Jerking in motion on same points of trajectory.

[CT63]

Before uccnc I was using mach 3, Ethernet Smoothstepper and Gecko 540 and never experienced a missed step. Changed over to ETH400, Gecko 540 and cucnc and experienced at least 3 occasions of missed X axis steps. Same pc used for both software. I know there could be many reasons and because it's random, it's impossible to determine. The only observable factor is it seems to happen when the X axis is moving towards the home position at the same time the Y axis is moving while running gcode. I switched the X and A axis driver cards in the Gecko and still had the random missed step. All of the gcode programs are generated with Aspire.

[cncdrive]

The UCCNC is designed in a way that the steps sent out are feedback as the coordinate to the PC. It was designed this way for easy debugging.
So if using servos then it is relatively easy to see if the UCCNC caused lost steps or not, because if the UCCNC coordinates are where they should be then it is sure that the UCCNC sent out the proper amount of steps.

However with steppers it is harder to know, you can still verify the coordinates, but a stepper works in open loop, so it can loose steps from too high running speed where there is not enough torque, or too high accelerations/deccelerations or can be a mechanical problem which causes higher torque requirement on the axis, or a combination of these. Or it can be a powersupply problem that the Voltage drops down too much when more axis moving together. Or if could be the capacitors in the powersupply not holding the voltage (electrolit dry out) causing more EMI which causes lost or extra steps. It is hard to find the reason out.

With the UCCNC we run complex test g-codes before each release with a counter electronics which we built only for this purpose connected to the step/dir lines of all axis and with that we always verify that the motion control API and the firmware is fine, sending our the correct amount of step/dirs and not causing lost or extra steps and so far we never had a problem with this.
The same electronics also calculates the accelerations and the deccelerations and registers the higher values so we can verify if anything runs out of control in the motion planner or the step/dir generator.
We take the about 4-5 hours of time per motion controller to run this test every time we make a new API version and before we officially release them.

[Derek]

One thing to remember. Machs trajectory planner is totally different from UCCNCs. So what worked in Mach as far as motion is concerned may not work the same in UCCNC.

I converted 4 machines over from Mach. three servo systems and one stepper. The stepper needed the most tweaking and diligence.

[beefy]

One thing to remember. Machs trajectory planner is totally different from UCCNCs. So what worked in Mach as far as motion is concerned may not work the same in UCCNC.

I converted 4 machines over from Mach. three servo systems and one stepper. The stepper needed the most tweaking and diligence.

Dereck,

any chance of expanding on that a little. What sort of "tweaking" was required for steppers. At the moment all I've ever used are steppers.

Keith.

[A_Camera]

One thing to remember. Machs trajectory planner is totally different from UCCNCs. So what worked in Mach as far as motion is concerned may not work the same in UCCNC.

I converted 4 machines over from Mach. three servo systems and one stepper. The stepper needed the most tweaking and diligence.

What do you mean by that? I went from Mach3 to UCCNC without any tweaking, just straight forward importing of the configuration.

But yes, just because a G-code worked on Mach3 it does not mean it works on UCCNC, since UCCNC does not support every command, but that is pretty clear when you load the code and is not related to stepper vs. servo.

[ger21]

I would think he's referring to the CV settings?

[Derek]

One thing to remember. Machs trajectory planner is totally different from UCCNCs. So what worked in Mach as far as motion is concerned may not work the same in UCCNC.

I converted 4 machines over from Mach. three servo systems and one stepper. The stepper needed the most tweaking and diligence.

Dereck,

any chance of expanding on that a little. What sort of "tweaking" was required for steppers. At the moment all I've ever used are steppers.

Keith.

I had to work with the accelerations so it wouldn't stall. Keep in mind it may have been marginal with Mach as It is my first and only stepper system. It's underpowered for what I'm trying to do. What I was trying to point out is the planer is completely different. It's not like turning an input on and off. So converting from Mach isn't simply importing the old mach settings and letting it rip. You need to tune your system like it was new.

[A_Camera]

I had to work with the accelerations so it wouldn't stall. Keep in mind it may have been marginal with Mach as It is my first and only stepper system. It's underpowered for what I'm trying to do. What I was trying to point out is the planer is completely different. It's not like turning an input on and off. So converting from Mach isn't simply importing the old mach settings and letting it rip. You need to tune your system like it was new.

Perhaps I misunderstand you, but you make it sound over complicated, which is not. If you have a stepper based CNC running under Mach3 you can run it with EXACTLY the same settings under UCCNC as well. It will work just as well, in fact, even better, also if you don't do any tweaking. So, what sort of tweaking you had to do to make it work? Of course, since UCCNC has a better trajectory planner, you can tweak it so that it is EVEN better, but that is really not necessary, or am I missing something? UCCNC outperforms Mach3 in CV mode, and performs equal in exact stop mode.

Of course, I assume that you have had otherwise the exact same hardware, like the motion controller as well, otherwise it is apples and oranges... but if your motors stall under UCCNC they stall also under Mach3.

[Derek]

I had to work with the accelerations so it wouldn't stall. Keep in mind it may have been marginal with Mach as It is my first and only stepper system. It's underpowered for what I'm trying to do. What I was trying to point out is the planer is completely different. It's not like turning an input on and off. So converting from Mach isn't simply importing the old mach settings and letting it rip. You need to tune your system like it was new.

Perhaps I misunderstand you, but you make it sound over complicated, which is not. If you have a stepper based CNC running under Mach3 you can run it with EXACTLY the same settings under UCCNC as well. It will work just as well, in fact, even better, also if you don't do any tweaking. So, what sort of tweaking you had to do to make it work? Of course, since UCCNC has a better trajectory planner, you can tweak it so that it is EVEN better, but that is really not necessary, or am I missing something? UCCNC outperforms Mach3 in CV mode, and performs equal in exact stop mode.

Of course, I assume that you have had otherwise the exact same hardware, like the motion controller as well, otherwise it is apples and oranges... but if your motors stall under UCCNC they stall also under Mach3.

Since you seem to be an expert at these things I'll defer to you at this point.

[CT63]

Ok... I've been beat up pretty bad over my post, not only from several members, along with told by CUCNC admins that missing steps because their "software is tested" and "can't happen", so I will pose the following:

Missing axis steps while moving in the direct to the home position on one axis when two axis's are moving, while running gcode does happen. I assumed it was a fluke but after 3 incidents it's not a fluke. I'm not the only one to experience this. It is a real problem.

I'm not new to the cnc steppers world as member Derek would like to you to believe, while I value his comments he is uninformed and making assumptions based on no facts to support his comments. I understand the multiple reasons why a stepper misses steps. I've been a long time advocate and supporter of uccnc... but to be told "it can't happen with our software" by uccnc reminds me of ArtSoft's attitude of screw the customer.

I hope this isn't cucnc's new business model to customer service.