> >
> > The comment on the link about Ethernet latency is what interested me.
> > Again if electrical noise causes retries or delays in the Ethernet
> > signalling then that could cause this sort of error under heavy load.
> >
> > John
> 
> And this is what the high frequency gains do, FF2 and Dgain, giving a
> boost when the velocity changes. All this assumes that the actual power
> available is sufficient to achieve what the controller is asked to make
> it do. If the available power sags with the initial accell but the speed
> being asked is so high the power available cannot catch it up and is
> limiting you then following errors creep into long distance moves.
> 
Hi Gene,
My AC Servo is run open loop with step/dir.  It's not the Servo Driver that is 
calling out a following error.  It's LinuxCNC that also has no clue where the 
motor actually is unless the AC servo asserts a fault output.  And it can't do 
that because I disconnected it for this test.

I don't know how the Ethernet Communications with the MESA 7i92H works.  Page 8 
of the user manual says it's UDP which means launch and forget.  The sender has 
no way of knowing if the message was received although with only two devices 
perhaps there's a lower level that does verify that.

In either case Page 31 states there are error counters so if there is a packet 
error that is detected after a new motion command is issued; like change 
direction for example.  Then LinuxCNC might well decide to throw up a following 
error.

I will admit that I misread the HAL instructions and had the MAX_ACCELERATION 
parameter set to 125% higher.  But with Backlash it was supposed to be 200%.  
Once I changed that the problem went away.  So bad HAL info generates a 
following error?  Except that never happened without the extra 92 lbs on the 
table.

But if electrical noise from the higher load did cause a damaged UDP packet 
then there's the answer.  A quick change in direction to handle backlash and a 
missing message throws out the timing and results in a following error.
John




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

Reply via email to