NEMA 23, weird velocity results.. need help

fannlow wrote:

When inputting in stepconf, we insert the number in servo thread or base thread?.

Base thread.

fannlow wrote:

NIn that case I assume servo thread is for servo motors and base thread is for steppers?

It's not quite that simple.
The "servo thread" is where all the complex calculations are done, and that runs at about 1kHz.
The "base thread" is where all the simple but fast stuff is done, and that runs at 10kHz to 50kHz.

Systems that use hardware step generators (such as FPGA cards) generally don't even have a "base thread"

Let me tests and see how it runs

Stepconf might still decide on a 100uS base thread, even if you tell it the latency is lower. I don't know what logic it uses.

fannlow wrote:

What else I can try?

What base period is stepconf giving you?

fannlow wrote:

Base period is 100, 000.

So change it!

Let stepconf finish the configuration, with quite a high max velocity.

Then edit the config files to change the base period (the reason it is choosing 100,000 is probably because you are only asking for 8mm/sec so it doesn't think it needs to be any lower. A bit of a chicken-and-egg situation).

Then start LinuxCNC with your new config and see how fast you can jog axes (adjusting the speed sliders).

I don't think that "test axis" is the best place to optimise the axis speed with your particular (fine leadscrew) setup.

fannlow wrote:

Steplen 1
Stepspace 0
Dirhold 56742
Dirsetup 56742

1, 0 is normal for a config that uses "reset" as a way to get a step pulse every cycle, rather than ever second cycle. That part is right.
The dir hold and dir setup seem strangely long, but they are in the region of one servo-period, so that's fine.

Also changing and saving .ini doesnt reflect out when I run the machine?
I have changed the base period to 50000
And up the velocity from 8 to 20 now and still same though..?
Something I did wrong?

In the screenshot the speed limit is still 8 for [AXIS_0]. Which one are you jogging?