additive offset error

Well...... i seem to have found a software error.
my mazak has twin turrets, and therefore has some tools on the negative side, and some on the positive side. i noticed that before one of the last patches, my rear turret tools couldn't have an offset value with more than one decimal place, (eg 456.1 -tis a long way between turrets). this changed with one of the previous patches. -But ever since i have occasionally had problems that i thought was the "X" axis encoder count going bad. ie. every part i ran shifted the axis a little. till today, when i noticed that if i limit all rear turret offsets to two decimal places, it stops creeping.

anyone else had this?

im using the latest stable version, doing lotsa overtime at the moment, but ill check the repeatability of this in a couple of days time.
ill also check the math (if the values add up.)

basicly when i limited the rear offsetts to two decimal places (out of sheer desperation, and some straw clutching) the error stopped.

so, a brief description,

im running a program that uses 5 tools in the rear turret, and 7 front turret tools.

when i ran it, on every run the x axis grew by 0.1mm.

(the part was larger)

if i re zeroed the x axis i was back to my first turning size.

i have no usb thumb drives inserted, and all screensavers/sleep modes are disabled.

upon changing the rear turret offsets to two decimal places (0.00) the error stopped.

the offsets were set using the touch off button on the axis screen, and were 5 decimal places long. (0,00000)

otherwise the ol' mazak (1984 model) has been doing 12 hr days for the last coupla weeks (bar feeding parts) much better than the old T1 control!

ill report back on this when i have some time, -changing setups in two days probably.

thanks!

synaptic says 1:2.4.6

and the dro dosent change, but the axis shifts, -only detectable by measuring the parts.
but the size increased by 0.1mm every run. (0.08~0.12) i havent had time to do a proper test yet -our welders are waiting on the parts im making

oh -2.4.6

and yes, i expected it to be a mech error, i should be able to to try to reproduce it today. i did find out that the operator had a usb thumb drive in use at the time, but no errors were being reported.....

hmmm.....

so i had an error of 0.03 adding to x on every part for the last 20 mins. (not great on a 1min cycle time on a barfed job) if i homed x it would reset the error, i don't believe there is somthing loose -as i went thru all the mechanicals of this mc during its retrofit...

after 20mins now, all is well, so i geuss i betta check that x axis ...

maybe after this job run..... -next week.

just checked out the encoder mounting, and all is as it should be- but if i slow the rapid traverse rate the error stops... looks like it is a counter/card/read speed issue..

more on this later......