[bonestock]

I am having some issues with my servos moving when no signal is being applied.

I have a Teco TSTA-20C Drive with a TSB08751C 750W servo. My troubleshooting has distilled down to running nothing more then 240V 3 phase power to the drive and just the minimum of pins jumped on the CN1 connector so as to enable the servo.

These drives are pretty close reproductions of the Yaskawa Sigma II so I have been scouring the Yaskawa documentation for any helpful troubleshooting tips with little success.

The drives are configured to run in Speed mode. I have jumpered the Analog signal input to the Analog ground on the Drive which according to the teco and yaskawa documentation should provide me with a 0v signal. Additionally I run the auto-offset adjustment so that the drive can compensate for the analog command voltage, which should be 0v as just mentioned.

Following the steps above per the documentation should yield me a stationary servo, but what I have is a very slow bouncing between counts and a gradual CW or CCW servo movement.

On further inspection I see is that offset value being set from the above auto adjustment is around 16mV. Manually setting this value +/-1mv only starts the servo spinning faster in one direction or in the opposite direction all together.

I have checked and even tested an alternate ground connection to rule out any possible grounding issues.

Some of the questions I am asking myself are: How sensitive are these or the Yaskawa drives to noise in the input power? Are filters absolutely necessary with these Drives/Servos or other AC powered servos? Are the problems I am having related to external noise or something internal to the drive? I am focusing mostly on line noise due to the fact that these problems only occur when the servo is configured in a mode to accept analog +10/-10V input.

[Al_The_Man]

It sounds like you just operating the drives in simple open loop control? If so you would normally get this kind of drift?
When used in conjunction with a CNC control where the encoder loop returns (or also returns) to the control the position is constantly updated in µsec via the PID loop, so you would not see this drift occur.
Al.

[hub]

I'm driving my spidle (through a VFD) using analogue 0-10V.
The RPMs are not perfectly stable, they vary a little.. In my case it's not so bad if the spindle turns +- a few rpm wring.
But in your case it matters because you want positional accuracy. If the drivers only accept analogue signals then you really need to connect encoders to correct the errors in voltage..
Digital control would be a lot better if the drives support it.. Even then, encoders would be good.

[bonestock]

It sounds like you just operating the drives in simple open loop control? If so you would normally get this kind of drift?
When used in conjunction with a CNC control where the encoder loop returns (or also returns) to the control the position is constantly updated in µsec via the PID loop, so you would not see this drift occur.
Al.

Does this also pertain to the Ajax MPU11/GPIO4D boards? I keep getting faults due to the drift which is what started me on the path elimination, troubleshooting and tuning the drive and my input signal.

If the drivers only accept analogue signals then you really need to connect encoders to correct the errors in voltage..
Digital control would be a lot better if the drives support it.. Even then, encoders would be good.

I would prefer digital the drives support it but I do not believe the MPU11 or its addon boards will easily support it.

[Al_The_Man]

I have never used that system, but don't they have their own DC servo drives?
Are you hooking some other servo unit to the Ajax?
Or are you using this drive as a spindle??
Al.

[bonestock]

I have never used that system, but don't they have their own DC servo drives?
Are you hooking some other servo unit to the Ajax?
Or are you using this drive as a spindle??
Al.

This is the Ajax system for 3rd party drives. I am using this servo for my Z axis and similar but larger servos for my X and Y.

Reading over your response of the PID loop and your various posts related to servo tuning over the years got me thinking. I was failing to see I have a PID loop nested within a PID loop and did not completely understanding the function of each.

These servo drives when set to speed mode have an internal PID loop only in reference to speed and not position. When I removed the controller in my troubleshooting I removed the outer PID loop that should have maintained position.

In navigating the servo drive diagnostic menus I noticed that I have a constant -20mv command signal when the controller is not connected. This signal completely explains the drift. With no outer PID loop to correct the position the drift continues.

I am now suspecting that the encoder was not wired properly to begin with so when a correction signal was being made from the controller the servo moved in the incorrect direction resulting in a fault. I will reconnect everything this evening and hopefully lay this issue to rest.

[Al_The_Man]

I am now suspecting that the encoder was not wired properly to begin with so when a correction signal was being made from the controller the servo moved in the incorrect direction resulting in a fault. I will reconnect everything this evening and hopefully lay this issue to rest.

Sounds Logical!
Al.

[bonestock]

After reconnecting the motion controller I found that the encoder's absolute position was incrementing in the opposite direction than the controller was expecting. Flipping my A and B signals resolved that.

Additionally I had another small problem in which I needed to configure the servo drive to reverse the direction of the command signal. It's working great now.

As Al pointed out the controller closes the PID loop and compensates for the drift.