a lathe or on top for a mill.
paul_c: Do you think that newstgmod takes longer to process than freqmod?
so I have to hit it out there..
anonimasu: Use a tapered sleeve? Outside is MT2 (MT3?) inside is cut for your collets.
Phydbleep: I still want somthing to retain the cutters in the mill..
I might be able to fit a machine taper, but I am unsure..
cradek: Never mind.. I see it's already built. So far looks way cool!
it depends if the last taper is the actual bearing contact surface..
or is it's just a reduction..
if it is I might be able to fit another drawbar..
toolchanger carousel *drools* ;)
are thoose good?
anonimasu: OR http://www.harborfreight.com/cpi/ctaf/Displayitem.taf?itemnumber=45047
what kind of precision do they keep?
well, still the drarbar is the trouble :9
Depends on how good a machinist you are.. I've seen guys hold 0.001 for 22" on a lathe with 0.15 wear in the bed.
as soon as my parts are done the machine is going down in molecules..
rayh: Most certainly, any stg*mod takes longer to run than freqmod (subject to context switching times for the pulse rate generator)
the gearbox is going away, and a 2hp 2900rpm spindle is going in..
maybe new bearings..
anonimasu: Good for you.. Get rid of those gears and all the slack in them. :)
Phydbleep: I dont like being scared of aluminium ;)
Okay. What do you mean by context switching times for the prg.
anonimasu: Of course a trip through the gearbox with the mig would also cure all that movement. :)
rt to user
I am going to replace it with a 1:2 reduction..
and run the motor 1:1 on top of the mill..
or the other way around
anonimasu, I've seen 3rd party tool changes, but way expensive.
I dont have trouble with anything else then the drawbar..
it's the only mess..
I could build the rest, with some pneumatics and a plc..
rayh: the time it takes between the shedualed time and the time the RT thread actually runs - Think of it as the latency & jitter times.
but well, without the drawbar, I cant even change tools manually without violence
Okay but doesn't that also apply equally to something like freqmod.
A-L-P-H-A: are there short holders like thoose cradek had to buy?
sure it does.
that can be adapted with another kind of retainer then a drawbar..
Doesn't freqmod increment and write an encoder position the same as stg?
anonimasu, I think it was meant for that kind of holder.
yes - But the encoder count is always in memory, so there is no IO read/writes to slow it down.
and freqmod only does two reads & writes (for more than four axis) per loop.
A-L-P-H-A: dosent matter too much in reality how I keep it in place, as long as the taper's right and shorter so I get less contact with the top of the taper..
And newstg reads?
STG: for each pair of encoders, three IO reads are required.
hm so stg is the bottleneck..
in that case..
[21:25:38] <A-L-P-H-A> http://www1.mscdirect.com/CGI/NNSRIT?PARTPG=NNLMK3&PMPXNO=2574862
<--- DAMN IT!!!!!!!!!!!
offset against the additional IO read/writes, stg*mod does not get pre-empted by the pulse generator thread.
A-L-P-H-A: why damn?
um sorry caps.
I've spent like 1/2 that amount already... and it's STILL not complete.
by the time I get the motor, and everything, argh.
Are the additional IO for limits and such and additional axis for *?
hm, you need pretty much air I belive
hm, I wonder if you could hog alu with that one..
or if it's too tiny..
there's a 0.88hp one, but thats 15hundred
rayh: Each home & limit switch read costs an inb()
well the 0.45 looks more reasonable..
but well expensive..
paul_c: Is the bottleneck the code execution speed or the port speed?
but it might be too little hp..
wow. 8CFM only.???
A-L-P-H-A: Well... If you talk as much as you type, 8CFM should be easy.. <JK> :)
well... I think my compressor should be able to do that without that much trouble.
but urk. the 525js is 21.5CFM.
that's lots more
Without something like HAL scope we have no way of measuring the times that these take?
Hehehe.. And here I was having evil thoughts about the piece of iron pipe in the stock pile and the fact that my compressor needs a full rebuild.
rayh: HALscope will affect the timings.
would Halscope perturb?
* Phydbleep tries the question again..
paul_c: Is the bottleneck the code execution speed or the port speed?
I don't think it is code execution
Phydbleep: Depends which RT module you are talking about.
Perhaps I ought to grab one of those fast ISA motherboards I meant to get
many things can affect the max speed you can run an RT thread at.
paul_c: You were talking about limit switches needing an 'inb()' like there was a real speed hit from reading them.
IO read/writes are only part of the problem.
les: Don't get an all intel
Think AMD might have a slight edge due to cache structure and pipelining?
I see two inb and one outb equivalent in the stg bridgeport extra.
Phydbleep: Any IO read/write over the ISA bus takes (typically) 2uSec.
Would freqmod with bridgeportio be about equivalent.
les: I have a K6-2/450(?) board and cpu here.. 2 ISA sockects.. I'll let it go real cheap. :)
paul_c: Ah, That's why I was going to use a PCI dual lpt for a buffer card. :)
I was thinking of one of those cheap 2 GHz isa boards
les: That's why I said real cheap. :)
I would want some pci too
I do not plan to buy any more stg
Phydbleep: Some of the PCI bridge chips have a long latency.
1 agp, 5 pci /2 isa /1 shared.
Thanks..but need faster
see pauls 500 went 400 microseconds
I need to see if it is possible to do 250
paul_c: Siig chipset on the lpt card.. 1mHz + out bioth port simultaneously.
PLX chip - PCI one side, ISA the other...
Not all PCI bridge chips are the same.
The Siig may be faster than an ordinary parport.
paul_c: Lots faster. :)
stick a scope on it and see how fast you can bit-bang an IO line.
My other REALLY RUDE thought was SCSI-II. :)
paul_c: I'll have to wait for one to get here.
how fast would a motenc be?
I wonder too
love local mom/pop shops. "Does CASH make a difference?" "umm... no really... lets see... $26... umm... I can make it $21." "done".
the vital card..
A-L-P-H-A: heh :)
The motenc has the advantage of a 32 bit data bus.
Ray has one there and a fast box
so you are four times faster already.
you could transfer more data faster couldnt you..
but just plugged in with no encoder signals etc would not run I think
That was why I was thinking dual lpt.. 16 bit data path (low-byte/high-byte) and double the input lines.
It might "idle"
On a PCI card, if the IO chip is slow, it will cause the system to wait should it request a read or write before the IO chip has finished.
but didn't we not see speed related errors until something was done?
paul_c: wouldnt it be possible to contact the makers of motenc and ask how it is?
I don't remember
if it does idle or what happens..
and what the max throughput rate is..
the max-real throughput..
Dac writes and encoder reads happen all the time right? even in estop?
yeah most likely
the card reads them to the buffer anyway..
But still....might take some actions to show too short servo time errors
if you could grab all input channels in one 32 bit read.. from the card you would have super short cycletimes..
I dont know of the typical speed of the PCI bus:es on modern systems..
It would take two I think
encoders 16 bit?
then to 32 counters in software?
let me see the specs
[21:55:42] <anonimasu> http://vitalsystem.com/motion/mtref.pdf
anonimasu: 33 mHz as a minimun.. 66 mHz for PCI-2.0
maybe also HAL helps, because the drivers for Hal, are doing the whole IO in one cycle.
les: have a look
it seems like the encoders are 32 bit
yeah that is what I see
so reading 3 axis:es will take 3 cycles..
or well 3 reads..
I have a PCI-Proto board from Kolter. They have a very interesting test of some mainboards
correct me if I am wrong I havent been dealing/caring about stuff like this
since I bought a laptop
they have mesuered the pci troughput and it was very different
[22:01:10] <Imperator_> http://www.kolter.de/pciboards.html
in german sorry
the question is how much throughput do you need in todays cnc machines..
and how much can motenc provide..
I know what I need
hm, is that the limit of motenc..
I could live with that
That is what I need to know...the emc speed limit
how fast can that push along a maching with full control?
if we assume optimal trajectory planning
Well as I studied the dsp boards and got by the hype....few were better than 250
now ... today i have lasered *interesting* things
endplates for a BIG electric motor
did you mean to?
these are the thick laminations (3mm) on a BIG stack of thin ones
for some huge locomotives in .ca
whatvever, they looked damn cool :)
Robin I got the call and sales hype from Camsoft today
Bad sales technique
Talked nasty about emc and stg heh
I just had a look at heidenhain's controllers
les: hehee :)
oh the software price suddenly went down up to 75% though
did they know you knew emc?
Cycle time of position controller: 1.8 ms
Cycle time of speed controller: 600 �s
Cycle time of current controller: minimum 100 �s
but well that stuff is $
He said STG was Don in a bedroom.
selling off the last batch of cards for replacement.
He was not too good all talk and no listen
the best way to deal with that sort ...
is to let them go on and on
oh, I had some salesmen at work, that came..
and then ...
and wanted to sell stuff..
they asked and asked about what I was doing..
ask them some stunningly in depth question
I said "I write code"
then they kept asking..
err the girl with the..
"are you doing acccounting?"
it ended with me walking away into the server room..
for some peace :)
His product was prob better than his salemanship
but I got the bottom line
A look ahead splined closed source 3000 ipm system is about...
$us3500 or so
oh, thats not bad
well, that's pretty cheap
or rather very cheap..
thats controllers and no drives right?
yeah...kills my product concept though
I could buy it but I like emc.
Well if you want to make a very good $10,000 machine...
and well, I belive emc can match that someday..
les: and step-servos are out too?
even $3500 does not work
no not out
but not preferred
better spend $ on the stuff that moves
and .. are you certain of the market?
robin_sz: g100 ?
I mean, does the market want a "very good" 10K machine or will a "good" 10K machine be fine too?
No I am not certain...that's why I won't do this unless I have something DRAMATICALLY better than what is out there
well, thats going to be tough
as I see it
tjhe biggest user group is signmaking right?
a stepper driven router gives them the accuracy they need
and the speed they need
you are trying to sell them something that gives them more than they can use ...
robin_sz: at a lower price.
I think in terms of show demonstrations...
what draws the crowds?
someone can do a les-type machine, with steppers for less than les can do it with servos
les: Girls in bikini's?
heh I mean besides that
very high speed
Phydbleep: no, girls *out of* bikinis
robin_sz: And in beer! :)
anonimasu: that draws the crowds, they watch, they gasp, they wander off and sopend half the money on soemthing perfectly adequate
at the big shows...shopbot always has the most bodies at their booth
reduction in cycletimes = more money
les: and that aitn on speed, thats on price
I could beat a shopbot with a hand chisel
anonimasu: you only cut acrylic at up to certian feeds and speeds, you can acheive those with steppers, and them some more too.
les: How many of those 'bodies' are housing brains?
mmmm at a professional only show...I don't know
heres the thing
singmakers wont give a flying fsck whetehr its servos, steppers or steam powered
does it cut acrylic
les: How many times have you seen some moron spend big $ just so he can brag about how much he spent?
and how much does it cost
Phydbleep: yes, but not in business. people want value for money, and *someone* is going to give it to them
in the UK, thats Pacer ...
robin_sz: How much does a pacer machine cost?
I was thinking stepper price...servo performance
that gets you ahead of the pack
whether they need it or not
anonimasu: the previous model was 12-14-16k .. depepndign on spec
not sure on this revised model, not had chance to check ...
been busy trying to run a business :)
les: my view is that the "luxury" machine you suggest will have the legs pulled out from under it by cheap imports. not as good, but 25% of the price. a good and cheap machine may hold up agaisnt that, a luxury machine .. I dont know,
if it was a real "rolls royce" of a machine
you may still ahve a niche
my motor end plates should be going onto a cmm about now
by tomorow I'll klnow if we have the contract
I have to remember the machine duty cycle doing signs vs turkey calls
robin_sz: are they within specs?
<robin_sz> my motor end plates should be going onto a cmm about now
I didnt ask about their measurements..
I was asking if you belived they would..
does your stuff have enough accuracy for it?
but the spec is *tight*
like +- 0.1
I cant measure that over 800mm
so I wont know till it comes out the cmm
yeah then it's tight..
it is ...
for cut stuff it is.. :)
0.1mm over 800mm is tight by any standards
we do however have a trick :)
we know which 4 holes out of 100 or so they use as the references
so we cut those first, before the plate warms up ...
hm just 8dm..
well, if it was on a mill that's be large tolerances..
yeah, 4 thou
I just can't get out of my head the idea that servos can be cheaper than steppers
50 pole brushless motors are expensive if big
encoders are free
it's the software.
if oem treadmill type motors cog...so what?
zero order FF
it's just code.
is servo a buzzword?
If at a show folks were at the shopbot booth seeing something that they thought met their needs...
and went to mine and saw someting twice as fast...for the same price...
would that sway them? Whether they needed it or not?
same orice? yes
but .. same price, same speed and lookign prettyier than shopbot would sway them too
they wont care a fig if its servos or steppers, most of them wont know the difference anyway
Extreme speed I know is not a big issue as far as business costs
but would it be an emotional issue?
in the sign business
you're limited to 25m/minute anyway I presume
hm, in milling it seems like rapids adds up to pretty much money..
well prettier then shotbot, isnt hard ;)
true but sign machines are idle 90% of the time
in eu, below 25m/min you can get away without light guards
yeah, then it wouldnt matter..
above 25m you needs gaurds or full enclosure
thats how you can stand alongside my plasmas when they are running
no one goes over 25m for that reason
25 m would knock the socks off of sign guys
if the price was right
easy to do with steppers
Sometimes I wonder if I should talk to shopbot
but not yet
I do not have a product
someone give me a shout when emc is fixed :)
I should forget this and become a pimp.
les: Become a bicyclist.. Then you can pedal your own ass all over town. :)
too bad cycling is boring ;)
it's strenuous here
Phydbleep: Do you have any idea how long it takes to travel the length of Tiger ?
all the slopes
haha not long
paul_c: I'd imagine it takes a while if you're tip-toeing and trying not to wake it.
tiger is 0.4 miles radius
pop 304 I think
no 303 my great grandma died 3 weeks ago
half great grandma
was younger than my grandma!
Paul met my great aunt
her dauter in law
les: Your great looking aunt? :)
still with a full set of teeth by the looks of it.
18 years yuonger than me
I found a movie of a toolchange on a real cnc machine..
yeah...especially since I ...well...kinda lost another tooth last night
robin_sz: BTW. NML is here to stay. The decision was made at the codeFest.
les: Bummer :(
well I documented the amateur dentistry last night right here!
root canal shattered
les: At least a car didn't shatter 70% of them in one go.
pulled out the shards
no problem since the fractures extended only halfway down to the tip of the root
but not repairable
you know, the bond of gum tissue to tooth enamel is pretty strong
you guys should come up north once a year for your dental work.
surely it'd be worth the expense.
* anonimasu nods
I removed a tooth a while back..
pretty soon I am not going to need any more dental work Chris
the dentist had to throw 2 bits away.. because they went dull..
do you have diamond teeth?
les: what's so different? We had flouridated water here I guess
hard bones.. I gues..
me too. I have lived in big cities most of my life
I have two very small fillings at 31
they were done at the same time - facing teeth
a good teeth person
I went all through college with no dental work
As a kid I had several fillings every 6 months
man the first tune-up after that was a nightmare
huh. Must be genes.
or diet or environment
les: I had all mine till the car got me..
with seat belts and air bags I didn't think anyone lost teeth in car accidents anymore
I did have wisdom ones out at shands medical center in college
that explains it...haha
right, i must go.
but I guess I need to sleep
cradek: I was on a bicycle. :\
I tried to ride the ten speed home after general anesthesia...
I fell over.
I haven't been to the university of Florida in a bit
so how do I check for an exhaust leak in my car if I can't get the potato to stay in?
um...huh? you mean the potato valve leak test?
in the tail pipe - I am hoping I can hear if there is a leak at the manifold
darn car won't idle right and I'm too cheap to have the dealer screw around with it for hours
manifold leaks usually sound like a clicking noise
yeah it has that at first when I start it
clean idle control valve!!
could be valves too though, I can't tell.
IAC is clean
ICV is about 99% of all idle problems it seems
it maintains idle speed, but it misses
none, but it does show a few misfires
they sort of move around though - not on one particular cylinder
I checked for intake leaks as well as I could (with propane)
I keep a code reader around
me too (I built mine myself)
paid for itself already, it fits both our cars (fords)
misfire code then?
no, it has never coded
but, it does register some misfires (not high enough threshold to code)
no codes at all then?
it had an EGR code once but I fixed it (the flow sensor)
what car? I have one ford
I thought for sure that would fix the idle but it didn't
coil on plug?
no, coil pack at the end of long wires
I should go pull the codes again and if it's one cylinder, check the plug again
I had bad idle on the 4.6 v-8 from water in the plug wells
but I'm sure it will be fine
yep, I've heard they do that
seems like coil on plug is a nightmare
yeah there is a lin eabove it that drips condensate
I have not had problems
so, back to the potato plan I guess
Both my cars are coil on plug
can you think of anything else I've missed?
let me think
only rough at idle?
yeah, runs great otherwise
great mileage - I sometimes get 33 highway
damn sounds just like ICV
not like it's missing at speed
Ford ICV usually doesn't pull a code
Have an employee with a 2000 ranger...
you're talking about the solenoid that bypasses the throttle plate, right?
does thgat all the time
always gunked up ICV
pretty sure mine is fine
it holds exactly the right idle speed at all loads (fan, AC, lights)
I've set the idle at 750 and 850 and it is right on the money
coil pack is one thing I haven't replaced but I did test it with ohmmeter and everything is according to the book
6 coil packs in a box on the side right?
well there are only 3
it fires pairs together
and the misfires aren't always on the same pair
crankshaft position sensor?
had not thought of that...
naw...that would throw a code
see if it has a code for that
well it can't run at all without that
it has codes for EVERYTHING
unless it is glitching
I can't figure out how to read the CPS correction like I can read the fuel trim
I know it learns the imperfections in the CPS output
but I don't know how to tell if that learning is successful
What are you pulling engine history data?
yeah, fuel trim is a regular old mode 2 PID
remove the battery ground...leave a while...and restart?
nah, I've reset it lots of times
I think go back to regular old non-computer things: vacuum leak, intake leak, exhaust leak
as you can see I've worked on this for a while...
has to be a glitched sensor if it is random cylinders or what you mention
I'll check the CPS, make sure it's not loose or something stupid
it's right on the front I think
yeah that is the thing that comes to mind
it's a good idea
I have seen it so much
sometimes it does not throw a code
I'll pull the test (misfire logging) data again
on 3.8 and 4.6 v-6 it is two bolts right on top
are you saying you have seen a bad cps cause rough running?
I think on 4.6
I think it's behind the fan on the bottom
3.8 I am not sure
very easy to identify.
I never thought I would buy or learn about one of these computerized cars.
but I sure like the easy starting, good mileage, etc.
Well I make a few of the sensors
yeah they are good
engines last forever these days
they really are better
and mine's an ultra-low-emission-vehicle which makes me feel all warm and fuzzy
and I laugh at the SUV drivers when I fill up my tiny 13 gallon gas tank
I sure am glad I didn't get the GT v-8 like I almost did.
this thing goes so damn fast as it is
I am having some ouch factor filling the truck
but the bimmer gets excellent mileage
yeah, it doesn't bother me yet - ask again at $4/gal though
then I'll just use the motorcycle all the time, and still no problem
at $8 or $16, I'll use the bicycle, still no problem.
well except for the soccer moms shooting each other.
well it will go up.
sure it will
we will be paying per liter what we pay per gallon.
well time for dinner for me
I'll report back