petev: I tried, but cannot catch an accel constraint violation.
cradek, what was yuur test case?
I'm not convinced I have a handle on all the different scaling and what will actually show up in the HAL yet
lots of test cases
was the feed at max velocity when the decel started?
you should test without me telling you how to do it
I know what the test case should be, but it's not clear to me what actually shows up in HAL
there is too much scaling going on in different places
hal position commands are in machine units
did u see signals that looked like they had the right values for velocity, etc.
run it thru 2 ddt blocks and you get velocity in machine units/sec and accel in machine units/sec^2
which are the same as the units in the inifile
sim has the ddt set up already
ok, I'll try and take a look at it
u guys there?
not me, I'm not here at all
ok, I got a voliation on the first test case
also, the accel waveforms look rather deformed from what I expected
screenshot on imagebin?
ok, hang on
hmm, it's very big uncompressed
how about I email it
wtf is bmp!
I'm on a doze box
it's a straight bitmap
I know - thats why I'm giving you a hard time
you can't do a "save as" .gif or .png?
let me see
bmp: bloated mega pixels ;-)
no, all I have is the crappy MS paint program and it won't do it
you need gimp
email on it's way
all I did is calc the raduis that should cause the problem from the values in the INI
[03:15:18] <jmkasunich> http://www.gimp.org/windows/
a real image editor
then I ran it and it was there
I don't do graphics ;-)
I'm serious - gimp is nice, there is an easy windows install, whats not to like?
I expected the accel to look like a square wave with trapezoidal velocity, but it's quite distorted
what do you think is going on there?
I dunno about the windows version, but my gimp has File->Acquire->Screenshot, lets me capture the whole screen or a window - very convenient
ok, what is your accel limit?
if my monitor wasn't out on the mill, I would run it on the linux box
but as it is, I'm using X from doze
straight from servo_sim.ini
accel is 20
vel is 1.5
so you are hitting the accel limit at the bottom
imagemagick is available for Windows
I calc the test case for the default sim INI file
did u get the email yet?
yeah, I'm looking at the pic
for a circle, shouldn't the accels be sin and cos?
thats a 90 degree arc?
why are they so distorted
or is that due to the tangential acell/decell ?
yeah, it will be distorted by the initial accel and the final decel (tangential)
no, look at the G-cod
takes about 28.6 degrees for accel and decel portions
if you say so (I don't read g-code arcs that well - remember, I do low level stuff)
the rest is cruise
you'll only have the nice sin/cos during cruise
ok, I can buy that
and with a 73 degree total, and 57 degrees used up in accel and decel, you ain't got much cruise
true, but I wanted to start decel near 45 degrees
next time do 163 degrees
that will get you a cruise segment long enough to get your bearings from
oh, another distortion source is just time
it takes about three times as long to do the first 28 degrees as it does to do the 16 degree cruise portion
your traj velocity is less than your axis velocity
I made mine higher so the axis vel would limit it
yeah, I saw that, but that's the default file
some of our default files are not great
if the axis limits are 20, then regardless of what the traj limit is, the -22ish value that he's seing is a violation
if you use g64 does it go away?
I wish there was a longer cruise portion, I'm having trouble correlating the traces to the move
I didn't think of trying g61 but I bet that's the key to triggering it
don't know, I wanted exact path, but it really shouldn't matter with a single arc
those are blending codes?
do a semicircle plus this
how about we put in the cos(45) factor and run the same test again?
I bet it won't violate
sure that's fine
it would be interesting to add a hypot component there, to get the total accel
sim may have those already
I suppose I should fire it up... but I'd rather work on vhdl
( 2007-04-07 18:57:24 - Project CVS Service ) On 2007-04-07, the CVS server for project letters m, n, and q has a disk problem that has caused a lapse in service. We are working to repair this.
their notices always remind me of sesame street
today's sesame street was brought to you by the letters w, n, and z, and the number 3
hmm.. maybe if I would watch I might get it..
but we didn't have it around here..
cradek: seems vmware server now supports SMP
if the host has 2 processors, it can provide those to the guest too..
I might give rtai & SMP another go
(although I'm not sure how much this would prove..)