So, while I am reinstalling, are rotary encoders worth installing? It seems that they should help mitigate missed steps, and I think ought to allow use as DROs if twirling handles by hand.
I have found 200ppr incremental encoders at around the $20 price-point, anyone tried them? http://uk.farnell.com/1161169/passives/product.us0?sku=avago-technologies-hedr-5420-es214
using encoders to fix missed steps is not really practical - if you miss a step, the worst possible thing to do is try stepping faster to make up for it
they could be used to _detect_ the missed step and trip out on following error
this topic has been discussed many times on the mailing list
it is possible to experiment with encoders + steppers, but that should be considered "advanced"
The system can't note the error and correct it's internal idea of where it is in space then?
the system can correct its idea of where it is, but what then?
Put in an extra step?
logical next step is to tell the motor "hurry up. you should be over there"
if the motor lost a step because it was begin asked to deliver too much torque for the speed it was running, asking it to go even faster will just make it worse
Not so much faster, as "hang on a bit, see if we get there"
you mean slow down the other axes to let the one that lost steps catch up?
Aye, pretty much
in theory you could do something like that, using feedhold or adaptive feed inputs
in practice, when a stepper loses it, it doesn't usually drop one step
I don't like open-loop, my day-job is engine controllers.
most of the time, an overloaded stepper stalls. just sits there and buzzes or howls untill the step rate gets low enough for it to get going again
often that doesn't happen till the command stops completely, or at least slows down a LOT
if you want closed loop, you want servos
I have been thinking about that. It seems that, in general, you will at least end up taking off too little metal rather than too much.
tripping on a following error (with encoder) certainly beats running the rest of the program blindly, with some unknown offset between where you are and where you should be
There seems no reason why servos can't be steppers, and vice-versa. It seems more of a naing convention than anything else.
not entirely true
driving a brush DC motor as a servo is pretty easy - torque is proportional to current, speed is proportional to voltage
brushless DC or AC servo motors are more complex - they need three phase sinusoidal drive, which needs to be synchronized to rotor position
Indeed, but when what you want is angular position..
steppers are like brushless DC, but with much higher pole counts, which makes the electronics that much more complex
With Sinusoidal Microstepping they effectively become three-phase or two-phase motors, with huge pole count. which seems very well suited to the sort of fractional-revolution duty-cycle we want
microstepping is still open loop though
im about to run the truck over thr acer..... how can i make it look like an accident
And whilst I accept that the electronics is more complex, you can just buy that on a chip.
the drive sets the winding currents such that you have a magnetic field in the motor at some angle, and you hope that the rotor aligns itself with that field
I think we may be vehemently agreeing here, whilst thinking we are arguing
the difference is that a servo drive _knows_ if the rotor is aligned with the field
and if its not, it can increase the field strength
in fact, AC servos normally do not have the field and the rotor aligned
the maximum torque is when field and rotor are 90 electrical degrees apart
What I am saying is that stepper+encoder running closed-loop is probably better than brushed-dc motor+encoder running closed loop.
AC servo drives keep the field at +/- 90 degrees and vary the strength, stepper drives keep the field strength constant and the angle varies with torque
for some value of "better", yes
Ah, I think you are hinting at the point I am missing, that size-for-size a DC motor can make a whole lot more torque.
<digress> My Y-axis leadscrew has a 6mm plain section. I can't find an angular contact bearing with that bore. I am planning to preload a pair of conventional deep-groove ball bearings. My thinking is that the actual internal geometry is not that much different. (in effect a DGB bearing is an AC bearing with an extra shoulder. Thoughts?
for light loads and moderate precision that should work
the geometry isn't quite the same - the contact angle on the angular contact bearing is controlled by the design, and is probably high enough to handle axial loads well
the contact angle for an axially loaded deep groove is unknown and likely to be small
I thought that too, but reading the skf website they seemed to say that the contact angle was actually higher (or at least the groove width is deeper)
depth of groove isn't the same as contact angle
However, the ball-count is a fair bit lower
Depending on the actual shape of the groove, and the preload, I am fairly sure that it _can_ be higher.
but you never know what it actually is
Whether it will last, time will tell.
the ball groove radius is always larger than the ball radius, so you don't get line contact and rubbing
so when you apply an axial load, the balls sort of roll up the sides of the grooves
Aye, I know.
if the groove radius is much larger than the ball radius, and there is very little free space in the grooves, they can't roll very far up
otoh, if the groove radius is nearly the same as the ball radius, and there is lots of space between balls and races, the balls can climb to a very high angle
Indeed, it was a cheap experiment. The backup plan is a needle roller thrust arrangement.
I wouldn't hesitate to use deep-groove bearings with an axial load of say 20% of the radial rating
maybe even a bit higher
but over 50% is probably asking for trouble
This is a hobby project, I am converting a cheap-and-nasty Chinese lathe/mill. One side-effect of this is that I am horribly constrained on Y-axis bearing size. 6mm bore and 18mm OD is all that can be fitted in.
that's not much of a bore, are you using rolling ball?
Actually, there is no reason I couldn't move the thrust assembly outboard to the twirly-knob (technical term) and leave the drive inboard.
Yes, this is an 8mm ballscrew.
oh i see
whats the ball screw's rated load?
No idea :-)
you using like a 618/6?
or do u know for sure what bearing you're using
you may not need thrust bearings is why i'm asking
(looks back at where dareposte logged in) No, I am trying my luck with a pair of preloaded deep-groove balls.
sorry did i come in and miss all the important facts again?
I have concluded that a pair of 6x16x5 bearings with a .1mm shim between them have not much backlash and still turn smoothly.
It's not engineering, I admit, and I do know better.
I wouldn't do it this way if someone was paying me, I would do it right and pass the cost on,
sounds right to me
i checked and a 3/8" ballscrew is only rated for about 135 lbs working, not too far from 8mm
It is a lathe Y-axis, that is probably enough.
i don't have any 6x16's in my bearing table, but for a pair of regular 626's they would be good for 45 million revolutions
The original leadscrew is 6mm, and not even trapezoidal.
Actually, probably 1/4" as it is a chinese Imperial lathe
i'd say you'll be in good shape
If not, I will mill out more space in the crosslide and try again.
deep groove's are underappreciated for their axial load handling, especially at lower speeds
you said you had pre-loaded ones already?
It's all that makes a motorbike go round a corner...
No, I took some standard ones and measured the axial float with a DTI.
mine has angular contacts in it
i'm pretty sure
but close enough right
And then assembled them with a matching feeler-gauge shim. And it felt about right.
pressed them into the shoulder?
Your bike or your lathe?
No, Yamaha use deep-groove balls.
expensive ones then, for sure
Except on the R1, where, at the back, they use a DGB on the brake side and a needle roller on the chain side.
No, totally standard 6203 and 6304
mine is an R1
sounds like you know your motorbikes!
any reason you wouldn't thread a bit of the rod and put a pre-load nut on the end?
Then make it very clear to your tyre fitter that the chain-side space is actually the needle-roller inner, and dropping it on the floor in the muck is not good.
i changed the last 2 sets, but didn't pay any special attention to that
just had the dealer break the bead and seat the new one on their hydraulic machine, seemed to work pretty well
Actually, the conventional motorcycle bearing arrangement is wrong.
You have two DGB bearings, both against shoulders on the outer and against the spader in the inners.
not familiar with DGB
So some degree of preload is unavoidable, unless the outer shoulders are narrower than than the inner spacer, which it might well be (I have never checked)
DGB == lazy Deep Groove Ball.
wouldn't a small bit of pre-load be good, both for motorcycle and cnc
In any case, a good reason to do the axle bolt up to the right torque. Squashing the spacer would be bad (and they tend to be aluminium)
yeah i was thinking you could "pre-load" them properly with the axle torque anyway
we always torque it down until the wheel spins just so when you flip it by hand, kind of by feel but it works real well
A small bit, yes, but you would normally preload by force, not distance. I think.
Don't use that method with the R1 rear, that has the "conventional" 1 axia and 1 radial bearing arrangement.
Which R1, by the way?
yeah just for the front wheel
yzf-r1, 2005 raven
I have the 2004 in grey, so very similar I think.
yeah great bike
I had on odd prblem with mine
(insert vowels to suit)
i've got about 18000 miles
what problem did you get?
First valve check at 24000 miles a few clearances had closed slightly (normal) and 2 had increased (odd). At 50,000 miles those same two were out of the shim range.
turned out that the stems of those two valves were soft, and I was half-way to the collet-grove.
that's not good
was it possible to seat new ones?
i've never messed with mine but i hear motorcycle valves can be a bit touchy
Yes, I fitted two new valves and the bike is better than ever (and quieter)
i've been hearing a "ticking" sound from the valves some recently
wonder if it might be related
Check the clearances, it's not such a big job.
i haven't done any valve checks yet
yeah i probably should
maybe this winter when its parked anyway
Well, actually, you might s well wait till the 24k service.
the weather is just cooling off enough to make for great riding weather again
"A tappy valve is a happy valve" after all
ie, too big is noisy, but does no harm.
Also, the valves are surprisingly cheap. <$20
never heard it, but clever
that's surprisingly cheap indeed
Of course, there are 20 of them.
thats a strange number
still a whole new set of valves for $400 isn't bad
5 cylinders? or 5 valves per cylinder?
mine's got 4 cylinders, so i guess that would be 5 valves per cylinder
probably three intakes and two exhausts
most of the early 90's toyotas have 20 valves i think
for the 4 cylinder
maybe not "most", but the sportier ones
Yamaha have had 3I + 2E since about 1987 on their bigger engines.
you are a wealth of yamaha knowledge
in 1987 i was still riding hot wheels
They reverted to 4-valves for the latest R1s because with modern FI systems you can get a bit more top end with 4. But they have sacrificed the tradtional R1 terrifying midrange
i know i am disappointed with the mid-range in mine
once it gets of 8k it goes pretty good, but until then its a bit poky
i had an '02 FZ-1 before it, and it was the opposite
Well, your 2005 and my 2004 are a bit weak midrange compared to the 1998 models.
from idle to about 8k it was uncontrollable, then peaked out after that
The solution is to ride in 1st gear everywhere. You can break every national speed limit in 1st gear, after all
(if anyone else is reading, the bikes in question do 105mph on the speedo in 1st gear)
i did notice that
i put a speedo-healer on mine and by all indications it does in fact break 100 in first gear
my speedo was off about 13% from factory though, pretty crappy
topped it out and it was reading 208 which was obviously incorrect
legal stuff, speedos are allowed to over-read, bur absolutley must not under-read.
but 13% is a lot
i work in auto manufacturing and the standard is 3.5%
I use a GPS if I care
thats how i calibrated my speedo healer
after calibration it tops out at 186ish, still pulling slightly, but that seems to be about right
Which auto manufacturer?
Ah. I work for Ford.
Not directly, Diesel engine control
domestic markets only or global
andypugh: you're in the UK, right?
thats a good place to be i'd say
no i'm in the US
ie everywhere except US, and US from 2010
Aye, I am in the UK
You might have guessed from my spelling.
actually I guessed because you said "whilst"
No, it's a word.
its english for while
i said that?
Well, it's a different tense.
dareposte: I asked andy if he was in the UK, not you
Aye, I am in the UK.
i did notice "tyre"
way off topic somehow
But I type in a Yorkshire accent and I am a pirate in another window, so there may be certain linguistic oddities.
So, stepping yet further up the ladder.
My preferred 3D design package is Autodesk Inventor (I used it full time for a couple of years) How do I get from that to G-code?
no clue at all
there's a plug-in for it i believe
aptly named "inventor cam"
i have never used it though
cradek has some AutoCad (Autolisp?) thing called Realize that will generate g-code from an autocad drawing, but I believe its 2.5D only
(not purchased from Autocad - he wrote it, but it only runs with autocad)
Inventor is a whole new world. No Autolisp as far as I know (no loss, to be frank)
for mill or lathe?
Inventor has one way to do things, not one new way and 8 legacy ways.
i had good results using CAMBAM, which is free
for a 2.5 axis mill
draw it as a .dxf, load it into cambam to define tool paths, and let it rip
What I have is a cheap lathe/mill. Curently is is zero axes, plan is any three from 4.
well for the mill, for easy pocketing and profiling i'd recommend cambam, it's pretty solid for 2.5d work
on the lathe i've been writing g code manually which is not as hard as i thought it would be
got a copy of NC Plot that simulates it
my lathe isn't fully finished yet though, so there's lots to learn still
if you want a full 3d package you might be better off learning mastercam or something like that
OK, I have just noticed that it is 3am.
But this has been a very useful session. I will be back.
Dallur: been sailing yet?
good fun reading the scrollback each day
archivist_ub: If reading logs is what you call "fun", I have a suggestion for you... GET A LIFE!
actually in here it has useful info, you could learn a thing ot six
archivist_ub: Sure, but you said FUN, not enlighting =)
learning is fun
you've been listening to those commericals too much
JymmmEMC: tried your pcmcia parport?
good night all
hmm, is the enco website (www.use-enco.com) broken for anybody else?
jmkasunich; looks good here
click on the main catalog link, or the sale catalog
actually, I can't even get the main page now
oh, all the links I've tested have worked
get an instant redirect to http://www.use-enco.com/CGI/INSRHM,
which reports "Internal Server Error Cannot read script output pipe"
funny. try a proxy maybe?
even more funny:
jmkasunich@mahan:~/emcdev/emc2head$ ping www.use-enco.com
PING www.use-enco.com (184.108.40.206) 56(84) bytes of data.
--- www.use-enco.com ping statistics ---
10 packets transmitted, 0 received, 100% packet loss, time 9044ms
jmkasunich: this is a linux box - right (stupid question I am sure)
hey, thats funny. I get a ping timeout too
ok - There have been some nasty viruses lately that have been doing some funky re-directs
do you guys have the same IP for "www.use-enco.com" as I do? 220.127.116.11
18.104.22.168 for me
yeah, I just used "dig", and thats what it came up with
wtf is going on?
wth, my pings time out
enco has multiple IP addresses
;; ANSWER SECTION:
neither pings for me
so they got a backup server
* renesis shrug
or just load distribution
When you would ping yahoo - it would come back instead of the ip address - it would be akadns.net/something....
so they got no response to pings
however, the webpage does work
* renesis shrugs
when I type in either of the IP addresses to foxfire, they almost instantly get replaced by http://www.use-enco.com/CGI/INSRHM
and I get the error page
works here also
jmkasunich: did you get any boring done?
not yet, getting close tho
I had a lot of chores to do this afternoon
and I spent this morning sleeping ;-)
oh is new ubuntu emc livecd built yet?
im still running that new machine in parport+usb only mode, heh
sd cards are so neat
neat, you guys have it up now
[22:52:26] <jmkasunich> http://www.pastebin.ca/1195966
* jmkasunich googles HTTP 302
heh - I loke 300: "Multiple Choices" :)
this is weird
wget fetches the enco main page, but firefox fetches the error page
I've cleared firefox's cache, dunno what else to try
it works for me, could just be one of those random errors that come and go
reboot the router. ;)
or come and stay
well, I fixed it
by closing all open instances of firefox and then trying again
I have an odd slowness I was putting down to the router here but it seems more to do with firefox a reboot of firefox cures it
I wonder if one of the two servers is down, and firefox was caching the IP address of the broken one
ah ok - FF may do what Mozilla has always done - cache DNS itself
so thats what the crappy ff does
thats not firefoxes job!
I guess FF is like windows - needs restarted at least once a day
I leave it up for ages(well was)
I was too
now rebooting it nearly daily
I usually shut it down every few days or so - there are still some memory leaks (at least in the Windows version)
neat - Enco has the dual-needle (metric/inch) dial calipers in their "outlet" catalog
dial! digital rulez
till the batteries die - my dial calipers (even cheap ones) will still be working long after electronic ones quit
I keep multiple digitals about
I skipped the dials, too much backlash in the one I tried
I just don't like dials. Seen too many of them get screwed up from crap in the rack
verniers were better
I still have my vernier.. (my first caliper)
mine was a Mitutoyo nive V slides on it
(but it is the last resort)
cradek: were you the one who was astonished by the price of small boring bars?
Enco has small solid carbide ones for $15-20
[23:32:57] <jmkasunich> http://www.use-enco.com/CGI/INPDFF?PMPAGE=217
is .8 far enough for you?