Warner/Swasey QCM-6 lathe with Kflop/Kanalog

[TomKerekes]

Hi Charles,

Replacing the resolvers with encoders shouldn't have had any effect on the Drives. Something else must have changed. I would take a step back, electrically disconnect Kanalog and the Encoders, and go back to to 1.5V battery test until that works.

Maybe you have the X tach crossed to the Z Amplifier and vice versa?

You might disconnect the tachs and short the tach inputs and reference inputs to force 0V to the drives to see how it behaves. I would expect no or minimal motor drive.

I'm not familiar with those Drives.you would need to research and find a manual and specifications.

Hard to say about the sparks. A runaway is probably putting high currents to the motors. Sounds like you may need new brushes.

[Charles2]

Thanks Tom. I have taken the step back, disconnected everything. Working with only x axis in stand alone mode. I am in contact with Mike Kilroy, who knows these drives and motors. We are working through some things. Will update you when I know more.

[Charles2]

With the patient help of Mike Kilroy, It was discovered that there was a 5 volt difference in grounding between the drives (both) and the machine. This caused a 5 volt error in several test points on the drive cards, which caused motor run away and tripped breakers. After Mike supplied me with the most current manual, taught me how to use it, and coaxed me through many, many tests, we found that all that was needed was a proper ground from the drives to the machine. It seems that the drives were originally grounding through the factory GE2000 control. Now all appears to be well and I am ready to resume Kflop/Kanalog setup.

[Charles2]

Back to setting up Kflop/Kanalog. Servos work Fwd and Rev with battery. Connected the Z-axis encoder cable A+,A-,B+,B- to JP1, index + & - to JP2, ground to JP6, and +5V to JP8. Connected Z-axis servo drive output and ground to JP11. Z motor is drifting slowly towards lathe spindle. Axis screen shows encoder counting in negative direction. From Console screen I can move z axis both directions, although minus command moves away from spindle (DAC1=-20). In Step Response screen, Zero,100, Move results in axis moving towards spindle without stopping by itself. I have to hit servo Disable to stop it. Axis screen shows Position counting to and past Destination. Limit switches are connected to JP15 opto 2&3, bits 138,139. If I check Watch Limit boxes, bits 138,139, Kill Motor Drive, Move command will not work. I'm sure I have something wrong, but have not been able to find it. Have read and read other threads, Wiki, and manuals. Trying to get one axis working before connecting anything else.

[TomKerekes]

Hi Charles,

Connected Z-axis servo drive output and ground to JP11. Z motor is drifting slowly towards lathe spindle.

You might read this.

Z motor is drifting slowly towards lathe spindle

I don't understand what this means.

In Step Response screen, Zero,100, Move results in axis moving towards spindle without stopping by itself. I have to hit servo Disable to stop it. Axis screen shows Position counting to and past Destination.

Please post your Filters Screen, Configuration Screen, and Step Response Screen with a Plot of Command, Position, Output vs Time of the 100 count move.

Limit switches are connected to JP15 opto 2&3, bits 138,139. If I check Watch Limit boxes, bits 138,139, Kill Motor Drive, Move command will not work.

Do the bits change properly on the Digital IO Screen? What state are they normally? What state when hit? Do you have "Stop when low" selected?

[Charles2]

Thanks for answering Tom.

I should have said the Z axis is slowly drifting towards the spindle. I adjusted the speed pot on the drive and greatly reduced the drift. I have to figure out how to post screen prints. It is not plotting. Limit switches work when "Stop When Low" is checked. Bits do change properly. Switches are NC. Open when hit.

[TomKerekes]

I have to figure out how to post screen prints

One way is to:

#1 - Select a Screen
#2 Alt-Print Screen (copies current window to clipboard)
#3 open Paint.exe
#4 paste
#5 crop
#6 save as *.png file
#7 cnczone - Insert Image
#8 from computer - browse
#9 upload files

It is not plotting.

Not sure what you mean. No red, blue, green lines?

[Charles2]

Requested screen prints

[Charles2]

Nothing in the plot area. I'm sure it's me.

[TomKerekes]

Well it seems you are very Zoomed in and at someplace in negative time. Left-Click Drag will zoom in. You must have done that inadvertantly. Right Click on the Plot and select Zoom Out.

[Charles2]

ScrnPrnt.pdf

[TomKerekes]

Hi Charles,



Seems like classic positive feedback instead of negative feedback. As soon as there is any error, the motor drives the wrong way, error gets worse, quickly saturates output at -40 DAC counts, axis runs away. btw The Output limits at 40 counts is because max error 200 x P Gain 0.2 = 40.

Try reversing the sign of either the InputGain 0 or OutputGain to have negative feedback. Either one should be stable. Change either one. If the Axis works but moves the wrong direction, change the other instead.

HTH

[Charles2]

Changed the InputGain to -1. Axis did not visibly move, but it did plot.

StpResp4.pdf

Changed the OutputGain to -1, changed InputGain back to 1. Axis did not visibly move, plot is similar.

StpResp3.pdf

[TomKerekes]

Hi Charles,

Congrats you have Servo !!!



It is clearly trying to follow the commanded trajectory. See how the red tries to follow the blue?

A move of 100 counts is probably a tiny move so hard to physically see. What is the resolution of your machine in counts per inch?

Its nearly unstable and there is some small signal noise or oscillation. It would be helpful if you save and post the raw data. That way we can load, re-plot, zoom in, etc. to better see what is going on.

The next steps are to make bigger size moves and tune the servo. See here.

[Charles2]

Hi Tom,

Encoders are 4000 CPR. Axis motors are direct drive to .200 IPR ballscrews. I will plot larger moves and post the raw data.

[Charles2]

Hi Tom,

I changed the move to 1000 counts. It oscillates. Switching InputGain and OutputGain signs didn't appear to make a difference. See plot. I saved the data. I have to figure out how to post it.

Thanks

[TomKerekes]

Hi Charles,

Encoders are 4000 CPR. Axis motors are direct drive to .200 IPR ballscrews

There is often confusion whether CPR means Quadrature Cycle Per Rev or Quadrature Counts Per Rev. So the Resolution will be either 20000 counts/inch or 80000 counts/inch. I think we chose 4000 cycle/rev so 80000 counts/inch? This is easy to verify by moving ~1 inch and looking if the Position on the Axis Screen changes by ~80000 or ~20000.

I changed the move to 1000 counts. It oscillates.

Just needs tuning. I forget what type of Amplifiers you ended up with and whether you have tachometers connected or not. Do you know? It appears like the output is causing acceleration not velocity. Either would be ok but it is good to understand this. You might read this.

[Charles2]

Analog drives (amplifiers). Tach feedback from motors. It is possible that the tach wires are switched. Long story. The drives have a temporary jumper to limit current. I think the encoders are 4000 cycles per rev. I will verify this when I have better control.

- - - Updated - - -

Analog drives (amplifiers). Tach feedback from motors. It is possible that the tach wires are switched. Long story. The drives have a temporary jumper to limit current. I think the encoders are 4000 cycles per rev. I will verify this when I have better control.

[TomKerekes]

It is possible that the tach wires are switched.

I think that is unlikely as I would expect a runaway.

Actually from your plot in post 32 where the output is stuck at 40 DAC counts (0.2V) the Position moves at a fairly constant Velocity. That would indicate the Amplifier is controlling velocity on its own. Otherwise I would expect more continuous gradual acceleration.

You have a fairly high resolution encoder. This effectively adds gain to the control loop. So try reducing the P Gain by 10X or more to get stability.

[Charles2]

Hi Tom,

As you suggested, I reduced P Gain from 0.2 to 0.02.



Changed Move to 10000.



Changed Move to 100000



Changed P Gain to .002.



Hopefully I kept the screen shots as described. Check the setting boxes to be sure.

Thanks