Re: [Emc-users] Possible pncconf error?

2016-09-26 Thread Dewey Garrett 

> check_config: ERROR
>  (Using identity kinematics: trivkins)
>  [JOINT_1]MIN_LIMIT > [AXIS_Z]MIN_LIMIT (-200.0 > -1e99)
>  [JOINT_1]MAX_LIMIT < [AXIS_Z]MAX_LIMIT (800.0 < 1e99)

I believe this is due to a bug in pncconf dating from its
conversion for joints_axes.  For XZ lathe configs, an [AXIS_Y]
section was created instead of the required [AXIS_Z].  With no
valid [AXIS_Z] section, system default values for MIN_LIMIT,
MAX_LIMIT are used leading to an error when the config is
checked by check_config() in the linuxcnc script that looks
for inconsistencies in joints/axes limits for machines using
known identity kinematics (trivkins)..

I pushed a fix to the master branch (lightly tested):

http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=commitdiff;h=a055cb10e8d12595c9a004e299fe8a36fdad0b17
-- 
Dewey Garrett


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

Re: [Emc-users] Possible pncconf error?

2016-09-26 Thread Gene Heskett 
On Monday 26 September 2016 22:49:15 Gene Heskett wrote:

> Greetings all;
>
> I gave up on trying to just copy my TLM's config over to the Sheldons
> computer, mainly because I've not yet installed home switches and
> such.
>
> But I need to be able to exercise this BoB, and find all my signals if
> I can. Something I haven't had a lot of luck finding because mach uses
> totally different parport pin assignments.  But the board still has 17
> available pins, 11 outs and 5 in's, so I figure if I tickle it and
> watch with a scope, I ought to be able to find them, right?
>
> So I fired up pncconf and built me a bare bones config based on metric
> measurements. It has one pwmgen for feeding the vfd. Only one
> direction ATM, but I can sort that by grabbing pin 17 at some point.
> NBD to me.
>
> However, when I attempt to run this config, I get konsole messages:
> =
> check_config: ERROR
>   (Using identity kinematics: trivkins)
>   [JOINT_1]MIN_LIMIT > [AXIS_Z]MIN_LIMIT (-200.0 > -1e99)
>   [JOINT_1]MAX_LIMIT < [AXIS_Z]MAX_LIMIT (800.0 < 1e99)
>
> check_config validation failed
> LinuxCNC terminated with an error.  You can find more information in
> the log:
> /home/gene/linuxcnc_debug.txt (its empty)
> and
> /home/gene/linuxcnc_print.txt (the usual blather that is zero
> info) 
> So I fired up geany and added what I think are sane limits in both
> axis's, and theres an entry in each axis AND each joint (seems one
> ought to sufficient, and I suppose once I check the syntax, and add it
> to this hal file, the other copy can be removed).
>
> Zero difference.
>
> Fire up geany to look at the syntax used to bring that data into hal.
>
> No such critter, totally missing from the pncconf generated hal file. 
> So hal presumes everything is -1e99 from the message above.
>
> I think this is a pncconf bug. Can we call the Orkin Man? :-)
>
No, not an obvious one at least. TLM's config does not contain that 
either.  And it works, works well.  So big old dummy did something 
wrong.  Anybody got a clue-bat?

Thanks all.

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] Possible pncconf error?

2016-09-26 Thread Gene Heskett 
Greetings all;

I gave up on trying to just copy my TLM's config over to the Sheldons 
computer, mainly because I've not yet installed home switches and such.

But I need to be able to exercise this BoB, and find all my signals if I 
can. Something I haven't had a lot of luck finding because mach uses 
totally different parport pin assignments.  But the board still has 17 
available pins, 11 outs and 5 in's, so I figure if I tickle it and watch 
with a scope, I ought to be able to find them, right?

So I fired up pncconf and built me a bare bones config based on metric 
measurements. It has one pwmgen for feeding the vfd. Only one direction 
ATM, but I can sort that by grabbing pin 17 at some point. NBD to me.

However, when I attempt to run this config, I get konsole messages:
=
check_config: ERROR
  (Using identity kinematics: trivkins)
  [JOINT_1]MIN_LIMIT > [AXIS_Z]MIN_LIMIT (-200.0 > -1e99)
  [JOINT_1]MAX_LIMIT < [AXIS_Z]MAX_LIMIT (800.0 < 1e99)

check_config validation failed
LinuxCNC terminated with an error.  You can find more information in the 
log:
/home/gene/linuxcnc_debug.txt (its empty)
and
/home/gene/linuxcnc_print.txt (the usual blather that is zero info)

So I fired up geany and added what I think are sane limits in both 
axis's, and theres an entry in each axis AND each joint (seems one ought 
to sufficient, and I suppose once I check the syntax, and add it to this 
hal file, the other copy can be removed).

Zero difference.

Fire up geany to look at the syntax used to bring that data into hal.

No such critter, totally missing from the pncconf generated hal file.  So 
hal presumes everything is -1e99 from the message above.

I think this is a pncconf bug. Can we call the Orkin Man? :-)

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] Rotary table indexer project

2016-09-26 Thread Chris Albertson 
I read the newspaper stories.  They are accurate but only talk about
package delivery using the normal post office reciprocity agreements that
all post offices have with each other.Some packages are sent this way
but if we are talking about small padded envelopes and boxes that fit on
your hand from eBay resellers then they typically use the e-packet system.
The packet system is much different from the "normal" system.

How would I know this.  Just go on eBay can buy small parts and you see the
shipping options.   They can send it normal post or FedEx or e-packet and
when you see the difference in price most will select e-packet.  Most small
packages are sent this way.

The articles you liked to were correct but they just did not talk about
this.   The media does this a LOT, they report on something that is a small
exception and leave off the normal case, because "normal" is not news.

Another example is the US southern border.  The media would have you think
all the people who are here undocumented walked across the border.   But a
little math:  12M people came here over about 30 years.  So roughly 400,000
per year.  or about 1,100 people walking over the desert every DAY. That's
nuts.  Its thousands per YEAR not per DAY. Only the most ignorant and poor
walked.  It's clear that most did not walk.  Truth is they took airplanes
and drove in cars, mostly airplanes.   But again the normal case is not
news.

The news reports news,  They will rarely tell you the average woman has 2.1
kids but the one odd-ball case of that woman with 8 kids all at once stayed
on the news for months.

What this e-packet is, it's an attempt by China, the USPS and eBay to
invent a more efficient and cheaper system.   The USPS wanted their mail
volume up and their costs down, eBay wants to close more sales and China is
looking for economic activity.   Its a win all around.

For the Chinese seller it works just like it works for me here where I sell
on eBay.  I click a link and print a pre-filled out and pre-paid shipping
table, slap it on the box and place it out for pickup.  The US Post office
gives me a discount.  The same in China except he prints a label that has
bar codes that "work" in both China and in the US. Data enters the USPO
tracking system as soon as the label is printed I see the tracking number.
The Chinese bar codes to show payment, the US based tracking bar codes and
the customs declaration are ALL on the label the Chinese seller prints.
That get sorted by US zip in China and bulk shipped.  Typically the FIRST
US based scan I see was done by the US mailman when he dropped it at the
house.  This means he got carrier route presorted mail that did not go
through US mail sorting equipment.  USPO and China Post share tracking
data, so I can see the China Post scans.

One Catch:  This ONLY works for package sold over high volume on-line
e-commerce because they can capture the data at point of sale.  They know
the weight and both postal codes and the value and have ability to broker
the money and make large batch payments saving the post office from
handling a few million $1 transactions.  eBay can offer to do all this with
no extra work any anyone because they have to capture this data anyways so
it was just a matter on linking the USPO, China Post and eBay computers
together.

But, some small sellers have to go the old fashion way that your newspaper
article talks about.  They walk the package to China Post, pay for stamps
and depend on the US/China reciprocal agreement.  EVERY country looses
money in reciprocal shipments but in theory they all agree to loose because
it's better then trying to do the accounting.   But it's need been fair.
The country with the highest number of inbound packs looses the most.

On Mon, Sep 26, 2016 at 9:36 AM, Dave Cole  wrote:

> Chris,
>
>  >>It pushes some of the
>
> manual work, except the "last mile delivery" onto the sender.
>
> Where do you get your e-packet info from?  That's not accurate.
>
> https://www.washingtonpost.com/news/storyline/wp/2014/09/
> 12/the-postal-service-is-losing-millions-a-year-to-
> help-you-buy-cheap-stuff-from-china/
>
> http://www.practicalecommerce.com/columns/hand-painted-
> ecommerce/86294-China-based-merchants-ship-to-U-S-for-free
>
> Dave
>
> On 9/26/2016 6:09 AM, Chris Albertson wrote:
> > Actually it is more sophisticated than that.  The e-packet delivery
> system
> > is a deal between China and the US.   The packages are collected and
> sorted
> > in china by US zip code and delivered to the final post office.  The US
> > post office skips the normal sorting and shipping and saves some money.
> > Kind of the way bulk mail is sent.   Also the US post office saves money
> > because the tracking information data entry is done in China and sent to
> > the US postal computer system electronically.   It pushes some of the
> > manual work, except the "last mile delivery" onto the sender. They
> also
> > send the packages

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Chris Albertson 
On Mon, Sep 26, 2016 at 12:24 PM, Greg Bentzinger 
wrote:

> Chris;
>
> I only would request B-Directional as a time saver.  I had a HAAS 5C
> indexer which used stepper motors and it was reasonably fast - even so,
> indexing often was the largest portion of total program run time. For small
> indexers that most of us will make from small rotary tables, they will
> likely be driven by low end steppers. These cheap rotary tables are also
> notorious for having tight spots because the gear was not mounted perfectly
> concentric. As the home builder/ converter we have to go to the lowest
> common denominator and keep the max step rate down to where the motor has
> excess torque at the tightest area.


It's good to discuss this here because I get ideas I'd not think of on my
own.  You are right about there being a tighter spot in one place.   But
isn't this 100% repeatable?  I don't have to go slow over the whole circle
just because one spot requires it.  Don't need to use lowest common
denominator speed.  Should be able to make the maximum speed a function of
the table position.

Actually my table is pretty good this way, light finger pressure on the
crack works all the way around.

If Linux CNC has a "Cycle Start" signal then I don't need to let LinuxCNC
actually run the stepper motor.  It can simply say "go to the next step"
and my controller can push the motor up to the physical limits of the
motor/table system.   The cost to add "smarts" to a stepper driver is about
1/10th the cost of the typical stepper driver.  The small computer costs
about $3.50.

Yes I know this feature could also go inside the FPGA on the Mesa card (if
there is room left)
--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Gregg Eshelman 
You want your order to be the final one on the final pallet going into the 
shipment, not the first order on the first pallet.

 
  From: Chris Albertson 
 To: Enhanced Machine Controller (EMC)  
 Sent: Monday, September 26, 2016 4:09 AM
 Subject: Re: [Emc-users] Rotary table indexer project
   
Actually it is more sophisticated than that.  The e-packet delivery system
is a deal between China and the US.  The packages are collected and sorted
in china by US zip code and delivered to the final post office.  The US
post office skips the normal sorting and shipping and saves some money.
Kind of the way bulk mail is sent.  Also the US post office saves money
because the tracking information data entry is done in China and sent to
the US postal computer system electronically.  It pushes some of the
manual work, except the "last mile delivery" onto the sender.    They also
send the packages in batches so that many packages are moved at once.  Like
palletized fright.

The down side is that it can take a few days, maybe two weeks for a batch
to accumulate in China before it ships.  Sometimes I luck out and get a
package in three or four days, I guess because my package was the last one
to fill the pallet.

it's air freight, not on a ship.

There is a private company that does something like this in India.  They
place first class US domestic postage on packages then place the packages
on pallets and ship them to the US where the pallets are busted up and
dropped in the US mail.  So Indian companies can ship to the US for little
more than the US domestic rate

So really they CAN air freight a small padded envelope from China to
California for a few cents because thousands of these are stuffed into one
box
   
 
--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Greg Bentzinger 
Chris;

I only would request B-Directional as a time saver.  I had a HAAS 5C indexer 
which used stepper motors and it was reasonably fast - even so, indexing often 
was the largest portion of total program run time. For small indexers that most 
of us will make from small rotary tables, they will likely be driven by low end 
steppers. These cheap rotary tables are also notorious for having tight spots 
because the gear was not mounted perfectly concentric. As the home builder/ 
converter we have to go to the lowest common denominator and keep the max step 
rate down to where the motor has excess torque at the tightest area.  Moving 
Negative 60 degrees is alot faster than moving 300 degrees. This is fine for a 
hobby user, but I earn a major portion of my income from production machine 
jobs. Some say time is money, but more importantly you can never buy more time. 
I have worked 20hr days to meet delivery deadlines and God as my witness you 
can look at me now and know I really needed that beauty slee
 p!


The remote cycle start and cycle ready/complete are function options on most 
commercial CNC machines - how they are used is up to the machine integrator. 
Same in LCNC.There is no specific LCNC code for this, but there are definable M 
codes and depending on your hardware will determine how you would configure 
this. LCNC would have no "Knowledge" of indexer position, only if its ready or 
moving.  I guess you could add another output for home that could be used to 
verify that all moves in the cycle had completed properly. ( The sum of all 
incremental moves would bring indexer back to its starting point. )

You are also correct about the mechanics - every table seems to have its own 
individual custom requirements. A company I worked for bought 5 "identical" 
units and was intending to use a single servo to drive them all. Each one had 
bolt holes in different locations. I swear I think most Chicom machines are 
just free hand drilled with a drill press without even doing hole position 
layout.

Greg


From: Chris Albertson 
To: Greg Bentzinger ; Enhanced Machine Controller (EMC) 
 
Sent: Monday, September 26, 2016 4:25 AM
Subject: Re: [Emc-users] Rotary table indexer project



This is the kind of ideas I was looking for.  Two questions:

(1) What does bi-directional get you that can't be done by going all the way 
around.  Maybe it takes longer to rotate 360+ degrees?Is there any other 
reason?  Technically it is very easy to run bidirectional but I could not see a 
reason.   

(2) I did not know about "remote cycle start input and a cycle complete" in 
LinuxCNC.  Got a pointer to the documents?   If it could be done with a TTL 
level pins then adding this is free.  A serial interface is also easy to add.  
Does LinuxCNC have a way to send data?

The hardest part of building this is the mechanics.   Cutting metal has to be 
custom engineered for each rotary table. I can host the software and 
schematics on GitHub.


On Mon, Sep 26, 2016 at 2:35 AM, Greg Bentzinger  wrote:

Chris;
>
>I would add a remote cycle start input and a cycle complete output that could 
>be optionally hooked into LCNC.
>
>
>Likewise I would allow bi-directional positioning moves. Just add a small 
>overshoot value and and feed back into position to maintain consistent preload 
>in one direction.
>
>I would be very interested in duplicating your Arduino indexer controller as I 
>often have parts I need to cut a hex drive feature on.
>
>Thanks;
>
>Greg
>
>
>-- -- 
>--
>__ _
>Emc-users mailing list
>Emc-users@lists.sourceforge. net
>https://lists.sourceforge.net/ lists/listinfo/emc-users
>


-- 


Chris Albertson
Redondo Beach, California

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

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Dave Cole 
Chris,

 >>It pushes some of the

manual work, except the "last mile delivery" onto the sender.

Where do you get your e-packet info from?  That's not accurate.

https://www.washingtonpost.com/news/storyline/wp/2014/09/12/the-postal-service-is-losing-millions-a-year-to-help-you-buy-cheap-stuff-from-china/

http://www.practicalecommerce.com/columns/hand-painted-ecommerce/86294-China-based-merchants-ship-to-U-S-for-free

Dave

On 9/26/2016 6:09 AM, Chris Albertson wrote:
> Actually it is more sophisticated than that.  The e-packet delivery system
> is a deal between China and the US.   The packages are collected and sorted
> in china by US zip code and delivered to the final post office.  The US
> post office skips the normal sorting and shipping and saves some money.
> Kind of the way bulk mail is sent.   Also the US post office saves money
> because the tracking information data entry is done in China and sent to
> the US postal computer system electronically.   It pushes some of the
> manual work, except the "last mile delivery" onto the sender. They also
> send the packages in batches so that many packages are moved at once.  Like
> palletized fright.
>
> The down side is that it can take a few days, maybe two weeks for a batch
> to accumulate in China before it ships.   Sometimes I luck out and get a
> package in three or four days, I guess because my package was the last one
> to fill the pallet.
>
> it's air freight, not on a ship.
>
> There is a private company that does something like this in India.  They
> place first class US domestic postage on packages then place the packages
> on pallets and ship them to the US where the pallets are busted up and
> dropped in the US mail.  So Indian companies can ship to the US for little
> more than the US domestic rate
>
> So really they CAN air freight a small padded envelope from China to
> California for a few cents because thousands of these are stuffed into one
> box
>
>
> So
>
> On Mon, Sep 26, 2016 at 1:50 AM, andy pugh  wrote:
>
>> On 26 September 2016 at 09:20, Chris Albertson 
>> wrote:
>>
>>> I can't see how the postage alone is that low.
>>
>> Because they only need to get it on a ship, and then it is delivered for
>> free (at a loss) by your national mail delivery co.
>> It's meant to be a mutual arrangement, but it is far from fair at the
>> moment.
>>
>> --
>> 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
>>
>
>

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

Re: [Emc-users] Strange Z value In DRO

2016-09-26 Thread Jim Craig 
On 9/26/2016 8:20 AM, Chris Radek wrote:
> On Mon, Sep 26, 2016 at 07:54:30AM -0500, Jim Craig wrote:
>> This is all with version 2.7.4.
> Are you sure of the version?  There are some problems in 2.7.5-2.7.6
> that seem like they could be related to this behavior (fixed now in
> 2.7.7) but I would not expect 2.7.4 to have problems with losing
> offsets.
>
> If you could figure out how to make this happen in a sim config and
> give us instructions for reproducing it, especially if that's on
> 2.7.7, it would be really helpful.
>
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
Chris,

It is with 2.7.4. We have not updated that machine in a while as it is 
not connected to the internet. Actually I have not updated any of my 
machines from 2.7.4 yet. . .

I don't have time to test with 2.7.7 right now. Maybe later this fall. 
We are farming now so not much time to play with CNC machines. : (

Thanks,

Jim


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

Re: [Emc-users] Strange Z value In DRO

2016-09-26 Thread Chris Radek 
On Mon, Sep 26, 2016 at 07:54:30AM -0500, Jim Craig wrote:
> 
> This is all with version 2.7.4. 

Are you sure of the version?  There are some problems in 2.7.5-2.7.6
that seem like they could be related to this behavior (fixed now in
2.7.7) but I would not expect 2.7.4 to have problems with losing
offsets.

If you could figure out how to make this happen in a sim config and
give us instructions for reproducing it, especially if that's on
2.7.7, it would be really helpful.

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

[Emc-users] Strange Z value In DRO

2016-09-26 Thread Jim Craig 
While working with the plasma cutter this weekend we noticed strange 
behavior on the DRO overlay on the preview plot in Axis.

When attempting to run a program we were getting -z limit switch trips 
due to needing more debounce on the switch. This switch is used for 
touch off probing and for -z limit when not doing a probing move.

So while debugging the debounce issue we noticed some strange issues 
with the Z axis readings after the limit switch was tripped. Once the 
trip would occur we would do a manual Z axis touchoff to get a 
preliminary rough Z zero. However once the touchoff button was pushed 
the value shown on the DRO overlay on the preview plot would show some 
random number from 0.012 to 0.240 inches. There was no offset applied to 
the touchoff and there are no tool length offsets applied.

The stranger thing is that if you switch to the DRO tab then the value 
shown for the Z axis is 0. every time. To make the preview overlay 
dro show the 0. value I would have to manually change to G55 then 
back to G54 and the value would show properly.

This is all with version 2.7.4. Note that this machine is operating with 
stepper motors and no encoder feedback. Also note that the strange value 
would not happen every time but ~90% of the time.

Let me know what you think. Is this a bug. It was very repeatable on 
this machine but I am not sure if it is repeatable on other machines. I 
have not had time to try it on any other machine yet.

Thanks,

Jim


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

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Dave Caroline 
Too me this is nuts, just use linucnc to index, no need for add on
code and hardware, goes both ways, coordinated moves (hobbing, helical
cuts and radii), indexing is trivial in an Ocode routine.
When I first set up my machine it was for cutting gears, using
linuxcnc got rid of all the errors of the manual methods we were using
in the clockworks. I started with only 2 axes under cnc the rest
manual (rotary and cutter pass)

Dave Caroline

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

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Chris Albertson 
This is the kind of ideas I was looking for.  Two questions:

(1) What does bi-directional get you that can't be done by going all the
way around.  Maybe it takes longer to rotate 360+ degrees?Is there any
other reason?  Technically it is very easy to run bidirectional but I could
not see a reason.

(2) I did not know about "remote cycle start input and a cycle complete" in
LinuxCNC.  Got a pointer to the documents?   If it could be done with a TTL
level pins then adding this is free.  A serial interface is also easy to
add.  Does LinuxCNC have a way to send data?

The hardest part of building this is the mechanics.   Cutting metal has to
be custom engineered for each rotary table. I can host the software
and schematics on GitHub.

On Mon, Sep 26, 2016 at 2:35 AM, Greg Bentzinger 
wrote:

> Chris;
>
> I would add a remote cycle start input and a cycle complete output that
> could be optionally hooked into LCNC.
>
>
> Likewise I would allow bi-directional positioning moves. Just add a small
> overshoot value and and feed back into position to maintain consistent
> preload in one direction.
>
> I would be very interested in duplicating your Arduino indexer controller
> as I often have parts I need to cut a hex drive feature on.
>
> Thanks;
>
> Greg
>
> 
> --
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>



-- 

Chris Albertson
Redondo Beach, California
--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Chris Albertson 
Actually it is more sophisticated than that.  The e-packet delivery system
is a deal between China and the US.   The packages are collected and sorted
in china by US zip code and delivered to the final post office.  The US
post office skips the normal sorting and shipping and saves some money.
Kind of the way bulk mail is sent.   Also the US post office saves money
because the tracking information data entry is done in China and sent to
the US postal computer system electronically.   It pushes some of the
manual work, except the "last mile delivery" onto the sender. They also
send the packages in batches so that many packages are moved at once.  Like
palletized fright.

The down side is that it can take a few days, maybe two weeks for a batch
to accumulate in China before it ships.   Sometimes I luck out and get a
package in three or four days, I guess because my package was the last one
to fill the pallet.

it's air freight, not on a ship.

There is a private company that does something like this in India.  They
place first class US domestic postage on packages then place the packages
on pallets and ship them to the US where the pallets are busted up and
dropped in the US mail.  So Indian companies can ship to the US for little
more than the US domestic rate

So really they CAN air freight a small padded envelope from China to
California for a few cents because thousands of these are stuffed into one
box


So

On Mon, Sep 26, 2016 at 1:50 AM, andy pugh  wrote:

> On 26 September 2016 at 09:20, Chris Albertson 
> wrote:
>
> > I can't see how the postage alone is that low.
>
>
> Because they only need to get it on a ship, and then it is delivered for
> free (at a loss) by your national mail delivery co.
> It's meant to be a mutual arrangement, but it is far from fair at the
> moment.
>
> --
> 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
>



-- 

Chris Albertson
Redondo Beach, California
--
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Greg Bentzinger 
Chris;

I would add a remote cycle start input and a cycle complete output that could 
be optionally hooked into LCNC.


Likewise I would allow bi-directional positioning moves. Just add a small 
overshoot value and and feed back into position to maintain consistent preload 
in one direction.

I would be very interested in duplicating your Arduino indexer controller as I 
often have parts I need to cut a hex drive feature on.

Thanks;

Greg

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

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread andy pugh 
On 26 September 2016 at 09:20, Chris Albertson 
wrote:

>
> I'll connect it to Linux CNC later.  But the PC box is huge and not so
> portable.  The indexer controller can be 1/3rd the size of this 4" table.


My rotary axes connect to the controller with a hybrid D-sub. There isn't
any separate controller. If I am not using the rotary then it just isn't
connected.
I wrote a post for cnczone on the subject
http://www.cnczone.com/forums/uncategorised-metalworking-machines/172314-cnc-post1227167.html#post1227167

-- 
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] Rotary table indexer project

2016-09-26 Thread andy pugh 
On 26 September 2016 at 09:20, Chris Albertson 
wrote:

> I can't see how the postage alone is that low.


Because they only need to get it on a ship, and then it is delivered for
free (at a loss) by your national mail delivery co.
It's meant to be a mutual arrangement, but it is far from fair at the
moment.

-- 
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] Rotary table indexer project

2016-09-26 Thread Chris Albertson 
On Sun, Sep 25, 2016 at 11:28 PM, Marcus Bowman <
marcus.bow...@visible.eclipse.co.uk> wrote:

>
> On 26 Sep 2016, at 04:45, Chris Albertson wrote:
>
> > I've started work on a motorized indexer for my rotary table.
>
> Like DivisionMaster?
>
> >   At this
> > point I am not calling this a fourth axis.  The way this will be used is
> to
> > (1) rotate the table, (2) clamp it down, (3) perform the drill or mill
> > operation (4) unlock) then if not finished go to #1.
> >
> > A fourth axis would required more mechanical precision.  I plan to drive
> > the table only on one direction and clamp down the clamps when cutting.
> > It should work well for making bolt circles and the like.
> >
>
> Limiting rotation to one direction is a huge limitation on what can be
> achieved with a table like this. Why not drive it with LinuxCNC and address
> any problems of backlash through either (or both) mechanics and backlash
> settings within LinuxCNC? Bolt circles and simple division are fine, but
> bi-directional motion allows full machining.
>

I can figure out backlash without a PC.  In fact I can do it automatically
the way inkjet printers do with a self calibration on power up.  Run a
metal tab past an optical sensor in both directions.   For now this is an
indexer.   Unlike a mill table, I can get to any location by driving in
just one direction.

I'll connect it to Linux CNC later.  But the PC box is huge and not so
portable.  The indexer controller can be 1/3rd the size of this 4" table.

>
> > Later I can see if my low-cost table can be used for more.
> >
> Better to address that during the design stage, I think.
>

I'm stuck with this rotary table.   It might be good enough that it can
move while the mill is cutting metal.  It's a small Chinese table  The
design of the table is fixed.

>
> > The project adds a stepper motor to replace the crank because I am not
> good
> > at counting or finding holes on an index plate.   The motor connects via
> > direct drive to a a worm that drives a 72 tooth wheel.  I can micro step
> at
> > about 8:1.  This gives me about 0.0001" of resolution at the edge of the
> > table and a lot more near the enter.
>
> Is resolution on a rotary table not better defined in terms of angles
> (degrees or radians)?
>

I think angular resolution might be a derived requirement.  The base
requirement is the tolerance on the hole placement, then you compute what
angular resolution you need to get the hole in the correct spot.   At least
this is the way I think about it.

Even with a gear.  You might look at one that is poorly made and say " that
one tooth is .010 to wide".  You's not say "the left side of that tooth is
radially displaced 17 seconds of arc".   Even if the two happen to be
equivalent.

But in any case, My goal was to have the table's mechanics be the weak link.


> >
> > For your amusement here is a photo of the controller electronics,
> computer
> > and all working on my electronics bench.   The user interface is a
> rotrary
> > knob that you can "push to select" and there is a small LCD for a
> display.
> > https://www.dropbox.com/...ElectBenchTest01.jpg
> >  dl=0>
>

Looks like a non-HTML aware mail reader hacked the link.  There is only one
link, the first is the text to be displayed to the reader, the second is
the actual link.  The bottom one SHOULD show the photo to anyone.


> >
> >
> Depending on the inertia of your table, you would need ramped acceleration
> and deceleration at the beginning and end of each set of steps.
>

Yes, thank you.   I have a motor setting on my bench and it jumps when I go
abruptly from zero to 10,000 steps per second.  So I find that even a bare
motor with nothing connected to its shaft needs to ramp up to speed.

What I do, under the hood, where the user does not see is give an end
point, maximum speed and a maximum acceleration.   The motor only gets up
to full speed if it moves a certain minimum distance.So I could add the
requirement "don't miss any steps".   But I only wanted to list user facing
features

>
> > PS.   This is only tangentially related to LinuxCNC, connection to that
> > comes later.   Another project I want to do is to add a replica of the
> > physical hand wheels that go on a manual machine tool.  It would be self
> > contained not using a PC.  Likely just three wheels with one of those
> small
> > LCDs above each one to serve as a digital readout.   Basically what you
> see
> > in that photo times three.
>
> A more practical solution might be to use a single wheel and three
> buttons. Or a ShuttlePro or Shuttle Express. The Express will be cheaper
> than the parts needed for your plan involving 3 or everything. Driver for
> LinuxCNC is available through the main LinuxCNC site.
>

I just looked, the ShuttleXpress is about $50 and it lacks an LCD display.
  I'm into about $8 in parts for a single axis control with a two line text
display. (Not

Re: [Emc-users] Anybody have an idea how...

2016-09-26 Thread Erik Christiansen 
On 25.09.16 13:11, Danny Miller wrote:
> Some of our folks built a metal kiln and we tried to melt some aluminum 
> swarf the neighboring metalshop discarded, at least what we thought was 
> aluminum (no confirmation, but it wasn't steel). We got almost no metal, 
> just inches of dross.  No inert gas or flux there, in fact the metal was 
> exposed to combustion gases.

It is some years back now, but I made a number of Al castings, using
scrap, and had no difficulty filling a 2 litre crucible with liquid
metal. That produced about a centimetre (call it 3/8") of dross, but
that much is just surface oxide from the significant proportion of drink
cans which went in. To really bring the dross to the surface, it is
necessary to stir in some flux - then you get a good layer. (I've read
of common table salt being used as flux, but I bought a commercial
product.)

The other thing which should be done is to plunge and stir a degassing
tablet (I just held it in a pair of domestic fire tongs, slightly
rusted, and well carbonised, to minimise iron contamination. You could
whitewash them instead.) That drives off bubbly amounts of dissolved
hydrogen, which will otherwise come out of solution on cooling, causing
cavities in the casting.

And, of course, add a couple of % of copper, early on, to make the
casting machinable. (Who wants a casting made of toffee?) A little bit
of zinc helps to make the alloy age-hardening, I found, and those
castings are still sound 40 years later. (The zinc I used came from
scraps of pot-metal, or "die-cast" as it's called down under. So all I
know is that 1% or 2% of Zn, or something like that, went in.)
(If you have a tub of accumulated brass swarf, not enough to make a
useful brass casting, then it is a source of both Cu and Zn, ideal for
Al alloying, I figure. Mine is exclusively hard brass, which is around
60:40, IIUC. The ductile stuff is 70:30, I think. Also near enough.)

I would expect a good layer of dross from swarf, as there's a lot of
surface area. And there must be some further loss of metal when all that
dross is scraped over the side of the crucible, as it burns as bright as
the sun when it falls onto the charcoal fuel I used.

It is possible that a deep layer of charcoal, and only gentle forced
ventilation from a speed-controlled DC fan, contributed to a reducing
(or at least only mildly oxidising) atmosphere, but I kept the lid on
the crucible except when charging with fresh (PRE-DRIED!) metal.

Erik

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

Re: [Emc-users] Rotary table indexer project

2016-09-26 Thread Marcus Bowman 

On 26 Sep 2016, at 04:45, Chris Albertson wrote:

> I've started work on a motorized indexer for my rotary table.

Like DivisionMaster?

>   At this
> point I am not calling this a fourth axis.  The way this will be used is to
> (1) rotate the table, (2) clamp it down, (3) perform the drill or mill
> operation (4) unlock) then if not finished go to #1.
> 
> A fourth axis would required more mechanical precision.  I plan to drive
> the table only on one direction and clamp down the clamps when cutting.
> It should work well for making bolt circles and the like.
> 

Limiting rotation to one direction is a huge limitation on what can be achieved 
with a table like this. Why not drive it with LinuxCNC and address any problems 
of backlash through either (or both) mechanics and backlash settings within 
LinuxCNC? Bolt circles and simple division are fine, but bi-directional motion 
allows full machining.

> Later I can see if my low-cost table can be used for more.
> 
Better to address that during the design stage, I think.

> The project adds a stepper motor to replace the crank because I am not good
> at counting or finding holes on an index plate.   The motor connects via
> direct drive to a a worm that drives a 72 tooth wheel.  I can micro step at
> about 8:1.  This gives me about 0.0001" of resolution at the edge of the
> table and a lot more near the enter.

Is resolution on a rotary table not better defined in terms of angles (degrees 
or radians)?

> 
> For your amusement here is a photo of the controller electronics, computer
> and all working on my electronics bench.   The user interface is a rotrary
> knob that you can "push to select" and there is a small LCD for a display.
> https://www.dropbox.com/...ElectBenchTest01.jpg
> 
> 
> The little 8-bit computer can drive the stepper at almost 10,000 steps per
> second.  The steps have 105 uSec period with only a few tens of nanoseconds
> variation.  I'm currently using a 12 volt plug-in power cube to power the
> stepper.
> 
> QUESTION:  Before I get further along, within the limits of this hardware
> what features would be good to include?   The memory has tons more room for
> software and many unused pins on the Arduino.  I wonder if I missed any
> useful features?
> 
> 1) selectable units, either degrees/minutes/seconds or fraction of a circle
> 2) A physical button to set the zero point
> 3) A physical "step" button to move to the next point
> 4) can set the step size with rotary dial
> 5) controls lock out automatically when it does not make sense to use them
> to avoid accidental activation
> 6) A fine motion mode where the table moves 1 second of arc per click of
> the rotary dial, for fine tuning the zero point
> 7) the display will show the current position and the step size in
> selectable units
> 
> 
> Is there anything I missed?
> 
Depending on the inertia of your table, you would need ramped acceleration and 
deceleration at the beginning and end of each set of steps.

> PS.   This is only tangentially related to LinuxCNC, connection to that
> comes later.   Another project I want to do is to add a replica of the
> physical hand wheels that go on a manual machine tool.  It would be self
> contained not using a PC.  Likely just three wheels with one of those small
> LCDs above each one to serve as a digital readout.   Basically what you see
> in that photo times three.

A more practical solution might be to use a single wheel and three buttons. Or 
a ShuttlePro or Shuttle Express. The Express will be cheaper than the parts 
needed for your plan involving 3 or everything. Driver for LinuxCNC is 
available through the main LinuxCNC site.

> 
> -- 
> 
> Chris Albertson
> Redondo Beach, California
> --
> ___
> 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