Re: [Emc-users] Something to think about re the hack-a-day tool changer

[Gene Heskett]

On Wednesday 17 April 2019 15:06:02 Chris Albertson wrote:

[...]
>
> But if maybe you lock the spindle and turn the nut.

This is the case, except you are turnbing the nut by using the xy 
steppers to drive the carousel which is turning the nut.

> Then your spindle 
> lock needs to have a torque gauge fitted.  The gauge is either a
> spring and switch or a load cell.   The switch is much easier to
> interface with.

This is true, but before I start, how much power to I have in the 
steppers before they start slipping steps? In order to get enough torque 
on the nut, I need to know how much push is available w/o having to 
rehome the thing after I've tightened the nut.  That tells me how long a 
wrench "handle" I need to achieve the required torque.  That handle 
length is the radius of where the nut is from the center bearing of the 
carousel.

> On Wed, Apr 17, 2019 at 12:17 AM Gene Heskett  
wrote:
> > On Wednesday 17 April 2019 01:48:06 Chris Albertson wrote:
> > > I think you could eliminate the load cell and simply measure motor
> > > current.  Given that the motor is locked at zero RPM, the torque
> > > would be a function of current.  Once you calibrate current to
> > > torque it will not change.   You still need to get this data into
> > > a computer but at least you don't need to build a mechanical
> > > widget, just a hall effect sensor on more lead wire.

Take care Chris.


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 



___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Something to think about re the hack-a-day tool changer

[Gene Heskett]

On Wednesday 17 April 2019 15:32:09 Ken Strauss wrote:

> The cheap load cells often use a HX711 for the amplifier and A/D.
> See
> https://www.instructables.com/id/How-to-Interface-HX711-Balance-Module
>-With- Load-Ce/ if you're using an Arduino. If you want to roll your
> own interface code see
> https://www.mouser.com/ds/2/813/hx711_english-1022875.pdf
>
Thanks Ken, that is exactly what I needed.  Now all I  need to do is 
round up the bits and pieces.

> > -Original Message-
> > From: Gene Heskett [mailto:ghesk...@shentel.net]
> > Sent: Wednesday, April 17, 2019 10:01 AM
> > To: emc-users@lists.sourceforge.net
> > Subject: Re: [Emc-users] Something to think about re the hack-a-day
> > tool changer
> >
> > On Wednesday 17 April 2019 07:12:53 andy pugh wrote:
> > > On Wed, 17 Apr 2019 at 06:57, Gene Heskett 
> >
> > wrote:
> > > > Some more along the lines of finding out how much force I can
> > > > figure on as being available to tighten or loosen the collet
> > > > nut. I can get, for under a tenner, 4 ea 50Kg load cells and a
> > > > processor board that converts the very low level diffs of a
> > > > wheatstone bridge
> > >
> > > Why not set your torque wrench to the correct torque for your nuts
> > > and then see how far down the handle you have to be to trip the
> > > wrench?
> > >
> > > That tells you what radius of action you need to get the right
> > > torque.
> > >
> > > There is no question that your machine can produce the torque,
> > > given the classical "sufficiently long lever"
> >
> > And that is the final question as to whether or not the idea is
> > actually practical. It assuredly is not if the xy motors are so weak
> > it needs a carousel more then 18" in radius to the circle of sockets
> > just to tighten a tool well enough to dig off .010" of an alu chip
> > without slippage of tool in the collet.  Even that isn't going to be
> > practical but it does serve to outline exactly why the 3" radii
> > carousels we are seeing in the videos will turn out to be an obvious
> > disproof of concept. The videos of it going thru the motions on
> > hackaday would certainly have to suffer from the loose tools
> > syndrome if actually put to work on a usable thickness of alu stock,
> > say .03125" thick. Putting bigger motors on it would be one option,
> > but even 470 oz nema 23's, might not be enough, and most certainly
> > could/would bend the frame or bearing rods. I have the motors from
> > the broken HF I can move, and probably will as that would at least
> > double the power, but just haven't found the round tuit yet. Among
> > other things they have dampers on them. But I'll have to change the
> > A motor again, putting a 435oz on it. I already changed the 90 for a
> > 230, so a 435 can't fail to be even better at holding work. That
> > would leave the pair of 235's for XY, and would leave the longer Y
> > to be moved to Z duties.
> >
> > The 1600oz nema 34 supplied as Z motor in the kit for a G0704, which
> > was a too slow disaster on the G0704, often stalling at 29 ipm, but
> > was strong enough to stretch the bolts anchoring the nut carrier to
> > the z sled, causing those to need replacement quite early, but now
> > moves at 90 ipm with a 940oz motor that has not further damaged
> > those replacement bolts, would be a prime example of overkill.
> > However I reused that motor as Z drive on the Sheldon, where it has
> > no weight to lift, works well at 75 ipm on the Sheldon with the same
> > driver I took out of the G0704. I put in the $180 AC powered drive
> > to spin that 940oz on the GO704. Dead smooth and 20 db quieter. The
> > DM860H drive is a noisy drive, so noisy I had to install miniature
> > quarter round on the keyboard shelf edges to keep stuff on it,
> > including the keyboard and mouse. The 860 has very very poor step vs
> > amps in coils calibration, so there is no nice quiet all steps in
> > the microstep mapping regardless of the current setting. The thing
> > you notice most is that a /8 setting actually does 7 steps because
> > one of them is way too small.
> >
> > But thats not solving this measurement problem. Since this $7 kit,
> > needs an external clock, it seems like that could be done by siggen,
> > for a 500 baud baud rate, and a software uart could do the rest. But
> > the packet length is unk until I have such a kit in my hot little
> > hands. The fleabay adv's don't say.
> >
> > 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 
> >
> >
> >
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> 

Re: [Emc-users] Something to think about re the hack-a-day tool changer

[Ken Strauss]

The cheap load cells often use a HX711 for the amplifier and A/D.
See
https://www.instructables.com/id/How-to-Interface-HX711-Balance-Module-With-
Load-Ce/ if you're using an Arduino. If you want to roll your own interface
code see https://www.mouser.com/ds/2/813/hx711_english-1022875.pdf


> -Original Message-
> From: Gene Heskett [mailto:ghesk...@shentel.net]
> Sent: Wednesday, April 17, 2019 10:01 AM
> To: emc-users@lists.sourceforge.net
> Subject: Re: [Emc-users] Something to think about re the hack-a-day tool
> changer
>
> On Wednesday 17 April 2019 07:12:53 andy pugh wrote:
>
> > On Wed, 17 Apr 2019 at 06:57, Gene Heskett 
> wrote:
> > > Some more along the lines of finding out how much force I can figure
> > > on as being available to tighten or loosen the collet nut. I can
> > > get, for under a tenner, 4 ea 50Kg load cells and a processor board
> > > that converts the very low level diffs of a wheatstone bridge
> >
> > Why not set your torque wrench to the correct torque for your nuts and
> > then see how far down the handle you have to be to trip the wrench?
> >
> > That tells you what radius of action you need to get the right torque.
> >
> > There is no question that your machine can produce the torque, given
> > the classical "sufficiently long lever"
>
> And that is the final question as to whether or not the idea is actually
> practical. It assuredly is not if the xy motors are so weak it needs a
> carousel more then 18" in radius to the circle of sockets just to
> tighten a tool well enough to dig off .010" of an alu chip without
> slippage of tool in the collet.  Even that isn't going to be practical
> but it does serve to outline exactly why the 3" radii carousels we are
> seeing in the videos will turn out to be an obvious disproof of concept.
> The videos of it going thru the motions on hackaday would certainly have
> to suffer from the loose tools syndrome if actually put to work on a
> usable thickness of alu stock, say .03125" thick. Putting bigger motors
> on it would be one option, but even 470 oz nema 23's, might not be
> enough, and most certainly could/would bend the frame or bearing rods. I
> have the motors from the broken HF I can move, and probably will as that
> would at least double the power, but just haven't found the round tuit
> yet. Among other things they have dampers on them. But I'll have to
> change the A motor again, putting a 435oz on it. I already changed the
> 90 for a 230, so a 435 can't fail to be even better at holding work.
> That would leave the pair of 235's for XY, and would leave the longer Y
> to be moved to Z duties.
>
> The 1600oz nema 34 supplied as Z motor in the kit for a G0704, which was
> a too slow disaster on the G0704, often stalling at 29 ipm, but was
> strong enough to stretch the bolts anchoring the nut carrier to the z
> sled, causing those to need replacement quite early, but now moves at 90
> ipm with a 940oz motor that has not further damaged those replacement
> bolts, would be a prime example of overkill. However I reused that motor
> as Z drive on the Sheldon, where it has no weight to lift, works well at
> 75 ipm on the Sheldon with the same driver I took out of the G0704. I
> put in the $180 AC powered drive to spin that 940oz on the GO704. Dead
> smooth and 20 db quieter. The DM860H drive is a noisy drive, so noisy I
> had to install miniature quarter round on the keyboard shelf edges to
> keep stuff on it, including the keyboard and mouse. The 860 has very
> very poor step vs amps in coils calibration, so there is no nice quiet
> all steps in the microstep mapping regardless of the current setting.
> The thing you notice most is that a /8 setting actually does 7 steps
> because one of them is way too small.
>
> But thats not solving this measurement problem. Since this $7 kit, needs
> an external clock, it seems like that could be done by siggen, for a 500
> baud baud rate, and a software uart could do the rest. But the packet
> length is unk until I have such a kit in my hot little hands. The
> fleabay adv's don't say.
>
> 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 
>
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users




___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Something to think about re the hack-a-day tool changer

[Chris Albertson]

I thought the nut that needed to be torqued down was on the spindle.   You
would not be using a stepper for the spindle motor.
So the "wrench" holds the nut and then you power up the spindle to tighten
the nut.   If that is the case then spindle motor current
tells you the torque on the nut.

But if maybe you lock the spindle and turn the nut.  Then your spindle lock
needs to have a torque gauge fitted.  The gauge is either a spring and
switch or a load cell.   The switch is much easier to interface with.


On Wed, Apr 17, 2019 at 12:17 AM Gene Heskett  wrote:

> On Wednesday 17 April 2019 01:48:06 Chris Albertson wrote:
>
> > I think you could eliminate the load cell and simply measure motor
> > current.  Given that the motor is locked at zero RPM, the torque would
> > be a function of current.  Once you calibrate current to torque it
> > will not change.   You still need to get this data into a computer but
> > at least you don't need to build a mechanical widget, just a hall
> > effect sensor on more lead wire.
> >
> > One thing is to be VERY careful when powering a stalled motor s you
> > can burn it up very easy.   Even if you do add the load cell be sure
> > to measure and limit the current and keep it will under the limit.
> > Then you need to cut the power quickly
>
> Ahh, Chris, this is not a servo, I am talking about the XY motors, which
> are steppers and not particularly strong ones.  They are effectively
> always stalled and the current used is set by the dipswitches on the
> driver.  So generally speaking, measureing the current is meaningless to
> a stepper that has remained locked to the steps issued.
> >
> > Coincides or not but just a second before reading your post I had
> > typed the line below to define where I'm connecting a current sensor
> > to an SMT32F411 chip pin PA4.  So I thought this could apply to your
> > project too.
> >
> > #define MM_CS1  PA4   // A2 Current Sense MUST LIMIT VOLTS TO 3V3
> >
> > One more idea that is even more simple.  I assume you are measuring
> > torque so you know the nut is on hard enough that it will not come
> > loose. Rather then a load cell why not just a simple coil spring and a
> > microswitch.  It will always require the same amount of force to
> > compress the spring and then the switch is tripped and yu can back off
> >   This is basically a load cell with one-bit resolution.
>
> That, with some monkey business incorporating the switch into the sliding
> wrench channel that is pushed out to engage the double flats on the
> chuck shank, might be doable, but the spring to absorb the torque would
> have to be 50x stronger, and if not restrained properly would prevent
> the wrench from being moved out to grab the flats, or retracting it to
> free the spindle. This wrench must be free to move in and out quickly as
> its used only the break the nut loose, then retracted and the motor spun
> yo finish the loosen and drop the tool or tighten to about finger tight,
> prior to using the xy motors, driven in an arc to match the radius the
> sockets are mounted on the carousel disk. This power from the xy motors
> is what breaks the nut loose or cinches it up to hold the tool while
> cutting. Spindle power is not applied during that circular move as the
> retractable wrench has it immobilized anyway, so the the nut is driven
> tight, or backed off depending on which direction the nut and carousel
> is being rotated by the xy motors.  I'd be forced to use an air cylinder
> to get the needed force to move the wrench. Could be done, but not in
> the physical space available since it would take at least a 1"bore x
> 3/4"stroke double action cylinder. I expect there will need to be a
> pause, and the spindle motor pulsed to turn until the wrench engages the
> flats, then pulsed the other way to make sure its fully engaged. How to
> do that while the nut is still tight, is another problem yet to be
> solved if the gator style socket isn't used. I don't know why, but I've
> an aversion to that as I have a set of those and they work but are a
> pain in the ass to place over the hex properly. The set I have do not
> have retractable pins like the gators either.  Different designs maybe?
> Probably.
>
> If you've not watched the videos on hackaday, that would explain it at
> least as well as I can.
>
> > On Tue, Apr 16, 2019 at 9:57 PM Gene Heskett 
> wrote:
> > > On Tuesday 16 April 2019 14:34:15 Gene Heskett wrote:
> > > > Greetings all;
> > > >
> > > > While adding a couple timedelays and a couple or2's to my hal
> > > > file, basically to prove that I can pulse the spindle for the
> > > > nominally 200 msecs used to spin the nut on/off to release the
> > > > collet and supposedly drop the tool, or just drop it all, I came
> > > > up against a designed in problem.  This vfd has a pid in it that
> > > > will no doubt wind up if the speed command is present for any
> > > > great amount of time, so its possible I might have to cobble up a
> > > > set of PID 

Re: [Emc-users] Something to think about re the hack-a-day tool changer

[Gene Heskett]

On Wednesday 17 April 2019 07:12:53 andy pugh wrote:

> On Wed, 17 Apr 2019 at 06:57, Gene Heskett  
wrote:
> > Some more along the lines of finding out how much force I can figure
> > on as being available to tighten or loosen the collet nut. I can
> > get, for under a tenner, 4 ea 50Kg load cells and a processor board
> > that converts the very low level diffs of a wheatstone bridge
>
> Why not set your torque wrench to the correct torque for your nuts and
> then see how far down the handle you have to be to trip the wrench?
>
> That tells you what radius of action you need to get the right torque.
>
> There is no question that your machine can produce the torque, given
> the classical "sufficiently long lever"

And that is the final question as to whether or not the idea is actually 
practical. It assuredly is not if the xy motors are so weak it needs a 
carousel more then 18" in radius to the circle of sockets just to 
tighten a tool well enough to dig off .010" of an alu chip without 
slippage of tool in the collet.  Even that isn't going to be practical 
but it does serve to outline exactly why the 3" radii carousels we are 
seeing in the videos will turn out to be an obvious disproof of concept.  
The videos of it going thru the motions on hackaday would certainly have 
to suffer from the loose tools syndrome if actually put to work on a 
usable thickness of alu stock, say .03125" thick. Putting bigger motors 
on it would be one option, but even 470 oz nema 23's, might not be 
enough, and most certainly could/would bend the frame or bearing rods. I 
have the motors from the broken HF I can move, and probably will as that 
would at least double the power, but just haven't found the round tuit 
yet. Among other things they have dampers on them. But I'll have to 
change the A motor again, putting a 435oz on it. I already changed the 
90 for a 230, so a 435 can't fail to be even better at holding work. 
That would leave the pair of 235's for XY, and would leave the longer Y 
to be moved to Z duties. 

The 1600oz nema 34 supplied as Z motor in the kit for a G0704, which was 
a too slow disaster on the G0704, often stalling at 29 ipm, but was 
strong enough to stretch the bolts anchoring the nut carrier to the z 
sled, causing those to need replacement quite early, but now moves at 90 
ipm with a 940oz motor that has not further damaged those replacement 
bolts, would be a prime example of overkill. However I reused that motor 
as Z drive on the Sheldon, where it has no weight to lift, works well at 
75 ipm on the Sheldon with the same driver I took out of the G0704. I 
put in the $180 AC powered drive to spin that 940oz on the GO704. Dead 
smooth and 20 db quieter. The DM860H drive is a noisy drive, so noisy I 
had to install miniature quarter round on the keyboard shelf edges to 
keep stuff on it, including the keyboard and mouse. The 860 has very 
very poor step vs amps in coils calibration, so there is no nice quiet 
all steps in the microstep mapping regardless of the current setting. 
The thing you notice most is that a /8 setting actually does 7 steps 
because one of them is way too small.

But thats not solving this measurement problem. Since this $7 kit, needs 
an external clock, it seems like that could be done by siggen, for a 500 
baud baud rate, and a software uart could do the rest. But the packet 
length is unk until I have such a kit in my hot little hands. The 
fleabay adv's don't say.

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 



___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Something to think about re the hack-a-day tool changer

[andy pugh]

On Wed, 17 Apr 2019 at 06:57, Gene Heskett  wrote:

> Some more along the lines of finding out how much force I can figure on
> as being available to tighten or loosen the collet nut. I can get, for
> under a tenner, 4 ea 50Kg load cells and a processor board that converts
> the very low level diffs of a wheatstone bridge

Why not set your torque wrench to the correct torque for your nuts and
then see how far down the handle you have to be to trip the wrench?

That tells you what radius of action you need to get the right torque.

There is no question that your machine can produce the torque, given
the classical "sufficiently long lever"

-- 
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


___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Something to think about re the hack-a-day tool changer

[Gene Heskett]

On Wednesday 17 April 2019 01:48:06 Chris Albertson wrote:

> I think you could eliminate the load cell and simply measure motor
> current.  Given that the motor is locked at zero RPM, the torque would
> be a function of current.  Once you calibrate current to torque it
> will not change.   You still need to get this data into a computer but
> at least you don't need to build a mechanical widget, just a hall
> effect sensor on more lead wire.
>
> One thing is to be VERY careful when powering a stalled motor s you
> can burn it up very easy.   Even if you do add the load cell be sure
> to measure and limit the current and keep it will under the limit. 
> Then you need to cut the power quickly

Ahh, Chris, this is not a servo, I am talking about the XY motors, which 
are steppers and not particularly strong ones.  They are effectively 
always stalled and the current used is set by the dipswitches on the 
driver.  So generally speaking, measureing the current is meaningless to 
a stepper that has remained locked to the steps issued.
>
> Coincides or not but just a second before reading your post I had
> typed the line below to define where I'm connecting a current sensor
> to an SMT32F411 chip pin PA4.  So I thought this could apply to your
> project too.
>
> #define MM_CS1  PA4   // A2 Current Sense MUST LIMIT VOLTS TO 3V3
>
> One more idea that is even more simple.  I assume you are measuring
> torque so you know the nut is on hard enough that it will not come
> loose. Rather then a load cell why not just a simple coil spring and a
> microswitch.  It will always require the same amount of force to
> compress the spring and then the switch is tripped and yu can back off
>   This is basically a load cell with one-bit resolution.

That, with some monkey business incorporating the switch into the sliding 
wrench channel that is pushed out to engage the double flats on the 
chuck shank, might be doable, but the spring to absorb the torque would 
have to be 50x stronger, and if not restrained properly would prevent 
the wrench from being moved out to grab the flats, or retracting it to 
free the spindle. This wrench must be free to move in and out quickly as 
its used only the break the nut loose, then retracted and the motor spun 
yo finish the loosen and drop the tool or tighten to about finger tight, 
prior to using the xy motors, driven in an arc to match the radius the 
sockets are mounted on the carousel disk. This power from the xy motors 
is what breaks the nut loose or cinches it up to hold the tool while 
cutting. Spindle power is not applied during that circular move as the 
retractable wrench has it immobilized anyway, so the the nut is driven 
tight, or backed off depending on which direction the nut and carousel 
is being rotated by the xy motors.  I'd be forced to use an air cylinder 
to get the needed force to move the wrench. Could be done, but not in 
the physical space available since it would take at least a 1"bore x 
3/4"stroke double action cylinder. I expect there will need to be a 
pause, and the spindle motor pulsed to turn until the wrench engages the 
flats, then pulsed the other way to make sure its fully engaged. How to 
do that while the nut is still tight, is another problem yet to be 
solved if the gator style socket isn't used. I don't know why, but I've 
an aversion to that as I have a set of those and they work but are a 
pain in the ass to place over the hex properly. The set I have do not 
have retractable pins like the gators either.  Different designs maybe? 
Probably.

If you've not watched the videos on hackaday, that would explain it at 
least as well as I can.

> On Tue, Apr 16, 2019 at 9:57 PM Gene Heskett  
wrote:
> > On Tuesday 16 April 2019 14:34:15 Gene Heskett wrote:
> > > Greetings all;
> > >
> > > While adding a couple timedelays and a couple or2's to my hal
> > > file, basically to prove that I can pulse the spindle for the
> > > nominally 200 msecs used to spin the nut on/off to release the
> > > collet and supposedly drop the tool, or just drop it all, I came
> > > up against a designed in problem.  This vfd has a pid in it that
> > > will no doubt wind up if the speed command is present for any
> > > great amount of time, so its possible I might have to cobble up a
> > > set of PID params to essentially disable that. Because I think I
> > > am going to need to have the speed present at the VI1 input before
> > > I enable the fwd or reverse commands for the timed 200 millisecs
> > > to drop or pick up the tool during the change. The wind up in that
> > > case might be advantageous in that it might jerk the nut loose, or
> > > tighten it tighter before the carousel puts the real tighten on
> > > it. But likely difficult to control too.
> > >
> > > But I find its not possible to send this speed request before the
> > > run/dir signals. I can mux2 the spindle speed inputs to the 7i76,
> > > and have the as yet unwritten tool changer code issue the speed
> > >