subject:"Re\: \[Emc\-users\] multiswitch Q's"

On Monday 27 February 2017 15:24:34 Sebastian Kuzminsky wrote:

> On 02/27/2017 01:14 PM, Gene Heskett wrote:
> > On Monday 27 February 2017 11:42:53 Sebastian Kuzminsky wrote:
> >> On 02/27/2017 09:35 AM, Gene Heskett wrote:
> >>> And, I just checked, formatted the .list file entry according to
> >>> the wheezy entries at http://buildbot.linuxcnc.org/ and there is
> >>> not an entry in the repos to match a wheezy i386 install, which
> >>> this is because yonks ago, when I installed wheezy, lcnc didn't
> >>> run on 64 bit installs, so I've been running an i386 install, with
> >>> whichever kernel had an actually works PAE, without which I'm a
> >>> gigabyte into swap in just a few hours.  The error message from
> >>> apt-get update is:
> >>>
> >>> W: Failed to fetch http://linuxcnc.org/dists/wheezy/Release 
> >>> Unable to find expected entry '2.7-rtpreempt/binary-i386/Packages'
> >>> in Release file (Wrong sources.list entry or malformed file)
> >>
> >> Unfortunately the apt sources for the buildbot and for the main
> >> archive at www.linuxcnc.org use slightly different URLs.
> >>
> >> The "2.7-rtpreempt" component you're using is for the buildbot, not
> >> for www.linuxcnc.org.
> >>
> >> The matching component on www.linuxcnc.org is "2.7-uspace".  The
> >> apt source line should look like this:
> >>
> >> deb http://linuxcnc.org/ wheezy base 2.7-uspace
> >
> > And that, Seb, while not identical to what I did have, offers me the
> > exact same thing that is currently installed. No 2.8-pre whatever in
> > sight. Which is whats running on the other 3 real machines.
>
> The apt-get error message you quoted above shows that you're looking
> for a component called "2.7-rtpreempt" at "http://linuxcnc.org;.
>
> I'm guessing you have a line in your apt source that looks like this:
>
>   deb http://linuxcnc.org/ wheezy 2.7-rtpreempt
>
> That's not going to work, since there is no component named
> "2.7-rtpreempt" in the main linuxcnc.org package archive.
>
> I now think you want the not-yet-released "2.8-pre" version.  This
> version is not available from www.linuxcnc.org (since it's not
> released yet) but it is available from the buildbot.  You just have to
> add an apt-source line exactly like it says on the buildbot front
> page:
>
>   deb http://buildbot.linuxcnc.org/ wheezy master-sim
>
Ahh, this worked, but I was looking at the buildbot's front page. Or 
thought I was 3 hours ago...
>
> It's maybe confusing that this is called "-sim" instead of "-uspace",
> I
>
> tried (probably poorly) to clear up that confusion in the description:
> > Wheezy (uspace: realtime with RT-Preempt, and simulation)
> > architectures: amd64, armhf, i386


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Monday 27 February 2017 15:24:34 Sebastian Kuzminsky wrote:

> wheezy master-sim

That sounded like it should, but:

W: Failed to fetch http://linuxcnc.org/dists/wheezy/Release  Unable to 
find expected entry 'master-sim/binary-i386/Packages' in Release file 
(Wrong sources.list entry or malformed file) 

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

2017-02-27 Thread Sebastian Kuzminsky

On 02/27/2017 01:14 PM, Gene Heskett wrote:
> On Monday 27 February 2017 11:42:53 Sebastian Kuzminsky wrote:
>
>> On 02/27/2017 09:35 AM, Gene Heskett wrote:
>>> And, I just checked, formatted the .list file entry according to the
>>> wheezy entries at http://buildbot.linuxcnc.org/ and there is not an
>>> entry in the repos to match a wheezy i386 install, which this is
>>> because yonks ago, when I installed wheezy, lcnc didn't run on 64
>>> bit installs, so I've been running an i386 install, with whichever
>>> kernel had an actually works PAE, without which I'm a gigabyte into
>>> swap in just a few hours.  The error message from apt-get update is:
>>>
>>> W: Failed to fetch http://linuxcnc.org/dists/wheezy/Release  Unable
>>> to find expected entry '2.7-rtpreempt/binary-i386/Packages' in
>>> Release file (Wrong sources.list entry or malformed file)
>>
>> Unfortunately the apt sources for the buildbot and for the main
>> archive at www.linuxcnc.org use slightly different URLs.
>>
>> The "2.7-rtpreempt" component you're using is for the buildbot, not
>> for www.linuxcnc.org.
>>
>> The matching component on www.linuxcnc.org is "2.7-uspace".  The apt
>> source line should look like this:
>>
>> deb http://linuxcnc.org/ wheezy base 2.7-uspace
>>
> And that, Seb, while not identical to what I did have, offers me the
> exact same thing that is currently installed. No 2.8-pre whatever in
> sight. Which is whats running on the other 3 real machines.

The apt-get error message you quoted above shows that you're looking for 
a component called "2.7-rtpreempt" at "http://linuxcnc.org;.

I'm guessing you have a line in your apt source that looks like this:

deb http://linuxcnc.org/ wheezy 2.7-rtpreempt

That's not going to work, since there is no component named 
"2.7-rtpreempt" in the main linuxcnc.org package archive.

I now think you want the not-yet-released "2.8-pre" version.  This 
version is not available from www.linuxcnc.org (since it's not released 
yet) but it is available from the buildbot.  You just have to add an 
apt-source line exactly like it says on the buildbot front page:

deb http://buildbot.linuxcnc.org/ wheezy master-sim


It's maybe confusing that this is called "-sim" instead of "-uspace", I 
tried (probably poorly) to clear up that confusion in the description:

> Wheezy (uspace: realtime with RT-Preempt, and simulation)
> architectures: amd64, armhf, i386


-- 
Sebastian Kuzminsky

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Monday 27 February 2017 11:42:53 Sebastian Kuzminsky wrote:

> On 02/27/2017 09:35 AM, Gene Heskett wrote:
> > And, I just checked, formatted the .list file entry according to the
> > wheezy entries at http://buildbot.linuxcnc.org/ and there is not an
> > entry in the repos to match a wheezy i386 install, which this is
> > because yonks ago, when I installed wheezy, lcnc didn't run on 64
> > bit installs, so I've been running an i386 install, with whichever
> > kernel had an actually works PAE, without which I'm a gigabyte into
> > swap in just a few hours.  The error message from apt-get update is:
> >
> > W: Failed to fetch http://linuxcnc.org/dists/wheezy/Release  Unable
> > to find expected entry '2.7-rtpreempt/binary-i386/Packages' in
> > Release file (Wrong sources.list entry or malformed file)
>
> Unfortunately the apt sources for the buildbot and for the main
> archive at www.linuxcnc.org use slightly different URLs.
>
> The "2.7-rtpreempt" component you're using is for the buildbot, not
> for www.linuxcnc.org.
>
> The matching component on www.linuxcnc.org is "2.7-uspace".  The apt
> source line should look like this:
>
> deb http://linuxcnc.org/ wheezy base 2.7-uspace
>
And that, Seb, while not identical to what I did have, offers me the 
exact same thing that is currently installed. No 2.8-pre whatever in 
sight. Which is whats running on the other 3 real machines.

>
> Here's the relevant docs, for reference:
>
> http://linuxcnc.org/docs/2.7/html/getting-started/getting-linuxcnc.htm
>l#_alternate_install_methods


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Monday 27 February 2017 11:05:23 andy pugh wrote:

> On 27 February 2017 at 15:52, Gene Heskett  
wrote:
> > How wide, in terms of inputs, can I make a single instance of this
> > mux_generic?
>
> The manual page says:
> "This component is a more general version of the other multiplexing
> components. It allows the creation of arbitrary-size multiplexers (up
> to 1024 entries)"
>
In a single instance? Wow!  That would be nearly 100 times what I need 
for this. It would also be one heck of a pile of hal code! Pages of 
setp's or...

Humm, I also need a scale function that runs on s32's. What we have runs 
on floats. OTOH, an s32-to-float in front of the whole chain, and a 
float-to-s32 at the output would work, but the execution time of all 
those floats in the middle might be a hurt on the pi.  Ideally, all this 
should run only if conditions have changed, but I've no clue how to 
write an if/else/else/else/end in hal.

> I suspect that you can derive up-pulse and down-pulse signals from
> logic and the raw A B phases.

As I had planned. But no progress yet, honeydo stuff in the way. ;)

> I think I would start with a pencil and paper and a self-referencing
> LUT5. But then LUT5 seems to be my go-to component for almost
> anything.

So I have noticed. :) If its low in execution time, and you wrote it, so 
you understand it intimately, why not? ;-) I tend to look for less 
complex bits and pieces I can stack up in the right sequence.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

2017-02-27 Thread Sebastian Kuzminsky

On 02/27/2017 09:35 AM, Gene Heskett wrote:
>
> And, I just checked, formatted the .list file entry according to the
> wheezy entries at http://buildbot.linuxcnc.org/ and there is not an
> entry in the repos to match a wheezy i386 install, which this is because
> yonks ago, when I installed wheezy, lcnc didn't run on 64 bit installs,
> so I've been running an i386 install, with whichever kernel had an
> actually works PAE, without which I'm a gigabyte into swap in just a few
> hours.  The error message from apt-get update is:
>
> W: Failed to fetch http://linuxcnc.org/dists/wheezy/Release  Unable to
> find expected entry '2.7-rtpreempt/binary-i386/Packages' in Release file
> (Wrong sources.list entry or malformed file)

Unfortunately the apt sources for the buildbot and for the main archive 
at www.linuxcnc.org use slightly different URLs.

The "2.7-rtpreempt" component you're using is for the buildbot, not for 
www.linuxcnc.org.

The matching component on www.linuxcnc.org is "2.7-uspace".  The apt 
source line should look like this:

deb http://linuxcnc.org/ wheezy base 2.7-uspace


Here's the relevant docs, for reference:

http://linuxcnc.org/docs/2.7/html/getting-started/getting-linuxcnc.html#_alternate_install_methods


-- 
Sebastian Kuzminsky

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Monday 27 February 2017 09:40:51 andy pugh wrote:

> On 27 February 2017 at 14:08, Gene Heskett  
wrote:
> > And on this machine running a sim;
> > 2.7.8-49-g5ce3159
>
> There you are, your sim config is running 2.7

And, I just checked, formatted the .list file entry according to the 
wheezy entries at http://buildbot.linuxcnc.org/ and there is not an 
entry in the repos to match a wheezy i386 install, which this is because 
yonks ago, when I installed wheezy, lcnc didn't run on 64 bit installs, 
so I've been running an i386 install, with whichever kernel had an 
actually works PAE, without which I'm a gigabyte into swap in just a few 
hours.  The error message from apt-get update is:

W: Failed to fetch http://linuxcnc.org/dists/wheezy/Release  Unable to 
find expected entry '2.7-rtpreempt/binary-i386/Packages' in Release file 
(Wrong sources.list entry or malformed file)

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On 27 February 2017 at 15:52, Gene Heskett  wrote:

> How wide, in terms of inputs, can I make a single instance of this
> mux_generic?

The manual page says:
"This component is a more general version of the other multiplexing
components. It allows the creation of arbitrary-size multiplexers (up
to 1024 entries)"

I suspect that you can derive up-pulse and down-pulse signals from
logic and the raw A B phases.

I think I would start with a pencil and paper and a self-referencing
LUT5. But then LUT5 seems to be my go-to component for almost
anything.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1916

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Monday 27 February 2017 09:01:47 andy pugh wrote:

> On 27 February 2017 at 13:49, Gene Heskett  
wrote:
> > What I wanted was a binary counter, whose individual bitN outputs
> > could drive the selN lines of a mux16, thereby selecting the gain to
> > be applied to the jogwheel output.
>
> Mux-generic can be driven by an integer (and, possibly usefully for
> you, you can also use the bit-inputs to "shift" the selection.)
> http://linuxcnc.org/docs/2.7/html/man/man9/mux_generic.9.html
>
> http://linuxcnc.org/docs/2.7/html/man/man9/updown.9.html
> Looks like the solution to the counting up and down part.
>
> One slight inconvenience is that updown is S32 out and mux_generic is
> U32 in.

Which the s32 says updown can count negative values, which is good. That 
means I can design for all speeds in both directions.

Humm, and what does it do when the input is an s32 representing a 
negative? Say of a -3 or a -14 (no floats allowed)?

I'll start with the updown and see if I can make the min/max behave as 
desired. I had rejected it originally because of the logic needed to 
determine the direction, derived by a wcomp from the encoders velocity 
output, and applied to a not and 2 and2's to derive the step up for that 
pin, or the step down for the other.

How wide, in terms of inputs, can I make a single instance of this 
mux_generic?  Sounds interesting but a bit of a bear to properly setup 
with all those options. 16 is likely overkill with a 1,2,5 sequence to 
the gain. In practice, 6 or 7 would be sufficient I think, if not 
overkill.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On 27 February 2017 at 14:08, Gene Heskett  wrote:
> And on this machine running a sim;
> 2.7.8-49-g5ce3159

There you are, your sim config is running 2.7

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1916

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Monday 27 February 2017 05:04:10 andy pugh wrote:

> On 26 February 2017 at 20:22, Gene Heskett  
wrote:
> > In the x86 version of man 9 motion, those are all there I believe. 
> > Yes. In my sim version on this base machine, only axis is listed,
> > joint stuff hasn't jumped over the fence into the simulators, yet. 
> > In the arm version, the whole section on axis[X|Y|Z] etc has been
> > expunged, and replaced with joint#'s and the axis stuffs are not
> > even visible to a halmeter, or to halshow.
>
> It seems rather more likely that you simply aren't using the version
> of LinuxCNC that you think you are.

On the G0704, which is x86;

2.8.0-pre1-2863-ga9120ef
from its sources.list;
deb http://buildbot.linuxcnc.org/ wheezy master-rt

On the pi;
2.8.0-pre1-2771-gdc2ff49
its sources.list entry;
deb http://buildbot.linuxcnc.org/ jessie master-sim

And on this machine running a sim;
2.7.8-49-g5ce3159
deb http://linuxcnc.org/ wheezy 2.7-uspace base 2.6 

Neither of the x86 boxes have had a sources edit in about a year.
Please show me any updated listings, bearing in mind that the pi is the 
only one running jessie, everything else is still on wheezy, but all are 
uptodate a/o this Friday past.

Thanks Andy

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On 27 February 2017 at 13:49, Gene Heskett  wrote:
> What I wanted was a binary counter, whose individual bitN outputs could
> drive the selN lines of a mux16, thereby selecting the gain to be
> applied to the jogwheel output.

Mux-generic can be driven by an integer (and, possibly usefully for
you, you can also use the bit-inputs to "shift" the selection.)
http://linuxcnc.org/docs/2.7/html/man/man9/mux_generic.9.html

http://linuxcnc.org/docs/2.7/html/man/man9/updown.9.html
Looks like the solution to the counting up and down part.

One slight inconvenience is that updown is S32 out and mux_generic is U32 in.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1916

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Monday 27 February 2017 05:02:29 andy pugh wrote:

> On 27 February 2017 at 03:27, Gene Heskett  
wrote:
> > So for this use, it resembles those appendages on the belly of a
> > boar hog in usefulness.
>
> Maybe it would help if you explained what you are trying to use it
> for.

Humm, I thought I had laid it out in prior messages, maybe not clear 
enough. My mind has been known to skip a whole paragraph when its 
occupied by the actual typing.  I am not a touch typist with my fingers 
running automatically in pace with my thoughts.

What I wanted was a binary counter, whose individual bitN outputs could 
drive the selN lines of a mux16, thereby selecting the gain to be 
applied to the jogwheel output. My plan is to use the nibble sized 
output, plus a sign bit, so I theoretically could have 16 speeds in 
either direction, with a spst button right beside the dial to switch 
from driving motions jog input, to moving the gain control counter one 
step per dial click when the button is pushed, this to scale the counts 
obtained when the button isn't pushed. Using a mux16 to select the gain 
multiplier on a 1,2,5 repeating sequence, allowing both very fine 
response of one motor microstep step per detent of the dial, up to 
faster than the machine can move with relatively few "clicks" of the 
dial wheel.

I plan to display the state of the speed settings with a pair of hbars in 
pyvcp.  But that bit of pyvcp/xml hasn't been written yet.

Last night, with only one eye open, I read lots more manpages, and it 
looks like I might be able to achieve this 4 bit mux16 address with 
bitslice and some limiting to keep the value of a u32 in-range, using 
similar logic to inc/dec the u32.

Write & test when I've input enough coffee to get both eyes open at the 
same time. I just woke up, got us each a cuppa, and I'll see what my 
lady, recovering from a hip replacement 3 weeks ago tomorrow, wants for 
breakfast, next, while a new pot is making, this was yesterdays cold and 
nuked leftovers.

She is getting around on a walker pretty good now, but there's still a 
bunch of distracting honeydo's a day too.

I don't recommend getting old Andy, it is not all its promoted to be by 
AARP.  Avoid it if you can. But its catching up with us, darnit. ;-/

Thanks Andy.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On 26 February 2017 at 20:22, Gene Heskett  wrote:
>
> In the x86 version of man 9 motion, those are all there I believe.  Yes.
> In my sim version on this base machine, only axis is listed, joint stuff
> hasn't jumped over the fence into the simulators, yet.  In the arm
> version, the whole section on axis[X|Y|Z] etc has been expunged, and
> replaced with joint#'s and the axis stuffs are not even visible to a
> halmeter, or to halshow.

It seems rather more likely that you simply aren't using the version
of LinuxCNC that you think you are.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1916

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On 27 February 2017 at 03:27, Gene Heskett  wrote:
> So for this use, it resembles those appendages on the belly of a boar
> hog in usefulness.

Maybe it would help if you explained what you are trying to use it for.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1916

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Sunday 26 February 2017 15:53:49 Gene Heskett wrote:

Further experimentation finally discloses that the bitN outputs are a 
single true bit, marching along however wide an array of bits is defined 
in the loadrt cfg= line.  And that overflows and underflows simply wrap. 
So for this use, it resembles those appendages on the belly of a boar 
hog in usefulness.

Short of a stack of flip-flops, does anyone else have an idea that would 
help?

Thanks.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Sunday 26 February 2017 10:45:19 Gene Heskett wrote:

Doing some clarifications and deletions as multiswitch is my REAL problem 
ATM

> On Saturday 25 February 2017 14:35:16 Gene Heskett wrote:
> Ping?
>
> > Hi all;
> >
> > Couple problems with multiswitch.
> >
> > Default output for no input is 0001, not . Why?
> >
> > Cannot be loadrt'd with the names= option. cfg=4,4 works fine.  Why?
> > I would really like to use the  state as the default, gain=1.0
> > position.
>
> Ack the multiswitch manpage, the "parameters" aren't clear to me.

> First, I'd like 
> to know what happens to the bit line logic progressions when it is
> asked to decrement to a below zero, aka a - position. Some
> experimentation may be required. But some clarification of the
> interaction between writing a u32 or s32 to either parameter, and its
> effect on the bit.N outputs from the author would be most appreciated.
> If they track each other that would be great. Humm, I can check that
> with halshow I believe. Yes.
>
> But, while I can read back what I set in the position register, I just
> spent much of an hour observing the bit.N lines, and only one actually
> changes its state with a change in position, bit.0 acts as a zero
> indicator being true if position=0.
>
> Any non-zero value written sets bit-0 false which the other 3 bits are
> and they remain false for any 0-255 value in the position register. A
> negative written to position is converted to a positive value in a
> halmeter, so it does not count below zero.
>
> Seems to me I either do not understand how to use it, or its broken.
>
> In either case, its broken for my use and cannot do this job that I
> expected it to be able to from the man page. So something is all
> aglay, me or the manpage?
>
> What I wanted was a nibble sized number of bits to drive the sel-0
> thru sel-3 lines of a mux16.
>
> Help on multiswitch!

Thanks.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

On Sunday 26 February 2017 12:13:15 Dewey Garrett wrote:

> > Also, the manpage for motion needs updated to reflect
> > that "axis letter of axis" is now "joint.#. of" with a
> > slight change in syntax to numerical. Hooking up to the
> > jog-inputs caught me out for a bit.
>
> In the master branch, as updated for joints_axes, there
> are hal pins for wheel jogging in jointmode and
> _different_ hal pins for jogging axes (coordinate
> letters) in world (teleop) mode.
>
> The motion man page identifies the hal pins correctly:
>
>   joint.N.jog-counts IN S32
>   joint.N.jog-enable IN BIT
>   joint.N.jog-scale IN FLOAT
>   joint.N.jog-vel-mode IN BIT
>
>   axis.L.jog-counts IN S32
>   axis.L.jog-enable IN BIT
>   axis.L.jog-scale IN FLOAT
>   axis.L.jog-vel-mode IN BIT
>
> N specifies numerical joint jumbers.
> L specifies axis coordinate letters.
>
This depends on the architecture, Dewey.

Read on.

> Support for wheel jogging in joint and world modes is
> specifically described in the documentation chapter for
> 'Updating LinuxCNC':
>
> http://linuxcnc.org/docs/devel/html/getting-started/updating-linuxcnc.
>html#_hal_changes_updates_for_joints_axes
>
> In general, joints are not the same as axes although
> the distinctions may be hidden from the user by the
> graphical user interface (the axis gui for instance)
> when using IDENTITY kinematics (trivkins for instance).
>
> Coordinates are 'homed' by moving joints in jointmode
> in accordance with home switch closures and the homing
> procedures as specified with inifile settings for each
> joint (settings like [JOINT_N]HOME_* = value).
>
> Upon successful homing of all joints, a gui (axis for
> instance), will typically change from joint to world
> mode operation.  This automatic transition may not
> be apparent to a user when identity kinematics are
> used.
>
> Once homed, returning to joint mode is seldom needed so
> in many configurations, wheel jogging is only wanted
> for world mode operations and the hal pins for joint
> jogging are not used.

In the x86 version of man 9 motion, those are all there I believe.  Yes. 
In my sim version on this base machine, only axis is listed, joint stuff 
hasn't jumped over the fence into the simulators, yet.  In the arm 
version, the whole section on axis[X|Y|Z] etc has been expunged, and 
replaced with joint#'s and the axis stuffs are not even visible to a 
halmeter, or to halshow.

So if I'm looking at the manpage from an x86 machine, and trying to write 
hal code for the R-Pi, its confusing, so I now have 2 ssh sessions 
running into the pi so that I am at least on the same page, I hope.  
That faint knocking sound? Yup. :)

All machines are updated to Master several times a week here. However, 
its been around 2 weeks now since the arm/jessie LCNC has actually been 
updated.

In fact I was about to ask if the buildbot was bailing out on the armhf 
stuff.  But armhf for jessie seems to have been thrown under the bus in 
recent times, with only one small library having been updated in the 
same time frame from debiam-arm.

Thanks Dewey.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's

2017-02-26 Thread Dewey Garrett

> Also, the manpage for motion needs updated to reflect
> that "axis letter of axis" is now "joint.#. of" with a
> slight change in syntax to numerical. Hooking up to the
> jog-inputs caught me out for a bit.

In the master branch, as updated for joints_axes, there
are hal pins for wheel jogging in jointmode and
_different_ hal pins for jogging axes (coordinate
letters) in world (teleop) mode.

The motion man page identifies the hal pins correctly:

joint.N.jog-counts IN S32
joint.N.jog-enable IN BIT
joint.N.jog-scale IN FLOAT
joint.N.jog-vel-mode IN BIT

axis.L.jog-counts IN S32
axis.L.jog-enable IN BIT
axis.L.jog-scale IN FLOAT
axis.L.jog-vel-mode IN BIT

N specifies numerical joint jumbers.
L specifies axis coordinate letters.

Support for wheel jogging in joint and world modes is
specifically described in the documentation chapter for
'Updating LinuxCNC':

http://linuxcnc.org/docs/devel/html/getting-started/updating-linuxcnc.html#_hal_changes_updates_for_joints_axes

In general, joints are not the same as axes although
the distinctions may be hidden from the user by the
graphical user interface (the axis gui for instance)
when using IDENTITY kinematics (trivkins for instance).

Coordinates are 'homed' by moving joints in jointmode
in accordance with home switch closures and the homing
procedures as specified with inifile settings for each
joint (settings like [JOINT_N]HOME_* = value).

Upon successful homing of all joints, a gui (axis for
instance), will typically change from joint to world
mode operation. This automatic transition may not
be apparent to a user when identity kinematics are
used.

Once homed, returning to joint mode is seldom needed so
in many configurations, wheel jogging is only wanted
for world mode operations and the hal pins for joint
jogging are not used.

--
Dewey Garrett

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] multiswitch Q's