retrofitting a New CNC router (Osai Open XS control to Linuxcnc)

andypugh wrote:

Looking throigh the INI I see that JOINT1 has HOME_USE_INDEX = YES whereas JOINT2 has HOME_USE_INDEX = no

The rest of the config looks like both should be set to "YES"

yea i just did that so i could actually home it. It was set to yes when it was doing what i described. I meant to change it back before posting, but forgot, sorry.

PCW wrote:

Los of control at higher speeds may mean that you have exceeded the count rate of
the encoders with the current sample rate/filter settings. A simple check would be to see if 
setting the encoder filter off increases the speed before runaway.

 

ok i will check. that may be likely due to the fact that the encoder scale is very high. good call. Since the encoder signal comes from the drives and not the actual encoder, i may be able to scale it down if it is anything like my other machine. 

You can speed the 7I95T encoder counting up quite a bit if needed
but with the default sample rate and filter settings, its pretty slow

PCW wrote:

You can speed the 7I95T encoder counting up quite a bit if needed
but with the default sample rate and filter settings, its pretty slow

So i just did some math, and i hope i'm not screwed here. 
The counts per second for each axis are as follows at max velocity.
Z 3,329,228.8
X 3,986,980.99
Y1 3,986,980.99
Y2 3,986,980.99

I saw the sample rate in the manual, is that per channel or total?

The sample rate applies to all channels (and defaults to 25 MHz)
With the digital filter on, 15 samples are required to detect a signal
so about 600ns per edge and 1.66 Mhz max count rate. With the filter off,
3 samples are required to detect an input so ~120 ns per edge or about
8 MHz maximum count rate. Because of quadrature distortion (the quadrature
from the drive is not exactly 50% duty cycle) you  need a lot of margin.

 

PCW wrote:

The sample rate applies to all channels (and defaults to 25 MHz)
With the digital filter on, 15 samples are required to detect a signal
so about 600ns per edge and 1.66 Mhz max count rate. With the filter off,
3 samples are required to detect an input so ~120 ns per edge or about
8 MHz maximum count rate. Because of quadrature distortion (the quadrature
from the drive is not exactly 50% duty cycle) you  need a lot of margin.


 

So if i'm understanding this right, i should be fine with filtering off? 

so i think i fixed the overrun issue by changing the encoder output from the servo drives to a smaller number. However, i am still having the homing issue