Hello Guest it is March 29, 2024, 01:26:49 AM

Author Topic: Why is there no effort for the true closed loop control on the mach3 level  (Read 20486 times)

0 Members and 1 Guest are viewing this topic.


I am building machine tools and have access and make use of   Heidenhein itc530 (germany) , Fanuc 18imB (japan), sytnec (taiwan) , LNC (China), GSK (china).  All use the true close loop control such as following error and velocity error in the Gcode level as standard. I am well informed that Mach3 is open loop controlled, But I wish that the MACH3 software come to the level of above with the help of improving technologies such as external motion hardware and availibity of very fast interconnection standards such as USB3.0 (5gbps) and sata III (6gbps).

What are the main concerns?

- Unreliability of USB3.0 and sataIII for real time control and feedback

-Development effort is huge that a 1-2 man team can not handle.

- Is the path too long (the feedback should come first from the servo to the motion control hardware and then transferred to the Mach3. ). How does heidenhein do this with a celeron 400Mhz PC and a motion controller?


Does a concept exist towards thşs implementation for future maybe in 3-4 years.  Or this is totally another type of business or method that does not/ can not belong to the future of Mach3.

Best regards.

tstbeyaz

Offline BR549

*
  •  6,965 6,965
    • View Profile
ALL those controllers use an independant MOTION controller for closing the loop. Some of them even run 2-4 processors JUST to keep up.

EVEN then NONE of them do axis corrections to correct a falut. They are just using a PID to help the servo keep up with the motion data stream. Let that servo LAG far enough and it just faults ands stops the controller.

You can have that with steppers in MACH. Rogers machine sells the board and provides the plugin to drive it. WORKS VERY WELL

You can have that same feature in mach by running a digital servo drive( the loop is close in the drive ) OR use a motion controller JUST like the big boys and clos it on the MC card.

NONE of them cost $175.00(;-)

IF you want Free closed loop PC control you have EMC2(;-)

Take your pick, (;-) TP
« Last Edit: January 08, 2011, 12:03:51 PM by BR549 »

Offline DaOne

*
  •  236 236
    • View Profile
I think the main reason why we don't have closed loop is it would require a major rewrite. I could be wrong but I think running on windows is the biggest problem. Timing issues are most likely the culprit.

Best bet is a hardware controller such as a galil. That closes the loop in the hardware control and will give you the same results as a real control.

Offline BR549

*
  •  6,965 6,965
    • View Profile
YEP you have many that can do the job.

Galil
Dsmpmc
Mesa
Kflop
I think I have missed the point that Kflop/Mesa/Dsmpmc/galil are more than smoothsteppers. It is good news so that I may start to investigate and try some setups thncoder inpute ones above. Do these cards have a seperate encoder input,   Do they function basically so that they get the encoder data from the servo drive without requiring to install additional encoders other than the ones readily installed at the back of servomotors. I believe I must investigate how they communicate with the servodrive control. Any comments which one has broad support for common servo drives from Fanuc/alfaseries,  Panasonic and Mitsubishi.

What abaout look ahead buffer, and increasing and decreasing the velocity dependant on the path and the inertia/power rating of the system. cancelling the jerk (ensuring that the derivative of the acceleration is not sharp to avoid vibrations) . How are these tasks coordinated (through mach3 or motion control card or mixed handling)

tstbeyaz.

Offline BR549

*
  •  6,965 6,965
    • View Profile
Mach has a simple bang bang trapazoidal planner you get what you get.

Most MC cards except the servo encoder signal and close the loop at the MC card.

As far as advanced motion control like you want?? I dought it. Most are used to control a Dumb servo amp that excepts analog +/-10v.

Most simple MC's just emulate the input signals they are given.  THE vey high end ones can convert the simple input into a complex output based on parameters like you wanted.

BUT that is why high$ controllers are high$$$$$ controllers.

Just a thought, (;-) TP

Offline rcaffin

*
  •  1,052 1,052
    • View Profile
Quote
- Unreliability of USB3.0 and sataIII for real time control and feedback
I don't think the reliability of USB or SATA is the issue. It's the UNreliability of Windows that is the biggest problem!

Some other points should be made here as well.

The older closed-loop control was created before the n86 chips reached their current level of speed. They relied totally on dedicated hard-wired logic chips to do the motion control for servos: there was no other way to get the control for linear servo motors.

Mach is, to some degree, based on the concept of a stepper motor. You do not (really) need closed loop control with steppers: either they have the power to drive the axes, or they don't. Feedback will not change this.

Working forward from the idea of steppers, some drive units for linear servo motors now emulate a stepper. You feed them 'stepper pulses' and they convert that to DC (or AC) servo drive. If the system is working OK all is well; if the servo drive cannot cope it flags a fault and Mach3 stops. This is conceptually the same as a stepper system.

You are right in your statement that SmoothStepper has nothing to do with this. In this context it is just a logic interface, nothing more.

Now, going on to smooth changes in acceleration (3rd derivative) - that is more difficult. If it was really necessary then customer demand might lead to Mach including it. But normally you get this for free from the power amplifier: it cannot produce instant acceleration in practice, as there are peak power limits. So for most people there is no need for Mach to do this.

Cheers

Offline kf2qd

*
  •  148 148
    • View Profile
Plain and simple - COST. Mach is a very nice controller at a very inexpensive price. It has the features that are appropriate for its cost. If you need these other features - Closed Loop Motion - you will have to buy (or invest the time and use EMC2) a controller at a higher price than a 'hobby' controller. Mach requires a minimal investment in hardware and thus you trade off some options - fully closed loop - for lower cost. Some have been able to use it successfully in production environments, but it is still a "hobby" control.

Offline rcaffin

*
  •  1,052 1,052
    • View Profile
Some have been able to use it successfully in production environments, but it is still a "hobby" control.

A bit more than 'hobby' today. One can do serious production with it IF you know how.

Cheers
I retrofited am Emco lathe using Mach 3 and it works quite well in an industrial environment.  My opinion is that if some aditional money is spent in industrial components such as servos, VFD and a PLC then one can come up with a decent system.  Industrial grade servos allow for closed loop (with their drives) and homing using their index pulse.  A VFD with braking resistors allow for good spindle control and the PLC is necessary for processing information from sensors, generating alarms and controlling periferal devices such as the hydraulic unit, tool changers, part catchers, bar feeders etc.