iHi, Today i proved the version 3.042.031 for first time, with a new program with enough interpolation line, i see with the Look Ahead of twenty no is enough for this and i see stopped and outburst frequently, i need a Look Ahead of 100 for see a fine movement. The buffer Load is 39% a 35khz of Kernel Speed.

With the version previous 3.037 i observe a fine and precise movement and the buffer Load is 19%, a 35khz of kernel Speed and Look Ahead of twenty.

Also the two versions have the following problems:

* A 60khz of kernel Speed o superior the movement of the mouse o the change of window can cause variation in the frequency of puses that originate jump and/or alarm in the axis.

* A 35khz the Feed Hold stop the axis a one distance superior which a 60khz, for example.

A 2000mm/min Feed Hold stop the axis in 4mm, this with 35khz of kernel. With 60khz of kernel
the axis stop in 2mm. the two a same aceleration.

I ran the Driver Test and was satisfactory.

I think that match3 take control of the PC almost all time, but can happen one time for day o one time for week that the window o the hardware of the PC interrupt the pulses.

this i proved with various computer that surpass the requirement minimun.

Maybe lack improve the real time software.

greeting

Maybe lack me intaller same software o driver that cancel the interrupt of windowns and the hardware.

Why the version R3.042.031 need more blocks Look Ahead for make the same work that the version R3.037? with the same PC and configuration.

The problem of variation o jump of pulses only happen me at kernel Speed superior a 60khz and move the mouse and change of screen.

At 35khz of kernerl speed almost never happen, also can happen when change the feedrate very quick of 100 a 170% (800mm/min) the axis X, Y and Z stop completely for 1seg and later return the movement, when this would should be continous.

I measure the pulses in real time with my driver, this can detect very high aceleration, with this i see that mach no lose any pulses, only happen jump.

greeting.