About ready to pull the plug on Mach and go to stand alone system but wanted to make one last attempt to solve issue before making the change. I have pretty much exhausted all of my ways of getting assistance as no one knows why or does not reply to request for assistance. I have spent a lot of days and hours trying to figure this out in the last year and a half. We all know the definition of insanity and I feel like I am headed that way. I have reached out to Newfangled solutions, CNC4pc and Dave Gatton (machine designer) in the past as well as those who have not replied to my request for help. It is frustrating.  I am technically savvy being a licensed aircraft mechanic, professional flight engineer and commercial pilot .

CNC Machine is a Gatton designed diy CNC router. Current cnc control is diy stepperonline systems using 542T drivers and nema 23 425oz motors with a UC100 motion controller. To test motion controllers I have tried three different units. Computers are currently Lenovo laptop with Windows 10 and Dell laptop using Windows 7.

1) The following is an example of what a test run produces: G-code generated in Vectric is for a 10" circle profile cut on .75 inch material. Using 10 passes to cut .5inch deep. No tabs. Results at end of run are as follows. New zero (machine) is approx. 2.5 inches off in BOTH X&Y axis and also off in z axis. The amount of offset is derived by jogging router bit back to the original zero and reading the DRO for the travel required to return to the original zero point. Considering the 10 passes there is .25 inches of consistent change per pass. I also get similar results with runs with less passes. Test are air cuts to save material and remove load from the equation.

2) To eliminate the possibility of this being a mechanical problem, I have conducted test running rectangular patterns using DRI commands which results with zero returning to the original start position after a series of passes.  Seems that the drift only occurs when running g-code with MACH.

Problem started after I had to replace defective, no longer supported xylotex controller with the diy build using Stepperonline drivers.

Motors are tuned to 6400 steps/ and velocity of 200 with pulse of 5/5 using 1/2-10 5 start lead screws.

Retightens all couplers, all wire connections, separated signal cables to reduce possibility of signal bleed.

Everything worked great on xylotex using original XP desktops and now laptops.

At 68 this is just a hobby and I do not want to invest another $700 into a Masso CNC controller if I can get Mach to work with my computer correctly.

About ready to pull the plug on Mach and go to stand alone system but wanted to make one last attempt to solve issue before making the change. I have pretty much exhausted all of my ways of getting assistance as no one knows why or does not reply to request for assistance.

I would suspect your self-build controller.
Perhaps consider a controller with proper manufacturer's support such as the 'Gecko 540' or similar quality product. https://www.geckodrive.com/g540-4-axis-digital-stepper-drive.html

Tweakie.

Motors are tuned to 6400 steps/ and velocity of 200 with pulse of 5/5 using 1/2-10 5 start lead screws.

that means your Motors are set to 3200Steps/rev.
i would try to half (1600) this and test.
for test run a simple G_code for one axis like:

F20
G1 X10
G1 X0
G1 X10
G1 X0
G1 X10
G1 X0
G1 X10
G1 X0
G1 X10
G1 X0
M30

and have a look. also reducing the acceleration might help.
how are the step/dir signal's  wired?

Thanks for the input from all.  Today I conducted some test runs similar to that suggested by TPS.  I created separate strait line profiles of 10 inches and 10 passes for both the on the X and Y axis which only used one axis at a time. In each case the router returned to the original zero with no offset.  I had previously did this using the MDI commands but had not done the next test. I also used my Windows 7 laptop for this test.   

I further created two different profile lines on a 45 degree diagonal across center of 10 inches and 10 passes running from left bottom quadrant to top right and vise versa.  I got totally different results.  One of the diagonals resulted in 3.5 inch offset of both axis to return to center.  The other had a much smaller offset. Less than 1 inch.  Hmmm. I ran several of these with similar results, but ran out of time.  Tomorrow I will conduct the same test using my Windows 10 laptop and take screenshots for the record.

With these results I am starting to lean towards this being a EMI noise problem.  I have been using the cables that came with the original xylotex unit since all worked well back when.  I went ahead and ordered 4 core shielded cable, copper tape and new aviation connectors.  This also maybe explains why occasionally I have had gantry racking on the Y axis which has the A axis slaved.  I did a little reading on EMI interference today and found that power supplies can also cause interference.  In the original xylotex there was only one pwr sply.  I now have 3 pwr supplies.  I may have to isolate their cables as well.

In hopes of getting this resolved and ultimately help someone else suffering from a similar problem, I will provide updates as this develops. It will be a week before the shielded cable arrives from Amazon.

I finally got back to the garage and rewired stepper cables with quad core shielded cable with each cable shield connected to earth ground. Test results had no change. Still.25inch change with every pass on a simple 10 diameter circle profile cut ending with a 2.5 inch offset in both the x & y axis after a 10 pass air cut.  I also changed stepper driver dip sw settings with no change.

As noted in previous post if a test is run only using the x or y axis the router returns to the original start point.

New steps to trouble shoot is take an older Dell desktop and wipe it clean with a fresh install of Linux and LinuxCNC using the same controller via parallel port or UC100 and USB.  I may partition the HD so that I can also test run Mach3 using a parallel point connection instead of UC100. If the problem still exist it is in fact the diy stepperonline controller, BOB etc. and I will go in a different direction.  This test will cost nothing but time.

Run the machine with the spindle turned OFF and see if you get the error.

Thanks for your reply.  I use a router motor which is not plugged in.  Its like there is interference but not sure where its coming from. Only other possibility is from 3 DC Pwr Splys. BOB 12V, Stepper drivers 5V & motors 40V tuned down to 38v.

Well we eliminated Mach3, Windows based computers and UC100 motion controller as potential problem sources. My son in law reworked a Dell desktop with a parallel port I had and installed Linux and LinuxCNC. Once we got the settings correct we ran a test file which was basically an oval profile cut since I had it on a stick in my pocket.  The good news is it is not the computers, Mach3 or the UC100 which narrows it down to the Stepperonline drivers or the 5 axis BOB. I have changed the dip switch settings several times in an attempt to find a cause so i have reached out to Stepperonline for guidance.  I will update this once I know more and hopefully finally have this resolved.

Oops! I forgot to mention the test results: The oval test profile cut had the same problem only to a much larger drift.  I had to shut down the run before it completed 2 passes to avoid exceeding machine x axis limit.

as mentoined earlier try a smaller ammount for microstepping and start with
low values for Velocity (50) and Acceleration (20).
then make single axis moves with a fixed amount to check and to get axis bay axis under control.

how are your dip Switches set for Motor current (1-3)?