.0005 is a small amount(;-) it could be backlash, Screw torque, bearing preload,  bed gibbs need adjusting, Vibration and THat long arm on your indicator, gravity and that large vise on one end. ANY and all could add up to .0005. I bet you can lean on the mill and get that amount.

If you can ref home then do so and run a program. Then do a verify and see how far it is off from  the home position.

Just a thought, (;-)

no problem.  I've tried changing the active state of the Dir pins and the result is that when I start the program that moves + first, When it gets done...  It still moved positive.  It seems to be about .0005" each time I run the repeatability test.  Yes it is cumulative.  Running the test with the accel at .1 just now after 6 runs, I've accumulated about .0032" error.  This agrees with the Dial on the right end of the leadscrew.  It is a standard Bridgeport type.  This is a very frustrating anomaly.  I haven't had this happen before until just this last week.  Could adding the 4th Axis possibly be related?  It was added about 2 weeks ago.  Pretty straight forward.  Thanks for the help, Don

To Vmax549, Thanks for your input.  See the above post.  The machine is pretty tight, but smooth.  We rescraped the ways and refitted the gibs about 2 years ago when w got it and prepared it for the retrofit.  There is about .0005" mechanical backlash in each axis.  If it was staying within that and not creeping off in one direction I would be fine.  I haven't hooked up any REF switches yet...  To lazy.  BTW, the small bronze handwheels were installed to prevent accidents, shuch as bruized legs, blown parts, etc.  They were raw castings I turned.  Sorry about the clutter on the table.  Don

Don,
Just so you know, i am are trying to help and understand frustration. So bear with me on information requests.
As i said before, it could be any part of the system or combination of the components. Sometimes easy to fiind
and sometimes just elusive as a fox. Little things can be additive and as you go deaper into the each and every
item it can even become un-nerving.

What kind / accuracy grade of ball screws are you using?
Are the results the same along the axis? (say try three different points along the screw travel, if 10 inches of screw travel,
 then say at the 2-3"  /  5-6"  / 8-9")
Can you check or verify the steps / unit over a longer distance of travel? ( say 10", 5" movements )
But before you aggrevate yourself, or we drive you crazy, try the following as it is easy to do.

The only way I know of isolating / checking the steps sent to a stepper without electronic gismos would be to
to put a index mark  ( fine line) on the motor pulley ( can increase the size via paper circle), send a G0 move for  say  1 rev of motor, set the pointer on the index mark,
do the same move,when it completes the move the indexed point should be right a pointer. Over and over again. if you do the same move for 15  or 50 times does it come off the mark? The system has been isolated down to software + sending devices+motor. Thus it eliminates the rest of mechanical items in checkout.  You can work your way down the mechanical train and look for changes in movement. You can also choose to not microstep.
Poor mans encoder readout. 
RICH
 

Hi, Rich.  Thanks for the suggestions and encouragement.  I'm also posting in the Yahoo! Gecko forum and Maris has just suggested the pointer on the motor too. he also suggests a couple of other tests for repeatability at different distances to see if it is drifting relative to the amount of travel.  I will try those steps soon and report back.  He also suggests trying "no auto standby" on the option header as another test.  Thanks, Don

Hi, again. below is a description of the tests Mariss had me try and my results.
Mariss, I just ran the tests you suggested. 1" and back and lost
about .0002" each repitition. Running the 10 reps lost .0020" and
100times lost .020" . The mark on the stepper pulley was off a
corresponding amount as well as the dial on the ballscrew end
opposite the drive. Then I set the No Auto-Standby mode and it had
no effect. What next? Thanks, Don

--- In geckodrive@yahoogroups.com, "Mariss Freimanis"
<mariss92705@...> wrote:
>
> 1) Set up a test program where you move some distance (1"?) and
back.
> Repeat it 1, 10 and 100 times. Does the error grow a repeatable
> amount; 1, 10 and 100 times bigger?
>
> 2) Take a Sharpie pen or a scribe. Mark a line from the motor
shaft to
> the motor body. Does the mark line-up after 1, 10 or 100 tries of
(1)?
> If not, is the angle error 1, 10 and a 100 times greater when you
run (1)?
>
> 3) Try setting the drive to "No Auto-Standby" on the option header.
>
> Mariss

In your pics of the drives I think I see caps connected to them, which terminals  are they between and are there meant to be caps there? I have only used 201 and 202 drives and I cant recall caps being needed there.
Hood

Hood,
You should have caps on the 201 drives as noted in the liiterature from Geco.
That cap can save you from blowing a drive.
Rich

should it be in the positions these drives have them? Its been so long since I had the 201's fitted I cant remember but seemed to remember they were on the Voltage Inputs from the PSU.
Hood

Here is a similar topic.   http://www.machsupport.com/forum/index.php/topic,7364.msg47127.html#msg47127

I had a similar problem that was resolved by increasing the PULSE width.
Here is a clip.

RC