Messages

1

Mach4 General Discussion / Mach 4 G01 interpolation

« on: August 23, 2016, 11:55:09 AM »

It seems that under certain situations, Mach doesn't brake before feeding into a totally different direction. I was milling simple squares, when I found out that, upwards from a certain feedrate, the corners of the squares became visibly round, but only certain corners (more on that later). After looking a bit more thoroughly, I found out that no matter what feedrate I was using, the round corners were there, but their size varied with feedrate. With an 8mm bit, 1500mm/min yielded small, maybe 0.1mm radius corners, 3000mm/min made 2mm radius corners. I'm pretty sure that Mach interpolated the toolpath to smooth out the 90 deg corners.

I had the milling separated into subprograms, the X and Y edges of the squares were separated (the two subprograms themselves only did a part of the respective edges, they were repeated multiple times). The interpolation only took place whilst the g-code executed was inside the subprogram, but in between two subprograms (or repeats of the same) the interpolation didn't take place. This lead me to believe that the error wasn't in the servo drivers, or in the mechanics of the CNC, but rather in either Mach 4 or VitalSystems's DSPMC controller (Couldn't rule it out).

Is G01 supposed to to this, or did I miss something like syntax? Interpolation seems like a thing that could be useful for smoothing out curves with small angles, but a 90 degree corner should be a 90 degree corner. G00 seems to work fine, braking just in time to stop at exactly where it was supposed to travel to.

Here are the subprograms, for Y and X edges, respectively:

Code: [Select]

%
O1337
G91
G01 Y-784 F1500
G01 X28
G01 Y784
G01 X28
M99
%
%
O1420
G91
G01 X-560 F1500
G01 Y28
G01 X560
G01 Y28
M99
%


2

dspMC/IP Motion Controller / Can't get servos to move

« on: April 21, 2016, 02:24:44 PM »

I'm currently retrofitting a large chinese CNC. I've replaced the chinese motion controller with Vitalsystems' DSPMC and Mach 4.

Here's the problem: I can't seem to get the servos working properly, if at all.

Here's what I'm working with:
Mach 4
DSPMC/IP with 3 pn7535 boards, 1 pn7737 board and 1 pn7711 board. The 7737 connected to J12 and configured for that in the VSI Manager.
Yaskawa SGDM-ADA servo drivers, 4 of them(X,Z,Y x2)

The symptoms of the issue are rather weird and inconsistent. After physically connecting everything a described in Vitalsystems' manuals and configuring everything as described and as I thought to be correct, I tried getting the servos to move using jog mode in Mach 4. Of course, it didn't work. After a lot of troubleshooting I finally traced it to inconsistent step/dir signals from the motion controller. In order to have a reference, I quickly connected up the old chinese motion control system, and had a look at the step/dir signals using an oscilloscope. The chinese one could move the servos without problems, despite the horrible ringing of the step signal (before you ask, it was using the same position control mode and the same signals to control the servo driver). It was very responisve to jog commands, the signal's frequency changed as it should, accounting for proper, smooth acceleration. The direction signal changed as it should, without any delays. In contrast, after hooking up the oscilloscope to the signals generated by the DSPMC, the signals were horrible in their content. The signals were incredible in actual signal fidelity (if they actually were there), the edges were fast and there was no ringing, even though the total cable lenght from the DSPMC to my probes was way over 3 meters. The horribleness was the signal's content. Sometimes, the first movement command (from a standstill to a complete stop) was perfect, only to be followed by a step signal without proper acceleration, no respective change in the direction signal, nothing, or total garbage. The last being rather infrequent, possibly not being caused by the main problem but something else. The direction signals was jerky too, it sometimes didn't change at all, sometimes too late. Same with the step signal, it just wasn't there at times. But here comes the even weirder thing: the first movement wasn't always there, requiring re-enabling of Mach 4 to come back, and only using step/dir channel SG5 of the DSPMC. I can't really recall what happened on the other channels besides SG5 when probed.

The power supply GND is connected to the chassis, and therefore to mains earth. The total cable lenght from the DSPMC is about 1.8 meters (1.2 from the DPSMC to the 7737, and 0.4 from there to the driver). The cable lenght shouldn't be the issue here, since the signal integrity was great, even after using a longer, 1,5m ethernet cable for probing (which adds up to about 3m total). Induced noise from the 380V stuff around the electrical cabinet, maybe. Since it was a retrofitting project, there was no efficient way of separating things as good practice might advise to. Still, the signal was quite clean of noise.

I tried using an example g-code to rule out the jog mode being at fault. It did a little better, now doing 2 proper moves and then sometimes leaving the servo moving at a low speed. It still failed to move it after these two though. I also modified the motor's acceleration curve to be a little less steep.

Could it be caused by the fact that I haven't yet applied the license key to Mach 4? I have one, but I'm not going to apply it until the machine's working if not necessary.

The computer's quite a beast, 4 core i5 2400, and 4GB of RAM. That can't be the limiting factor, unless something's incorrectly configured. Did a clean install of win 7 pro, and it hasn't got an internet connection, so Mach4 shouldn't be interfered with either.

Some screenshots of the config windows I though could be useful:
https://drive.google.com/file/d/0B_yAWIhEioTZS2N5ZXlfb1duaU0/view?usp=sharing
https://drive.google.com/file/d/0B_yAWIhEioTZcFFGNGc1cDN0cDg/view?usp=sharing
https://drive.google.com/file/d/0B_yAWIhEioTZclRpOXk5NTl1SHc/view?usp=sharing
https://drive.google.com/file/d/0B_yAWIhEioTZREt6NlgzYUVEMlE/view?usp=sharing
https://drive.google.com/file/d/0B_yAWIhEioTZUnVtamNsOWtHSUk/view?usp=sharing
https://drive.google.com/file/d/0B_yAWIhEioTZQzY0UG1UOHp0Q0U/view?usp=sharing