Re: [Emc-users] Odd encoder problem

[Peter C. Wallace]

On Wed, 1 Nov 2017, Eric H. Johnson wrote:


Date: Wed, 1 Nov 2017 19:25:36 -0400
From: Eric H. Johnson 
Reply-To: "Enhanced Machine Controller (EMC)"

To: "'Enhanced Machine Controller (EMC)'" 
Subject: [Emc-users] Odd encoder problem

All,



A machine (table) that has been running for about three months has developed
an odd problem in the last couple weeks. Not sure if it has been getting
worse or if just the nature of the patterns being run has made the problem
more pronounced. The machine is an XYYA configuration with a pneumatic Z,
and using mainly Mesa boards. The problem is in the A axis. It seems to be
losing encoder counts at very low speed, but not otherwise. It took a bit to
track down because when I initially had it moving multiple rotations I could
not get any errors running A by itself. Running with a feed rate in excess
of 2 (Degs per min), resulted in no errors. Slowing down to 1
started to show a very small amount of error, and from 5000 on down the
error was very pronounced.



The hardware is a Mesa 7i43 FPGA, 7i33 quad analog servo card and YEnc line
drivers. The table is 15' x 30', thus requiring line drivers on X and A, but
for uniformity, were put  on all four axes. The encoders are wired
differentially. The error is always in the same direction regardless of the
velocities used for the cw and ccw directions, thus would seem to be
dropping counts.



I would check that you get decent differential voltages across the 7I33 inputs 
in both states of the inputs (no less than +-2.5 across the A,/A B,/B or Z,/Z 
pairs in any state)






I did not have a scope with me to check the encoder signals. While a low
probablility, I do have spare YEnc modules, but ran out of time before
trying to swap it out. I don't really see how this type of problem
(electrical) can develop over time, but do not see how it can be anything
else.



Anyone seen anything like this when using that hardware?



Thanks,

Eric



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



Peter Wallace
Mesa Electronics

(\__/)
(='.'=) This is Bunny. Copy and paste bunny into your
(")_(") signature to help him gain world domination.


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

Re: [Emc-users] Odd encoder problem

[Chris Albertson]

Once have I a connection where only fast signal worked.  In turned out the
signals were coupled with a capacitor that was to small to pass low
frequency.   Of course no capacitor was designed into the system but a
loose solder joint was acting like a capacitor.  Actually what else is a
cap, but two metal plates with a tiny air gap between them.   The fix was
easy, apply heat with solder iron.   Once you know what look for it is easy
to find.

It could be something like that, Some place there is an unexpected high
pass filter or a low pass the shunts to ground or Vcc



On Wed, Nov 1, 2017 at 4:25 PM, Eric H. Johnson 
wrote:

> All,
>
>
>
> A machine (table) that has been running for about three months has
> developed
> an odd problem in the last couple weeks. Not sure if it has been getting
> worse or if just the nature of the patterns being run has made the problem
> more pronounced. The machine is an XYYA configuration with a pneumatic Z,
> and using mainly Mesa boards. The problem is in the A axis. It seems to be
> losing encoder counts at very low speed, but not otherwise. It took a bit
> to
> track down because when I initially had it moving multiple rotations I
> could
> not get any errors running A by itself. Running with a feed rate in excess
> of 2 (Degs per min), resulted in no errors. Slowing down to 1
> started to show a very small amount of error, and from 5000 on down the
> error was very pronounced.
>
>
>
> The hardware is a Mesa 7i43 FPGA, 7i33 quad analog servo card and YEnc line
> drivers. The table is 15' x 30', thus requiring line drivers on X and A,
> but
> for uniformity, were put  on all four axes. The encoders are wired
> differentially. The error is always in the same direction regardless of the
> velocities used for the cw and ccw directions, thus would seem to be
> dropping counts.
>
>
>
> I did not have a scope with me to check the encoder signals. While a low
> probablility, I do have spare YEnc modules, but ran out of time before
> trying to swap it out. I don't really see how this type of problem
> (electrical) can develop over time, but do not see how it can be anything
> else.
>
>
>
> Anyone seen anything like this when using that hardware?
>
>
>
> Thanks,
>
> Eric
>
>
>
> 
> --
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>



-- 

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

Re: [Emc-users] Odd encoder problem

[Gene Heskett]

On Wednesday 01 November 2017 20:14:07 andy pugh wrote:

> On 1 November 2017 at 23:25, Eric H. Johnson  
wrote:
> > The problem is in the A axis. It seems to be
> > losing encoder counts at very low speed, but not otherwise. It took
> > a bit to track down because when I initially had it moving multiple
> > rotations I could not get any errors running A by itself. Running
> > with a feed rate in excess of 2 (Degs per min), resulted in no
> > errors. Slowing down to 1 started to show a very small amount of
> > error, and from 5000 on down the error was very pronounced.
>
> I think that this might be a floating GND or a half-disconnected
> differential signal.
>
> My thinking is that at high pulse rates you get AC coupling and the
> data gets through, but at low pulse rates the coupling breaks down.

Speaking as a CET, that is also my first assumption.

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

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

Re: [Emc-users] Spindle pid problems

[Gene Heskett]

On Wednesday 01 November 2017 12:30:20 andy pugh wrote:

> On 1 November 2017 at 15:53, Gene Heskett  wrote:
> > Now I have additional switches to make work to control it, in
> > particular I'll use 2 switches and 2 mux2's in the end so that motor
> > commands are blocked if the knob is not properly homed in one
> > position or the other.
>
> I would suggest a single mux4 instead (set to output 0 for 00 and 11
> at the switches)

I'd forgotten we had that, good idea as it also gets rid of a sum2 to 
pass the output of the two mux2's to a single path to the motor.

The backing plate, due in friday, also fell out of the little white USPO 
van today.  The registration ring will need reduced by about 1/4 inch, 
and new bolt holes will need drilled & tapped. Expected when the backer 
is by Shars, and the chuck is a Steelex with a shop fox logo on it.  
Drilling and tapping those 4 bolt holes will test the G0704 as the bolts 
are 10mm's. Looks like 1.25mm fine thread, but will measure for sure. I 
know I've an 8x1.0, but  not sure about the 10x1.25 without taking 
inventory. It might be a 1.50mm tap.

Shars who made this plate, included a spare catalog, and I note their 
tapping arsenal does not include a way to grab a tap by the square on 
the end of it with an r8 key stub on the other end. I wish there was a 
std size for the rear of a tap. But in 15 lbs of assorted taps, I doubt 
it a lot if I have 2 with the same square butt stub size.  And getting a 
truly non slip grip on one of them in an er32 collet is often very 
frustrating. Particularly when you note after the spindle has started, 
the tip of the tap is going in 1/16" circles because the tap wasn't ever 
straight in the first place.

So boring and tapping the new holes will probably discolor the garage 
air. But first  make the register fit so I can use the chuck to locate 
the new  holes, the whole bolt hole pattern is out of square by about 25 
thou means the first thing I do after making the register fit, is put 
witness marks on everything lined up with jaw #0.  But at that price, I 
don't think they would ever bother to polish the sow's ear.  And while 
its well finished, its not that precisely made. But its a 4 jaw 
independent, so who's got a quarter to call somebody who actually cares? 
I don't... :(


Thanks, Andy

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

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

Re: [Emc-users] Odd encoder problem

[Eric H. Johnson]

Andy,

That sounds plausible, I will check at the next opportunity.

Thanks,
Eric


On November 1, 2017 8:14:07 PM EDT, andy pugh  wrote:
>On 1 November 2017 at 23:25, Eric H. Johnson 
>wrote:
>
>> The problem is in the A axis. It seems to be
>> losing encoder counts at very low speed, but not otherwise. It took a
>bit to
>> track down because when I initially had it moving multiple rotations
>I could
>> not get any errors running A by itself. Running with a feed rate in
>excess
>> of 2 (Degs per min), resulted in no errors. Slowing down to 1
>> started to show a very small amount of error, and from 5000 on down
>the
>> error was very pronounced.
>
>I think that this might be a floating GND or a half-disconnected
>differential signal.
>
>My thinking is that at high pulse rates you get AC coupling and the
>data gets through, but at low pulse rates the coupling breaks down.
>
>-- 
>atp
>"A motorcycle is a bicycle with a pandemonium attachment and is
>designed for the especial use of mechanical geniuses, daredevils and
>lunatics."
>— George Fitch, Atlanta Constitution Newspaper, 1916
>
>--
>Check out the vibrant tech community on one of the world's most
>engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>___
>Emc-users mailing list
>Emc-users@lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/emc-users

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Odd encoder problem

[andy pugh]

On 1 November 2017 at 23:25, Eric H. Johnson  wrote:

> The problem is in the A axis. It seems to be
> losing encoder counts at very low speed, but not otherwise. It took a bit to
> track down because when I initially had it moving multiple rotations I could
> not get any errors running A by itself. Running with a feed rate in excess
> of 2 (Degs per min), resulted in no errors. Slowing down to 1
> started to show a very small amount of error, and from 5000 on down the
> error was very pronounced.

I think that this might be a floating GND or a half-disconnected
differential signal.

My thinking is that at high pulse rates you get AC coupling and the
data gets through, but at low pulse rates the coupling breaks down.

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

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

Re: [Emc-users] Odd encoder problem

[Eric H. Johnson]

Gregg,

I did that, well mostly. I rechecked and tightened every encoder connector, 
although not necessarily disconnecting and reconnecting.

The housing for the line drivers rides in the gantry, so is in motion. Those 
were the first connectors I checked.

I can do a more thorough job, but saw no difference after tightening them up.

Thanks,
Eric

On November 1, 2017 7:54:32 PM EDT, Gregg Eshelman via Emc-users 
 wrote:
>Basics first. Disconnect power, unplug and re-connect all plugs, loosen
>and tighten all screw terminals. Re-test. Eliminates any minor
>corrosion or possible looseness causing an intermittent.
>
> 
>On Wednesday, November 1, 2017, 5:47:05 PM MDT, Eric H. Johnson
> wrote:  
> 
> All,
> 
>
>A machine (table) that has been running for about three months has
>developed
>an odd problem in the last couple weeks. Not sure if it has been
>getting
>worse or if just the nature of the patterns being run has made the
>problem
>more pronounced. The machine is an XYYA configuration with a pneumatic
>Z,
>and using mainly Mesa boards. The problem is in the A axis. It seems to
>be
>losing encoder counts at very low speed, but not otherwise. It took a
>bit to
>track down because when I initially had it moving multiple rotations I
>could
>not get any errors running A by itself. Running with a feed rate in
>excess
>of 2 (Degs per min), resulted in no errors. Slowing down to 1
>started to show a very small amount of error, and from 5000 on down the
>error was very pronounced.  
>--
>Check out the vibrant tech community on one of the world's most
>engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>___
>Emc-users mailing list
>Emc-users@lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/emc-users

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Odd encoder problem

[Gregg Eshelman via Emc-users]

Basics first. Disconnect power, unplug and re-connect all plugs, loosen and 
tighten all screw terminals. Re-test. Eliminates any minor corrosion or 
possible looseness causing an intermittent.

 
On Wednesday, November 1, 2017, 5:47:05 PM MDT, Eric H. Johnson 
 wrote:  
 
 All,
 

A machine (table) that has been running for about three months has developed
an odd problem in the last couple weeks. Not sure if it has been getting
worse or if just the nature of the patterns being run has made the problem
more pronounced. The machine is an XYYA configuration with a pneumatic Z,
and using mainly Mesa boards. The problem is in the A axis. It seems to be
losing encoder counts at very low speed, but not otherwise. It took a bit to
track down because when I initially had it moving multiple rotations I could
not get any errors running A by itself. Running with a feed rate in excess
of 2 (Degs per min), resulted in no errors. Slowing down to 1
started to show a very small amount of error, and from 5000 on down the
error was very pronounced.  
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] Odd encoder problem

[Eric H. Johnson]

All,

 

A machine (table) that has been running for about three months has developed
an odd problem in the last couple weeks. Not sure if it has been getting
worse or if just the nature of the patterns being run has made the problem
more pronounced. The machine is an XYYA configuration with a pneumatic Z,
and using mainly Mesa boards. The problem is in the A axis. It seems to be
losing encoder counts at very low speed, but not otherwise. It took a bit to
track down because when I initially had it moving multiple rotations I could
not get any errors running A by itself. Running with a feed rate in excess
of 2 (Degs per min), resulted in no errors. Slowing down to 1
started to show a very small amount of error, and from 5000 on down the
error was very pronounced.

 

The hardware is a Mesa 7i43 FPGA, 7i33 quad analog servo card and YEnc line
drivers. The table is 15' x 30', thus requiring line drivers on X and A, but
for uniformity, were put  on all four axes. The encoders are wired
differentially. The error is always in the same direction regardless of the
velocities used for the cw and ccw directions, thus would seem to be
dropping counts.

 

I did not have a scope with me to check the encoder signals. While a low
probablility, I do have spare YEnc modules, but ran out of time before
trying to swap it out. I don't really see how this type of problem
(electrical) can develop over time, but do not see how it can be anything
else.

 

Anyone seen anything like this when using that hardware?

 

Thanks,

Eric

 

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

Re: [Emc-users] Spindle pid problems

[andy pugh]

On 1 November 2017 at 15:53, Gene Heskett  wrote:

> Now I have additional switches to make work to control it, in
> particular I'll use 2 switches and 2 mux2's in the end so that motor
> commands are blocked if the knob is not properly homed in one position
> or the other.

I would suggest a single mux4 instead (set to output 0 for 00 and 11
at the switches)

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

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

Re: [Emc-users] Spindle pid problems

[Chris Albertson]

On Tue, Oct 31, 2017 at 7:58 PM, Gene Heskett  wrote:

> On Tuesday 31 October 2017 21:20:22 Chris Albertson wrote:
>
> > Can you convert the gear position to a number, either 1.0 or 0.0
> >
> Thats easy enough.
>
> > Now you one need one PID, no MUX.  Multiple each parameter but ehvalue
> > of the switch or not-switch
> >
> > P =  Int(SW)*Plow + int(not(SW))*Phigh
>
> But this is code that would need to be halcompiled. It doesn't look like
> anything hal is known to be able to do.
>

I didn't mean that you cut and past the code.Just that some place you
multiply by the switch setting which is either one or zero.

-- 

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

Re: [Emc-users] Spindle pid problems

[Gene Heskett]

On Wednesday 01 November 2017 06:30:08 andy pugh wrote:

> On 1 November 2017 at 02:34, Gene Heskett  wrote:
> >> bit   FALSE  spindle-index-enable
> >>  <=> motion.spindle-index-enable
> >>  ==> pid.0.index-enable
> >>  ==> pid.1.index-enable
> >>  <=> sim_spindle.index-enable
> >
> > I'll have to look that one up, if it works, it will save a huge
> > rewrite tomorrow. I take it the bit FALSE is equal to a net in most
> > cases?
>
> Sorry for any confusion I caused, I just copied and pasted the output
> of  "show sig" in halcmd after I had created the net.
> The net name is spindle-index-enable, the other items are the
> connected pins, bit is the type and FALSE the value at the time of the
> query.

Not a lot of confusion other than I thought you were trying to teach me 
about a new hal keyword. In any event, without power to the motors, it 
worked. Now I have additional switches to make work to control it, in 
particular I'll use 2 switches and 2 mux2's in the end so that motor 
commands are blocked if the knob is not properly homed in one position 
or the other.

I saw a t-shirt, in ladies style at Wallies a while back, say's "I'll go 
to work when my coffee does". I told the clerk if she had it in mens 
style, I'd take 2 of them.  Still a pint low though.

Thanks Andy.

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

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

Re: [Emc-users] Ethercat master (development board)

[Nicklas Karlsson]

Infineon XMC4300 have builtin Ethercat peripheral but it require two
external drivers, two sets of MII pins and extra cost is roughly the same
as using external chip.

2017-10-31 22:27 GMT+01:00 Chris Albertson :

> Yes oddly enough it's a significanly cheaper product made from
> > significanlty more expensive components.
>
>
> Oddly?  No.  This is what always happens with electronics if you wait.
>
> Today Ethercat looks to be the way the world of automatic is going.   It
> was adopted as a preferred standard in China and there are many German
> companies turning out ICs.
>
> Wait long enough these will be commodity priced parts.   Remember with
> Gecko style stepper drivers we expensive?  What's going to happen is ST
> Micro or a company like it (TI?) will release an ARM M with Ethercat PHY as
> a built in peripheral.  Then you have can have a one chip controller. (just
> add RJ45 and magnetics)   They already build I2C, SPI and Ethernet into the
> $1 chips.  They will do Ethercat at some point.
>
>
> --
>
> Chris Albertson
> Redondo Beach, California
> 
> --
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Spindle pid problems

[andy pugh]

On 1 November 2017 at 02:34, Gene Heskett  wrote:

>> bit   FALSE  spindle-index-enable
>>  <=> motion.spindle-index-enable
>>  ==> pid.0.index-enable
>>  ==> pid.1.index-enable
>>  <=> sim_spindle.index-enable
>
> I'll have to look that one up, if it works, it will save a huge rewrite
> tomorrow. I take it the bit FALSE is equal to a net in most cases?

Sorry for any confusion I caused, I just copied and pasted the output
of  "show sig" in halcmd after I had created the net.
The net name is spindle-index-enable, the other items are the
connected pins, bit is the type and FALSE the value at the time of the
query.

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

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