Re: [Emc-users] losing encoder counts

2022-10-06 Thread gene heskett 

On 10/6/22 15:06, John Figie wrote:

A colleague once told me that the harder a bug is to find then the
easier it is to fix. I think there is a lot of truth to that.

John Figie
That is a given John. When a bug is hard to find, you start paying 
attention to every byte, and become one
with the code. And its always a face slapper when you finally recognize 
the exact miss-programmed byte.
I hate to admit its a lesson it took me 70 years to learn the basic 
truth of.



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, 1940)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 



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

Re: [Emc-users] losing encoder counts

2022-10-06 Thread John Figie 
A colleague once told me that the harder a bug is to find then the
easier it is to fix. I think there is a lot of truth to that.

John Figie

On Thu, Oct 6, 2022 at 12:44 PM Thaddeus Waldner  wrote:
>
> The solution to hard problems are always in the peripheral vision.
>
>
> > On Oct 6, 2022, at 12:20 PM, gene heskett  wrote:
> >
> > On 10/5/22 22:56, Jon Elson wrote:
> >> Wow, it gets deeper!
> >>
> >> It is NOT the encoder, or anything in the encoder-reading process.
> >>
> >> I marked the motor shaft, and the motor is returning to the exact same 
> >> position every cycle.
> >>
> >> Well, I tried a different indicator and mount, and the problem was in the 
> >> indicator mount!  I was using an old knock-off of an IndiCol that clamps 
> >> around the spindle nose or end mill holder and has several rods with 
> >> locking thumbnuts on them.  These swivel joints were slipping just a tiny 
> >> bit each time the vise touched the indicator, even though they were PRETTY 
> >> tight!  Ugh, 3 days of going around in circles trying to tell where the 
> >> issue was!
> >>
> >> Jon
> >>
> > Glad you found it, Jon. And don't feel like the lone stranger, I've been 
> > caught out by such stuff lots of times.
> >
> > 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, 1940)
> > If we desire respect for the law, we must first make the law respectable.
> > - Louis D. Brandeis
> > 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


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

Re: [Emc-users] losing encoder counts

2022-10-06 Thread Thaddeus Waldner 
The solution to hard problems are always in the peripheral vision.


> On Oct 6, 2022, at 12:20 PM, gene heskett  wrote:
> 
> On 10/5/22 22:56, Jon Elson wrote:
>> Wow, it gets deeper!
>> 
>> It is NOT the encoder, or anything in the encoder-reading process.
>> 
>> I marked the motor shaft, and the motor is returning to the exact same 
>> position every cycle.
>> 
>> Well, I tried a different indicator and mount, and the problem was in the 
>> indicator mount!  I was using an old knock-off of an IndiCol that clamps 
>> around the spindle nose or end mill holder and has several rods with locking 
>> thumbnuts on them.  These swivel joints were slipping just a tiny bit each 
>> time the vise touched the indicator, even though they were PRETTY tight!  
>> Ugh, 3 days of going around in circles trying to tell where the issue was!
>> 
>> Jon
>> 
> Glad you found it, Jon. And don't feel like the lone stranger, I've been 
> caught out by such stuff lots of times.
> 
> 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, 1940)
> If we desire respect for the law, we must first make the law respectable.
> - Louis D. Brandeis
> 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] losing encoder counts

2022-10-06 Thread gene heskett 

On 10/5/22 22:56, Jon Elson wrote:

Wow, it gets deeper!

It is NOT the encoder, or anything in the encoder-reading process.

I marked the motor shaft, and the motor is returning to the exact same 
position every cycle.


Well, I tried a different indicator and mount, and the problem was in 
the indicator mount!  I was using an old knock-off of an IndiCol that 
clamps around the spindle nose or end mill holder and has several rods 
with locking thumbnuts on them.  These swivel joints were slipping 
just a tiny bit each time the vise touched the indicator, even though 
they were PRETTY tight!  Ugh, 3 days of going around in circles trying 
to tell where the issue was!


Jon

Glad you found it, Jon. And don't feel like the lone stranger, I've been 
caught out by such stuff lots of times.


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, 1940)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 



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

Re: [Emc-users] losing encoder counts

2022-10-06 Thread dave engvall 
This make a good case for dual independent encoders. motor 
shaft/ballscrew, ballscrew/glass scale, steps/glass scale, etc.
Just for grins I tried the hand crank on my well used (Boeing then trade 
school, the auction) defunct tracer mill converter by a Russian engineer 
to cnc for the trade school. 1963 vntage cinci. Comparing the hand crank 
on the X vs a accurite 5 um glass scale they were dead on (of course 
only in one direction) ;-) The Y was a mess so I didn't try it.
It is relatively easy to clamp a glass scale to the bed, dial it in 
parallel to the axis then tie it to the spindle with the spindle power 
carefully disabled and proceed with measurements. Notice I did not say 
much about a Chinese glass scale. Some are good some not so good. You 
pays your money and takes your chances.


Dave

On 10/5/22 7:53 PM, Jon Elson wrote:

Wow, it gets deeper!

It is NOT the encoder, or anything in the encoder-reading process.

I marked the motor shaft, and the motor is returning to the exact same 
position every cycle.


Well, I tried a different indicator and mount, and the problem was in 
the indicator mount!  I was using an old knock-off of an IndiCol that 
clamps around the spindle nose or end mill holder and has several rods 
with locking thumbnuts on them.  These swivel joints were slipping 
just a tiny bit each time the vise touched the indicator, even though 
they were PRETTY tight!  Ugh, 3 days of going around in circles trying 
to tell where the issue was!


Jon



___
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] losing encoder counts

2022-10-05 Thread Sam Sokolik 
Lol..   I remember Chris R was working on rigid tapping at one of the fests
in galesberg..   He was using the mazak which had been converted to
linuxcnc..  (I think it had been worked on over a couple fests. ).  Anyway
- it didn't seem to be tracking the correct pitch..  (Tapping in plastic).
Chris was trying to think what could be wrong with his code and johnk
walked over and listened..  Then said 'could the z axis scale be off?'.  It
was..  I could see that situation cause you to go down a lot of rabbit
holes before thinking about the z axis scaling  (At least that is how I
remember it..). :)

On Wed, Oct 5, 2022, 9:59 PM Jon Elson  wrote:

> Wow, it gets deeper!
>
> It is NOT the encoder, or anything in the encoder-reading
> process.
>
> I marked the motor shaft, and the motor is returning to the
> exact same position every cycle.
>
> Well, I tried a different indicator and mount, and the
> problem was in the indicator mount!  I was using an old
> knock-off of an IndiCol that clamps around the spindle nose
> or end mill holder and has several rods with locking
> thumbnuts on them.  These swivel joints were slipping just a
> tiny bit each time the vise touched the indicator, even
> though they were PRETTY tight!  Ugh, 3 days of going around
> in circles trying to tell where the issue was!
>
> Jon
>
>
>
> ___
> 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] losing encoder counts

2022-10-05 Thread John Dammeyer 
You must be a happy camper other than the lost time.  In my case the DRO was 
telling me things were off.  And of course holes in the wrong place.
John

> -Original Message-
> From: Jon Elson [mailto:el...@pico-systems.com]
> Sent: October-05-22 7:54 PM
> To: emc-users@lists.sourceforge.net
> Subject: Re: [Emc-users] losing encoder counts
> 
> Wow, it gets deeper!
> 
> It is NOT the encoder, or anything in the encoder-reading
> process.
> 
> I marked the motor shaft, and the motor is returning to the
> exact same position every cycle.
> 
> Well, I tried a different indicator and mount, and the
> problem was in the indicator mount!� I was using an old
> knock-off of an IndiCol that clamps around the spindle nose
> or end mill holder and has several rods with locking
> thumbnuts on them.� These swivel joints were slipping just a
> tiny bit each time the vise touched the indicator, even
> though they were PRETTY tight!� Ugh, 3 days of going around
> in circles trying to tell where the issue was!
> 
> Jon
> 
> 
> 
> ___
> 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] losing encoder counts

2022-10-05 Thread Jon Elson 

Wow, it gets deeper!

It is NOT the encoder, or anything in the encoder-reading 
process.


I marked the motor shaft, and the motor is returning to the 
exact same position every cycle.


Well, I tried a different indicator and mount, and the 
problem was in the indicator mount!  I was using an old 
knock-off of an IndiCol that clamps around the spindle nose 
or end mill holder and has several rods with locking 
thumbnuts on them.  These swivel joints were slipping just a 
tiny bit each time the vise touched the indicator, even 
though they were PRETTY tight!  Ugh, 3 days of going around 
in circles trying to tell where the issue was!


Jon



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

Re: [Emc-users] losing encoder counts

2022-10-03 Thread andy pugh 
On Sun, 2 Oct 2022 at 23:51, Jon Elson  wrote:

> I am using my PPMC boards, and I've never
> seen this on my older Bridgeport using the same hardware.

Do you have quadrature error detection in your firmware? ie something
that sets a flag if the speed is non zero but you see an
out-of-sequence transition.

-- 
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, 1912


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

Re: [Emc-users] losing encoder counts

2022-10-03 Thread gene heskett 

On 10/3/22 07:48, Sam Sokolik wrote:

Our 80's vintage matsuura started losing counts when the 5v supply started
to lower..  don't remember how far down..
   Maybe 4.7v?
Wearing my C.E.T. hat, at 4.7 volts, all it can do is crash, and if 
there is a resistively terminated scsi drive
involved, it will get flaky below 4.95 volts.  The culprit there is the 
isolation diode, engineering speced
a schotky, spedificly for its very low fwd drop, and some MBA had a cow 
over the price and substituted
a std si power  diode on the way to production, killing  .7 volts of the 
5 it had.   By-by to any logic 1 noise
margin it might have had on the drawing board with the proper parts 
installed.  Your trivia factoid for the day.

On Sun, Oct 2, 2022, 11:54 PM gene heskett  wrote:


On 10/2/22 22:25, John Figie wrote:

So if you missed an A or a B pulse from the encoder I think you should

see

the position count go in the opposite direction for one count. Another
words, instead of a position sequence of 0,1,2,3,4,5,6,7 you might see
instead 0,1,2,3,2,3, and in addition the time spent in 3 followed by 2
would each be twice as long.  So maybe if you can trigger on a non
monotonic position sequence you can capture the event.

John Figie

On Sun, Oct 2, 2022, 7:49 PM Jon Elson  wrote:


On 10/2/22 19:34, John Dammeyer wrote:

Oh boy does that bring back memories.
I started with the US Digital Encoders on the DC Servos.  According to

the designer of the PIC upgrade board for the HP_UHU Servo drive the US
Digital were crap and caused nothing but problems.

I switched to CUI and it appeared to be better.  Until one day circle

milled inside a piece was off center.  See photo.

That’s when I started doing what you did and discovered that at very

slow speeds I could return to the same 0 position but faster moves lost
steps.  The error happened in only one direction.

Eventually I discovered that if I swapped the motor+encoder with the Y

axis the problem moved to the Y so it wasn’t the HP_UHU drive.  Further
research showed the X axis motor winding resistance was almost double

that

of the working Y axis motor.

It's possible I might have been able to fix the problem by doubling the

current setting on the motor.  They were bought at the same time and

were

supposed to be identical.

I don't know why the HP_UHU servo drive didn't catch the position loss.

Like yours, the encoder signals looked nice.The CUI encoders seem

to be

high

One other point.  I switched to the Bergerda AC Servo drives which to

date have been great.  Talking to Donald Chen the sales guy at Bergerda

he

mentioned that their encoders are Japanese and more expensive than ones
used on a lot of their competitors.  After that conversation I took one
apart and checked part numbers.  Sure enough, Japanese design made in

China.

So try different encoders.

Yes, I did see some really strange signals on the scope
before. I tore the whole motor/encoder set apart and cleaned
the inside of the encoder.  But, optical components in the
encoder are held in place with glue, 39 year old glue!  I'm
having doubts that everything is still properly aligned in
there.

So, I'm leaning toward replacing the encoders with newer
high-end encoders with higher line count.  But, that will
require machining adapter plates and making sure the
encoders are well-centered.

Thanks,

Jon

Jon and John, I'd vote for a lack of good "star' grounding. Somehow the
encoder is picking up noise from the
motors reversal. So you are getting an extra noise pulse into the
encoder wireng and its going to be direction
switch related. A good star ground, and all that just goes away. Even
with unshielded cables.

The whole point of the star ground is that this common ground may not be
quiet. But everything is
referencing the same bolt.  that bolt may not be grounded to earth, and
can be bouncing around 500 volts, but its
carrying ALL the logic with it and the logic is only sensitive to the
logic level between the wire from the encoder
and that star bolt.  As far as individual line cords are concerned, only
one third pin should actually get to the wall,
More than one third pin is a ground loop and a noise src.

.


___
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


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, 1940)
If we desire respect for the law, we must first make the law respectable.
   - Louis D. Brandeis
Genes Web page 



___
Emc-users mailing list
Emc-users@lists.sourceforge.net

Re: [Emc-users] losing encoder counts

2022-10-03 Thread Sam Sokolik 
Our 80's vintage matsuura started losing counts when the 5v supply started
to lower..  don't remember how far down..
  Maybe 4.7v?

On Sun, Oct 2, 2022, 11:54 PM gene heskett  wrote:

> On 10/2/22 22:25, John Figie wrote:
> > So if you missed an A or a B pulse from the encoder I think you should
> see
> > the position count go in the opposite direction for one count. Another
> > words, instead of a position sequence of 0,1,2,3,4,5,6,7 you might see
> > instead 0,1,2,3,2,3, and in addition the time spent in 3 followed by 2
> > would each be twice as long.  So maybe if you can trigger on a non
> > monotonic position sequence you can capture the event.
> >
> > John Figie
> >
> > On Sun, Oct 2, 2022, 7:49 PM Jon Elson  wrote:
> >
> >> On 10/2/22 19:34, John Dammeyer wrote:
> >>> Oh boy does that bring back memories.
> >>> I started with the US Digital Encoders on the DC Servos.  According to
> >> the designer of the PIC upgrade board for the HP_UHU Servo drive the US
> >> Digital were crap and caused nothing but problems.
> >>> I switched to CUI and it appeared to be better.  Until one day circle
> >> milled inside a piece was off center.  See photo.
> >>> That’s when I started doing what you did and discovered that at very
> >> slow speeds I could return to the same 0 position but faster moves lost
> >> steps.  The error happened in only one direction.
> >>> Eventually I discovered that if I swapped the motor+encoder with the Y
> >> axis the problem moved to the Y so it wasn’t the HP_UHU drive.  Further
> >> research showed the X axis motor winding resistance was almost double
> that
> >> of the working Y axis motor.
> >>> It's possible I might have been able to fix the problem by doubling the
> >> current setting on the motor.  They were bought at the same time and
> were
> >> supposed to be identical.
> >>> I don't know why the HP_UHU servo drive didn't catch the position loss.
> >> Like yours, the encoder signals looked nice.The CUI encoders seem
> to be
> >> high
> >>> One other point.  I switched to the Bergerda AC Servo drives which to
> >> date have been great.  Talking to Donald Chen the sales guy at Bergerda
> he
> >> mentioned that their encoders are Japanese and more expensive than ones
> >> used on a lot of their competitors.  After that conversation I took one
> >> apart and checked part numbers.  Sure enough, Japanese design made in
> China.
> >>> So try different encoders.
> >> Yes, I did see some really strange signals on the scope
> >> before. I tore the whole motor/encoder set apart and cleaned
> >> the inside of the encoder.  But, optical components in the
> >> encoder are held in place with glue, 39 year old glue!  I'm
> >> having doubts that everything is still properly aligned in
> >> there.
> >>
> >> So, I'm leaning toward replacing the encoders with newer
> >> high-end encoders with higher line count.  But, that will
> >> require machining adapter plates and making sure the
> >> encoders are well-centered.
> >>
> >> Thanks,
> >>
> >> Jon
> Jon and John, I'd vote for a lack of good "star' grounding. Somehow the
> encoder is picking up noise from the
> motors reversal. So you are getting an extra noise pulse into the
> encoder wireng and its going to be direction
> switch related. A good star ground, and all that just goes away. Even
> with unshielded cables.
>
> The whole point of the star ground is that this common ground may not be
> quiet. But everything is
> referencing the same bolt.  that bolt may not be grounded to earth, and
> can be bouncing around 500 volts, but its
> carrying ALL the logic with it and the logic is only sensitive to the
> logic level between the wire from the encoder
> and that star bolt.  As far as individual line cords are concerned, only
> one third pin should actually get to the wall,
> More than one third pin is a ground loop and a noise src.
> >> .
> >>
> >>
> >> ___
> >> 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
>
>
> 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, 1940)
> If we desire respect for the law, we must first make the law respectable.
>   - Louis D. Brandeis
> 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] losing encoder counts

2022-10-02 Thread gene heskett 

On 10/2/22 22:25, John Figie wrote:

So if you missed an A or a B pulse from the encoder I think you should see
the position count go in the opposite direction for one count. Another
words, instead of a position sequence of 0,1,2,3,4,5,6,7 you might see
instead 0,1,2,3,2,3, and in addition the time spent in 3 followed by 2
would each be twice as long.  So maybe if you can trigger on a non
monotonic position sequence you can capture the event.

John Figie

On Sun, Oct 2, 2022, 7:49 PM Jon Elson  wrote:


On 10/2/22 19:34, John Dammeyer wrote:

Oh boy does that bring back memories.
I started with the US Digital Encoders on the DC Servos.  According to

the designer of the PIC upgrade board for the HP_UHU Servo drive the US
Digital were crap and caused nothing but problems.

I switched to CUI and it appeared to be better.  Until one day circle

milled inside a piece was off center.  See photo.

That’s when I started doing what you did and discovered that at very

slow speeds I could return to the same 0 position but faster moves lost
steps.  The error happened in only one direction.

Eventually I discovered that if I swapped the motor+encoder with the Y

axis the problem moved to the Y so it wasn’t the HP_UHU drive.  Further
research showed the X axis motor winding resistance was almost double that
of the working Y axis motor.

It's possible I might have been able to fix the problem by doubling the

current setting on the motor.  They were bought at the same time and were
supposed to be identical.

I don't know why the HP_UHU servo drive didn't catch the position loss.

Like yours, the encoder signals looked nice.The CUI encoders seem to be
high

One other point.  I switched to the Bergerda AC Servo drives which to

date have been great.  Talking to Donald Chen the sales guy at Bergerda he
mentioned that their encoders are Japanese and more expensive than ones
used on a lot of their competitors.  After that conversation I took one
apart and checked part numbers.  Sure enough, Japanese design made in China.

So try different encoders.

Yes, I did see some really strange signals on the scope
before. I tore the whole motor/encoder set apart and cleaned
the inside of the encoder.  But, optical components in the
encoder are held in place with glue, 39 year old glue!  I'm
having doubts that everything is still properly aligned in
there.

So, I'm leaning toward replacing the encoders with newer
high-end encoders with higher line count.  But, that will
require machining adapter plates and making sure the
encoders are well-centered.

Thanks,

Jon
Jon and John, I'd vote for a lack of good "star' grounding. Somehow the 
encoder is picking up noise from the
motors reversal. So you are getting an extra noise pulse into the 
encoder wireng and its going to be direction
switch related. A good star ground, and all that just goes away. Even 
with unshielded cables.


The whole point of the star ground is that this common ground may not be 
quiet. But everything is
referencing the same bolt.  that bolt may not be grounded to earth, and 
can be bouncing around 500 volts, but its
carrying ALL the logic with it and the logic is only sensitive to the 
logic level between the wire from the encoder
and that star bolt.  As far as individual line cords are concerned, only 
one third pin should actually get to the wall,

More than one third pin is a ground loop and a noise src.

.


___
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



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, 1940)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 



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

Re: [Emc-users] losing encoder counts

2022-10-02 Thread John Figie 
So if you missed an A or a B pulse from the encoder I think you should see
the position count go in the opposite direction for one count. Another
words, instead of a position sequence of 0,1,2,3,4,5,6,7 you might see
instead 0,1,2,3,2,3, and in addition the time spent in 3 followed by 2
would each be twice as long.  So maybe if you can trigger on a non
monotonic position sequence you can capture the event.

John Figie

On Sun, Oct 2, 2022, 7:49 PM Jon Elson  wrote:

> On 10/2/22 19:34, John Dammeyer wrote:
> > Oh boy does that bring back memories.
> > I started with the US Digital Encoders on the DC Servos.  According to
> the designer of the PIC upgrade board for the HP_UHU Servo drive the US
> Digital were crap and caused nothing but problems.
> > I switched to CUI and it appeared to be better.  Until one day circle
> milled inside a piece was off center.  See photo.
> >
> > That’s when I started doing what you did and discovered that at very
> slow speeds I could return to the same 0 position but faster moves lost
> steps.  The error happened in only one direction.
> >
> > Eventually I discovered that if I swapped the motor+encoder with the Y
> axis the problem moved to the Y so it wasn’t the HP_UHU drive.  Further
> research showed the X axis motor winding resistance was almost double that
> of the working Y axis motor.
> > It's possible I might have been able to fix the problem by doubling the
> current setting on the motor.  They were bought at the same time and were
> supposed to be identical.
> >
> > I don't know why the HP_UHU servo drive didn't catch the position loss.
> Like yours, the encoder signals looked nice.The CUI encoders seem to be
> high
> >
> > One other point.  I switched to the Bergerda AC Servo drives which to
> date have been great.  Talking to Donald Chen the sales guy at Bergerda he
> mentioned that their encoders are Japanese and more expensive than ones
> used on a lot of their competitors.  After that conversation I took one
> apart and checked part numbers.  Sure enough, Japanese design made in China.
> >
> > So try different encoders.
>
> Yes, I did see some really strange signals on the scope
> before. I tore the whole motor/encoder set apart and cleaned
> the inside of the encoder.  But, optical components in the
> encoder are held in place with glue, 39 year old glue!  I'm
> having doubts that everything is still properly aligned in
> there.
>
> So, I'm leaning toward replacing the encoders with newer
> high-end encoders with higher line count.  But, that will
> require machining adapter plates and making sure the
> encoders are well-centered.
>
> Thanks,
>
> Jon
>
>
>
> ___
> 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] losing encoder counts

2022-10-02 Thread Jon Elson 

On 10/2/22 19:34, John Dammeyer wrote:

Oh boy does that bring back memories.
I started with the US Digital Encoders on the DC Servos.  According to the 
designer of the PIC upgrade board for the HP_UHU Servo drive the US Digital 
were crap and caused nothing but problems.
I switched to CUI and it appeared to be better.  Until one day circle milled 
inside a piece was off center.  See photo.

That’s when I started doing what you did and discovered that at very slow 
speeds I could return to the same 0 position but faster moves lost steps.  The 
error happened in only one direction.

Eventually I discovered that if I swapped the motor+encoder with the Y axis the 
problem moved to the Y so it wasn’t the HP_UHU drive.  Further research showed 
the X axis motor winding resistance was almost double that of the working Y 
axis motor.
It's possible I might have been able to fix the problem by doubling the current 
setting on the motor.  They were bought at the same time and were supposed to 
be identical.

I don't know why the HP_UHU servo drive didn't catch the position loss.  Like 
yours, the encoder signals looked nice.The CUI encoders seem to be high

One other point.  I switched to the Bergerda AC Servo drives which to date have 
been great.  Talking to Donald Chen the sales guy at Bergerda he mentioned that 
their encoders are Japanese and more expensive than ones used on a lot of their 
competitors.  After that conversation I took one apart and checked part 
numbers.  Sure enough, Japanese design made in China.

So try different encoders.


Yes, I did see some really strange signals on the scope 
before. I tore the whole motor/encoder set apart and cleaned 
the inside of the encoder.  But, optical components in the 
encoder are held in place with glue, 39 year old glue!  I'm 
having doubts that everything is still properly aligned in 
there.


So, I'm leaning toward replacing the encoders with newer 
high-end encoders with higher line count.  But, that will 
require machining adapter plates and making sure the 
encoders are well-centered.


Thanks,

Jon



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

Re: [Emc-users] losing encoder counts

2022-10-02 Thread John Dammeyer 
Oh boy does that bring back memories.
I started with the US Digital Encoders on the DC Servos.  According to the 
designer of the PIC upgrade board for the HP_UHU Servo drive the US Digital 
were crap and caused nothing but problems.
I switched to CUI and it appeared to be better.  Until one day circle milled 
inside a piece was off center.  See photo.

That’s when I started doing what you did and discovered that at very slow 
speeds I could return to the same 0 position but faster moves lost steps.  The 
error happened in only one direction.

Eventually I discovered that if I swapped the motor+encoder with the Y axis the 
problem moved to the Y so it wasn’t the HP_UHU drive.  Further research showed 
the X axis motor winding resistance was almost double that of the working Y 
axis motor.
It's possible I might have been able to fix the problem by doubling the current 
setting on the motor.  They were bought at the same time and were supposed to 
be identical.

I don't know why the HP_UHU servo drive didn't catch the position loss.  Like 
yours, the encoder signals looked nice.The CUI encoders seem to be high 

One other point.  I switched to the Bergerda AC Servo drives which to date have 
been great.  Talking to Donald Chen the sales guy at Bergerda he mentioned that 
their encoders are Japanese and more expensive than ones used on a lot of their 
competitors.  After that conversation I took one apart and checked part 
numbers.  Sure enough, Japanese design made in China.

So try different encoders.
John

> -Original Message-
> From: Jon Elson [mailto:el...@pico-systems.com]
> Sent: October-02-22 3:49 PM
> To: emc-users@lists.sourceforge.net
> Subject: [Emc-users] losing encoder counts
> 
> I thought I was about done with my R2E3 retrofit project.
> But, doing some testing, I found it was losing a little bit
> of position when moving. Every axis shows this, but it is
> worst on the Y axis. I have a program that moves Y 3" back
> and then returns to the indicator. It seems like at 12IPM or
> below, it doesn't seem to lose position, but above that
> speed it loses from a half to a full thousandth on each
> repeat. I've looked at the A and B quadrature signals with a
> scope, and they look good, the differential complements look
> like complements. I am using my PPMC boards, and I've never
> seen this on my older Bridgeport using the same hardware.
> 
> This retrofit is using AMC 30A20AC analog servo amps, and
> the original motors. I reterminated the original Bridgeport
> encoder cables to the PPMC encoder board, and have checked
> all that wiring.� The cables are individually shielded
> twisted pairs until the last couple inches at the controller
> end.
> 
> Anybody have any suggestions? I'm strongly considering
> replacing all the encoders, there are some nice ones on eBay
> right now, but I would have to make
> shaft and mounting adapters.
> 
> The encoders are DataMetrics S-9481A-250.
> 
> Thanks for any comments,
> Jon
> ___
> 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