Induce servo vibration, or how to achieve 100+ Hz controlled vibration 1

[RyreInc]

We have a servo press, directly coupled ball screw, that we would like to vibrate at 100+ Hz at a specified amplitude (0.010" range) and for a specified time (1-5s or so). I have been able to achieve up to about 30 Hz but the method I'm using performs a discrete move every scan of the PLC and so I am limited by its scan time. Even a dedicated task with sub millisecond scan times would have limited bandwidth.

Is there another way I can create a controlled vibration? Perhaps using tuning parameters? How would I calculate the frequency and amplitude?

If there is not a way to do this using the servo, what additional hardware will get me there? I am aware of voice coils but how will this work with a press?

I am struggling to find anything on Google - everything related to servo vibration seeks to eliminate it, not create it.

 

[IRstuff]

HAve you considered:
> Arduino with Modbus?

http://playground.arduino.cc/Code/ModbusMaster

> External function generator direct input to your press
> External function generator digitized by your PLC and sent as commands to your press?

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[Compositepro]

Are you trying to make a vibration table and hope to be able to use your press to do this? I think you will quickly wear-out or otherwise damage your press trying to do this. That is why you cannot find information on this except on how to prevent it.

 

[PNachtwey]

Do you know the bandwidth of your system?
Do you know what the peak acceleration is? You should calculate this.
How much mass are you moving?
Your controller should close the loop 10 times per sine wave or more.
There should be no problem finding a motion controller that can close the loop 4000 times a second. The motion controller is only moving electrons. Moving physical machinery is hard.
Motion controllers are relatively cheap compared to the mechanics.
I get involved with a lot of sinusoidal testing. Testing to 10 hz is easy. 100 Hz is very difficult.
There is no a PLC can do this. I doubt an Arduino can. A Raspberry PI may be able with good program and fast I/O.




Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[IRstuff]

There are lots of options; you can use a programmable function generator and use the Arduino only for I/O, or you can precalculate the output values and use the Arduino to look up the values.

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[PNachtwey]

The problem is the peak acceleration. 0.01*(2*pi*100)^2=3947.8 in/s^2 or about 10.225g
Do not pass go until you know the actuator can accelerate the load at 10.225g.
Notice that the frequency goes up proportional to the frequency squared.
Power goes up proportional to the frequency cubed.
I have seen too many engineers try to control something that will never meet their specifications.
If the mechanics and motor sizing are right this could be done easily with a motion controller that has a sine wave function.

Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[RyreInc]

IRStuff, I don't understand how your solution would overcome the PLC scan time issue. I am not able to directly feed a control signal into these servo drives so it would have to go through the PLC. Plus I can easily generate a sine wave within the PLC, so I don't see the advantage of doing that externally. This would be a great idea if I could feed the signal directly to the drive.

Compositepro, I am certainly aware of the mechanical issues here. I have suspected from the beginning that this would be an issue, but have been asked to "press" forward regardless.

PNachtway, the BW of the system is 4kHz. The peak acceleration which the system can handle is about 1000 m/s^2, moving about 20lbm. I do understand that the amplitude would be limited as frequency increases for this reason. However, it does sound like even if I can break the 30Hz barrier I would not be able to go too much higher than that, even with reduced amplitude.

A motion controller with a sine wave function, eh? I'll have to look into that. Too late for this particular project but would be good info for the future. Do you know of any particular brands that have this feature?

Thanks everyone for your input, some good stuff here.


So, it sounds like I would need an additional actuator to achieve the frequencies and amplitudes I desire. What can I tack onto the press platen to do this? (e.g. voice coil)

 

[RyreInc]

Does it seem possible that I could use a cam table to achieve a higher frequency vibration? I have no experience in executing camming with servos but if this seems possible then I will spend some time learning.

 

[PNachtwey]

RyreInc, yes you can use a cam table but it still will not be good enough if just using a PLC. If you get a full featured motion controller it will have cam tables. The problem is that cam tables can't be changed on the fly easily. A good servo controller will have a sine command where amplitude, frequency, phase and offsets can be changed on-the-fly. Like at my webpage and application notes then look for the fatigue testing app note. This application used adaptive sine wave control for higher frequencies.

It is possible to control the actuator precisely up to a point. After about 30 Hz it gets more difficult to do. The alternative is for the motion controller to simply output a sine wave voltage to the servo motor or servo valve. The sine wave adapts so the desired amplitude is maintained until the voltage saturates at +/- 10 volts at which point there is nothing that can be done.




Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[IRstuff]

"Plus I can easily generate a sine wave within the PLC, so I don't see the advantage of doing that externally. "

The main advantage is typically time. I don't know enough about your particular PLC, but most processors can do a DMA followed by an I/O write substantially faster than cranking any level of math precision. Even processors specifically optimized for math can chew up cycles getting data in and out of the ALU. A good DMA scheme can bypass the processor altogether, by dumping a lookup table directly to the I/O port for the servo.



TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[MikeHalloran]

So far you've outlined a good selection of ways to kill a ballscrew.


Mike Halloran
Pembroke Pines, FL, USA

 

[RyreInc]

Thanks IRstuff, that does make sense now.

I did ask the question in a more generic way, but in this particular case the PLC is an Allen Bradley 1769-L36ERM (CompactLogix), which is a motion controller + PLC. This controller cannot, as far as I know, pass analog signals directly to the servo drive. They need to be in the form of specific instructions such as MAM, MAJ, etc.

Furthermore the servo drives are Kinetix 5500s, which take high-level commands over Ethernet/IP and do some internal processing to generate the low-level commands used to actually drive the servos. My understanding is this is required since the Ethernet/IP interface does not have a high-enough bandwidth, or too great of a delay, to send the low-level commands that are typically used.

 

[RyreInc]

MikeHalloran, absolutely this is a great concern for me. Do you think that using a separate device mounted on the press platen would alleviate or at least reduce this issue? Or will the vibration force just feed back through the ball screw regardless?

The material we are vibrating is pretty compliant so it seems like the vibration will not necessarily transfer through the press, if the vibration is separately actuated.

 

[MikeHalloran]

If the ballscrew is carrying the press load, the vibration will Brinell it.

Mike Halloran
Pembroke Pines, FL, USA

 

[PNachtwey]

Some road equipment rollers have mechanical eccentric weights inside the roller that vibrate to help compact the dirt below. If you are trying to do something similar on a press then why not make a normal hydraulic press but add a motor that spins an eccentric weight attached to the platen? You would need to filter motion and force feed back to the motion controller. The motor spinning the eccentric would vibrate the platen. It would be easy to spin an eccentric weight at a constant 100 Hz IF the device doesn't shake itself to pieces. You would need some tough bearings. You might be able to achieve 100 Hz with an amplitude of 0.01 inches. At least you would be able to feel the vibration and that may be enough if you only need to make the material compress faster.


Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com