Mesa 5i20 servo configuration

This is turning into a greater challenge than i anticipated.

Tuning has become a bit of a problem I suspect due to nonlinearities in the linear motor.

Let me describe the setup.

The motor is a NORMAG ( northern magnetics of late owned by Baldor) iron cored linear motor.

The servo amp is a Science motion BLDC unit operating in current mode. Peak current is 6 amps at 140V DC with rated current at 33% of peak.

Encoder is a linear glass scale with 5 micron ( accounting for quadrature ) resolution.

Controller is a Mesa 5I20 , 7I33 and a home made breakout board consisting of 50 pin header similar number of scrw terminals and an adjustable 5 volt supply ( +/- 0.5 volts) based on LM317 regultaor.

Tuning software is a hal file found in one of 5i20 sample applications supplied with EMC2 and edited to suit my single motor test bench.

The file was set up with square wave sig geny to enable me to adjust the loop for optimum transient abd steady state error.

The software works a treat over short distances of motor movement ( corresponding to areas where cogging forces are minimal.)

On longer strokes the motor will reach programmed destination within four to ten counts of the encoder. It will th over a period of a second or so find the final position as attested to by the DRO and audible clicks coming from the motor as it homes in on programed value.

I guess I need to increase the integral gain to minimise this final homing in section. I am however worried I shall upset the loop elsewhere.

Any comments on how to overcome nonlinearities welcome.

Sounds like maybe the current control loop in your amp needs adjustment
(current=torque control is usually via a PI loop in the amplifier)
low gain in the current loop can cause cogging as the current loop is used to compensate
for non-linearities in the PWM voltage stage and low gain limits the ability to fix these errors

Another possibility is that your amp doesn't match your motor (maybe trapezoidal vs sine commutation issue)

A number of things have happened to prevent me from progressing with the experimentation.. that is till now.

I am in the process of building a couple of frequency to voltage converters ( tachos).

They will receive encoder quadrature signals from the IO card and generate analog velocity signals ( +/-10V) to provide velocity feedback signals to the servo amps.
The servo amps can operate in both voltage and current modes ( switch selectable ) .

There is a small delay in the tacho circuit of about 3 ms or more precisely the time constant of the tacho circuit is approximately 3 ms. I suspect this may present some problems during acceleration and deceleration phases.

I wish to set the motor acceleration to about 10 m/s^2. While this seems reasonably high the only load the motor will push is a work platten of about 10kilos and the motors can provide thrust of 30 lb continuous and 90 lb at 10% duty cycle.

Production speed of 100mm/s is present target.

Hope to have some additional info on the tach feedback performance during the coming week.

Well Board,

Four months later....

A bunch of things had transpired since. My father a terminal cancer patient passed away last week of May. We had set up a "ward" at home to be able to look after him and provide him paliative care at home. He wanted to live his final days at home not in a hospital ward. Things have settled down somewhat since and I have finally managed to make some progress.

I designed and built a duak channel quadrature decoder and tacho circuit which has been used to convert the current drive servos to velocity drive .

I am having some issues in getting the Iron cored linear motor to behave and operate at low speed with a small enough following error.At the moment about 3000mm/min is the lower limit at a following error setting oh 50 micron.

I guess I shall be playing arounfd with the loop a while longerto improve on this. In reality the task to hand calls for coordinated motion at about 5000 to 6000 mm/min.

Acceleration is set to 4000mm/s/s and I have had it operating at 10 000mm/s/s . It moves like a cut cat.

More on this soon.

Hello Andy,
Thanks for the comment and ideas.

Some months back there was a mention of the possibility of using the emcoder velocity output to stabilise the loop.
At the time I got the impression that the velocity component was not generating sufficiently stable and accurate velocity component hence the external hardware solution.

The current situation is that I am getting an external hardware velocity feedback signal to the servo.
The servo has a gain and itegral gain setting .

EMC2 n the other hand does thepositional loop control whihc generates the analog voltage to vcontrol the external velocity loop.

So far the results are far better than the current ( force) drive behaviour I had had in the past.

More progress and yet another onion peel to chew through and get rid of new issues.

Motors are operating reasonably well. Perhaps at a lower acceleration and velocity than i had hoped for but never the less well enough to perform tasks to hand.

MAJOR

problem is that the following error is still not low enough.

My expectation on minimum following error would have been down in the 50 micron region.
Presently minimum following error is 100 micron and even at this setting the loop drops out every now and then even when stationary especially after completing a move.
Stopping the program ( .ngc program) also causes the motor to trip out on following error.

ONLY PID coefficients are in use at the moment.

50 micron barier no more.
The motor is behaving quite OK.
Next stage enable homing on the index pulse.

More to come soon.

Index latches.

The secret to getting index only operation is ofcourse section 7.2.3 ( a big thanks to PCW).
Searc velocity must be set to zero.

Almost there...

Now that the motor is homing on index time to set up the machine coordinate system.

The axis in question is working back to front; positive and negative directions need to be swapped.

Other than altering phasing of motor and hall sensor wires i stumbled along trying to understand which pin and more importantly how to reverse the direction of output on PWM 0 generator.

Any hints and or pointers welcome....

PLEEEZZZZ