DC Servo gives me the "creeps"

[zoeper]

Hi,
I was just busy writing my first program down on paper when the machine tripped out without me doing anything! The machine was standing at the workpiece datum but after it tripped it seemed that the Y axis has crept all the way up to the machine column while the display was still showing the datum coordinates. since i was standing only a few feet away I am sure i would have noticed if it jumped or made a sudden move. I did notice that the y servo motor (SEM dc servo) was making a verry sharp whistling sound since i got the machine a couple of days ago and it was not very responsive when i moved the belt by hand and when at rest it was going forwards and backwards all the time.
Would this motor just need tuning? On the Anilam servo amp for that axis there are two pots (presumably for tuning)
What is the procedure for tuning dc servo's and what tools are required?
How is it possible for a motor to creep without the encoder/ Tacho picking up on it?
I would like to get this machine cutting soon and really apreciate your input.
PS there is quite a lot of muck collected on the servo amp cards. Is it safe to clean them or should they rather be left alone?

Pieter

[NC Cams]

Mucky cords: as long as you clean them with the power off, you should do fine. BrakeClean is a decent solvent BUT it may remove labels or some ink identifiers. Wipe the wires down with paper towels to get big muck/oil off them and then give them a shot with brake clean.

Motor hunting/buzzing issues: the motors can "hunt" when idle if the tune is off. My understanding is that the servo gain is asking for so little correction to achieve the position that the thing is buzzing back and forth. That or else it is asking for correction that essemtially isn't need which is why it "hunts"

The tach and encoder won't pick up on it if it is merely buzzing as the motor probably hasn't rotated enough to either create a tach signal or made enough of an encoder change to generate a corretion requirement.

How you adjust/tune the amp depends on who's amp it is. Ditto that for the tools needed to adjust. In some instances, a voltmeter is all that's needed. In others, you may need an oscilloscope. In some Bridgeports, there is a tuning program that will endable you to effectively tune the servos while in DOS and not every turn a pot on the amps.

You should hope that Machintek picks up on this thread - he's the expert in the servo tuning dept and if anybody has insight, he should.

[Al_The_Man]

I am suprised you only have two pots if it is a Velocity amplifier (tach feedback).
Normally you would have at least four,
1/ Input Gain
2/ Tach Gain
3/ Balance
4/ Current Limit
The two pots you have will be two of these for sure.
These velocity amps consist of two loops, the inner loop which is motor/tach/drive and the outer loop which is Motor/encoder/controller.
The problem is most likely with the inner loop.
Because the inner velocity loop tach feedback is sloppy around zero speed it is common to crank the gain to hold the motor in position at zero, this results in the amplifier outputing alternating DC of equal amplitude at zero, resulting in the 'Singing' which is normal with these drives.
The inner loop is always tuned first, and in the absence of a tuning procedure, it can be done crudely by disconnecting the analogue input signal and either shorting the input, which should result in zero speed and adjusting the balance for zero creep, and then gain until the motor oscillates back and forth and back off half a turn.
For this you have to identify which pot is which.
I usually use a battery box, this way I can test the drive with a ±10vdc signal.
These type of drives, the pots, especially if 10 or 20 turn, often go bad contact etc.
The normal reaction when you try and turn the motor when connected to the controller, should be that it supplies increasing resistance and the high pitch sound will increase in intensity.
Al.

[zoeper]

You're right Al, the two pots mentioned are single turn trimmers and then there are two other multiturn pots also - sorry i missed them first time round!

I caught the machine creeping again today and it seems like it only does it when you set the tool length and diameter in the tool table. After entering the tool data the machine displays a busy message for a couple of seconds as it loads the tool data. (normal acording to the manual) but then the controll seems to hang as it does not accept any further input and all 3 axis starts to creep. (the display coordinates does not show any movement at all, but the DTI does not lie!)
X moves at 38s/mm + dir
y moves at 30s/mm - dir
z moves at 19s/mm - dir
Controll as an Anilam GXM
Servo's are SEM dc servo's
Tacho's for feedback and the servo Amps are Anilam

I hope this rings a bell for someone, as i can only think it must be a controll thing (touch wood)
Pieter

[Al_The_Man]

It almost seems like the controller loses contact with the drives and they start to creep due to balance slightly off which is almost impossible to get it to remain dead constant due to temperature and component drift.
Normally you would not notice it if the controller remains in control and corrects it.
The current trend has gone away from velocity drives and were replaced by torque style amplifiers, with these you can do away with the tach feedback.
I am not sure if anyone has replaced the Anilam drive with a AMC or similar torque replacement drive, but it may be an option if you have a drive go bad.
Al.

[zoeper]

Machine is doing nothing other than creeping and tripping now.
I've cleaned the y axis drive belt housing and had to turn the servo by hand. Since switching on again the machine would creep in Y and Z for about 10seconds(100 micron) and then trip out (error code "servo dscnct"). Could this be caused by the fact that the axis moved while not energzed, or is it coincidence with the problem lying elsewhere? How can i test the system components individually to eliminate them as suspects?
P

[NC Cams]

Problems like you have can involve a process of elimination. To do that, you start at one end, evaluate the component and rule it out.

In many cases, you can have several items that are screwed up and/or intermittantly malfuctioning.

By all means, check the stuff individually - you may be surprised at what you'll find in the process.

[Xerxes]

It is probably an encoder problem. There migh be inferference or noise in encoder lines to cause false counts and position drifting. Controller thinks it is in position since it relies on encoder information and if that information is false, then position is incorrect.

[zoeper]

Thanx NC, that is what i had in mind. What i would like to know is if anyone has suggestions of how to test the individual components. As it stands now i switch the sytem on, it stays on for 10 seconds and then trips. During those rpecious 10 seconds you cannot execute any commands as the system is not ready to accept them.
For instance, how do i rule out the servo/tacho/amp combination?
P

[zoeper]

Xerxes,
If it is encoder problems (i'm using tacho's), why is it drifting in more than one axis? When i first detected the problem the display was not picking up on the drift, but now it shows it on the screen also?
P

[Al_The_Man]

Originally Posted by zoeper Xerxes, If it is encoder problems (i'm using tacho's), why is it drifting in more than one axis? P

If it just does this at power up time, it almost appears the drives are enabled, allowing the drift I mentioned earlier, but the system has not completed boot up and is ready, therefore does not yet have control.
It is not normal to enable the drives until the controller is ready, likewise if a fault occurs in the controller it should shut the drives off.
If all axis are drifting and the controller, for some reason is not IN control, when it thinks it should be, it could be a fault in the analogue signals not being output.
If it were a problem on the common encoder supply, you would get a encoder feedback fault immediatly or the analogue signals would climb to try and correct this and the drives would run away.
The fact it is in all axis points to something in the controller, or your drives enables are on way too early.
To test the drives individually, you can disconnect the analogue signal (two wires) in to the drive and short the analogue drive input Terminals together, with the drive enabled, there should be zero or next to zero motion.
If you had a small DC supply of a few volts (<10v), you could drive the motor by input to the drive analogue terminal, but it is best to disconnect the motor belt or? first.
Also if you can artificially disable the drives enable-pin at power up, (the controller won't know you have done this).
The servo's should not move so the drive should not put out any correction signal, so you should see zero voltage on the analogue command signals to each drive.
Al.

[Xerxes]

Originally Posted by zoeper Xerxes, If it is encoder problems (i'm using tacho's), why is it drifting in more than one axis? When i first detected the problem the display was not picking up on the drift, but now it shows it on the screen also? P

Are you using tachos as only feedback devices for positioning? Then the reason is clear. Tacho gives velocity signal and to get position, velocity must go thru integrator. Even smallest error or inaccuracy in velocity signal would lead to drifting. So unless you have ideal tachos, you need also some position sensor.