AC Vector Drive + Encoder, Motor's shaft oscilates at set up speed 5

[robertoesc2001]

HI, thanks for this helpfull forum

I'm asking for your help to solve this aparent problem we have on a brand new mill tube forming machine, the man who originaly designeed the electric part, just left, living us to solve any further problem, besides that i really like the chalenge and do have some electric and electronics hability.

The system consists of (2) Baldor 18h Ac flux VFD's controling two diferent motors, 15hp and 5hp, coupled directly to one 1024 ppm dynapar shaft encoder each, and the speed reference(+/- 10 vcd) generated by a Machine Motion Control module from GL controls.

The machine is still on its starting up and tests period, more than 1 year so far!
Diferent copper tube forming and welding problems had ocurred since then. So i dont have any reference to say the electric motor controls were working pretty good before.

The last idea the production guys had, was to try the stroboscopic lamp to discard any vibration/oscilation problems on dies, pulleys, belts, etc, so they got to the two motors shaft and found this oscilation at the set speed on the two independent motors, it ocurred with and without the main belt on them (with and without load), i guess it was about +/- 1/6" and suddenly jumps to +/- 1/8" on the motor's (4")pulley's external diam. To be shure, we tryed the strobo lamp on one Lathe and on one CNC Machine Center Spindle, they where exactly steady fixed at the set-up speed. So Im wonder if this would be posible with the combo we have on this new machine, I havent had experience on closed VFD-motors before, nevertheless it was posible on the CNC spindle wich has a VDF drive and encoder on it too, not to say i read some threads here that says so, regarding a VDF with the encoder feedback.

Tests where made in the 5hp VDF unit and were as follows:

1) Control speed signal is normaly over 5 volts and under osciloscope has a little amount of noice on it, we twisted and ground shielded, nothing happened. (Perhaps we should try to put a small disc capacitor if recomended.)

2) Encoder signals where also twisted and shield grounded, the osciloscope between A+ and A- shows a +/- 200ns period (time) variation on each pulse. This made me think on the acumulation, when mult. by 1024 pulses and perhaps this is the electronic measurement of the phisical oscilation.

3) The control speed and encoder signals where separated and kept 90º angle from supply and motor lines as Baldor Manual recomends.

4) i havent disasembled the motor's guard and by that niether verified the encoder's coupling to the motor, as I could read from other treads on this forum, it could be the problem, will doit on monday.

5) some relevant parameters on the VDF are set as follows;
control base speed=1826RPM
feedbackfilter=4
current prop gain=100
current int gain=50kz
speed prop gain=10
speed int gain=1hz
speed dif gain=0
position gain=31
slip freq.=.47h
stator X1=4.4ohm
prop gain#1=15000
int gain#1=5000

Motor data is as follows:
motor voltage=460; motor rated amps.=6.5A; motor rated spd.= 1750; motor rated freq.=60hz; motor mag amps.=2.91A; encoder counts=1024ppr;

Sorry for the amount of information above, I'll apreciate your help if posible, first to let me know if this oscilation will be posible to be eliminated or diminished, Any ideas will be very apreciated.

Thanks for your time.
Roberto

 

[itsmoked]

Nicely stated question with good info!

We'll get to the bottom of this.

Where exactly is the encoder in your system? On the motor right?

What you describe sounds like an oscillation rather than a noise problem which can result in random jumpiness.

Excessive gains can cause oscillations.

Try reducing the Prop gain#1 by a bunch. Maybe 5000 at a time. What does that result in?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[robertoesc2001]

Thanks Keith,
You are right, the encoder is on the motor, i haven't take the back guard off yet, but the encoder cable goes there. (I thought to disasemble it and verify a slipery coupling, isoating, etc...perhaps later)

Regarding your recomendation, it says on the Baldor's manual for this parameter: "The anti-saturation controller's proportional gain. Leave the gain at the factory setting. Do not change this gain unless authorized by Baldor", But if you think it is required, lets doit!

Besides your recomend. , I will also try to control from keypad: run, speed, etc. to avoid any control signal noice.

ok, will let you know the results. Any other idea 'll be apreciated
THank you very much.
Roberto

 

[itsmoked]

Hmmm. Go a little less then.
You can also wait a little as there are people here that design these things.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[sreid]

In your first post you said two motors and one encoder. You need to detail the exact mechanical configuration of motors, belts, shafts and encoder(s)for us.

 

[ScottyUK]

One encoder each...


----------------------------------

Sometimes I only open my mouth to swap feet...

 

[ozmosis]

Roberto
Are you aware if the vector control function within the Baldor 18H has been tuned? They have an autotune function and then, if this does not seem to stabilise the speed control loop, then you can try it manually. Rather than explain it here, there is a link to the Op.Manual:

http://www.baldor.com/downloads/manuals/_downloads/718-1205.pdf

Page 6-1
Also, about a year ago we were doing some tests on various competitor drives relating to the performance of vector control. One thing we noticed with the Baldor (although I can't be sure it was this particular drive) was that the motor temperature had a large factor in the performance of vector control. There didn't seem to be any parameter to adjust motor ambient temperature. There must be one somewhere I guess. So if your ambient temperature is high or very low, check the parameter within the drive (if you can find it) and then carry out an autotune again using as close to ambient temp as possible.

If you could switch your control speed to direct from the leypad, this will be direct digital and could rule out any noise from your analogue input.

 

[robertoesc2001]

HI, thank you for asking
It consist of two VDF-MOtor-Encoder systems, One is 15 HP and the other 5 HP. Each systems drives thru a pulley-belt one of two diferent sets of circular forming dies (thru its respective 5:1 speed reducer), where a copper tube is getting formed,

The 5hp motor-vdf-encoder drives the front dies and the 15hp the back ones. We can adjust thru the HMI the speed ratio between both, being the slower the 15hP drive. This HMI conects to an MMC from GL controls, wich in turns adjust the control signal on each drive.
BTW, the encoder signals go to the VDF (Bakldor Series 18H Ac vector) terminals and also buffered to other VDF output terminals and directed to the MMC. (I think for it to manipulate this variable also).

I think this is the concept described as Master-Slave.
Any way, we tested with the stroboscopic tach, and found both systems oscilating, with and without load.
(we took the belt out in both motors)

Each system has its own encoder integrated on the back side of the motor, you can see a conector labeled encoder on it. I havent take the motor back cover out since it requires to take the whole motor out, due to space limitations. So the encoder is attached to the motor always on the tests we made.

The oscilation can be noticed on the 4" sincronous pulley at the motor shaft. So you can see on the pulley most external surface at..lets say, 800 to 1420 RPM, a little +/- 1/16" foth and back movement, with less frecuent but more noticeable increments to an almost 1/8".
At first, We thought it was normal, but compared with another references, like this Lathe or the CNC Spinlde, it make us think it's not normal.

After reading some posts on this forum, i understand that the speed should be steady even with the load on it, that's the reason for the encoder feedback on a VDF Flux Drive.

I hope I was clearer this time. Please let me know what else should I describe. Dear Srs. I do appreciate a lot your help and time to solve this problem.

Roberto

 

[robertoesc2001]

Ok, Sed2, will try on monday and let you know
Thank you
Roberto

 

[ozmosis]

Just another point, it seems a little strange to me to use belts/pulleys on such a dynamic application, or are they toothed belts?

 

[robertoesc2001]

Yes they are toothed Sed.

 

[itsmoked]

cswilson where are you???

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[ozmosis]

I was also thinking that Keith. If there's anyone out there who would know about this, it's probably cswilson.

 

[DickDV]

I don't think you want to change the anti-saturation gains. You need to find the speed loop controller gains and retune those.

More than likely, there will be an autotune function associated with the speed loop controller and I would use this for initial tuning. It will be a little sluggish but the motor rotation should be nice and smooth.

 

[robertoesc2001]

Thank you for all the recomendations,
we had problems with the stroboscopic gun (it wasnt ours, it was rented), but we will have a new strobo gun tomorrow, and follow the ideas above described, and let you all know, i.e. setting the keyboard as the control input, and run the motor autotune,
Dick, above in my first post, I described the control actual parameters, once the autotune function is completed and had tested the motor, if nothing happens, then i will try to adjust independently those values described above, all of them except the last two, is that ok? is that what you mean?
Thanks again
Roberto

 

[itsmoked]

robertoesc2001; You can mess with ALL the settings you want to! Just write things down so you can reverse them if desired.

Too high gains, can all cause instabilities. Instability is good for rapid movement but too much quickly leads to oscillation and random movement(bad!).

I'd mess with them in this order:
prop gain#1=15000
position gain=31
speed prop gain=10
feedbackfilter=4
current prop gain=100

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[robertoesc2001]

Ok, Sir, will mess in this order, any other idea, will be welcome
thank you
Roberto

 

[cswilson]

Hey -- I've been busy! (Still don't have much time, so I'll make this quick.)

There have been some excellent suggestions in recent posts. I suspect a tuning issue, too. While an encoder coupling problem is theoretically possible, with factory-installed encoders, this is extremely unlikely.

Two key things I believe you should look at:

First, what is the role of PosGain? If the input voltage is a velocity command, there should be no need for any position control. Now, its possible that the gain term is there but not used. Is the MMC trying to close a position loop, or is it giving an open-loop velocity command? I have seen several systems where both controller and drive were effectively trying to close position loops (even if the drive was integrating the velocity command into a position). When both are trying to do this, they can fight each other and cause oscillations.

Second, without the above issue, velocity loop tuning is the likely culprit. Remember that the velocity loop tuning is load dependent, so the drive vendor cannot do it for you. This also means that if you tune first for the unloaded motor (which I strongly suggest, if only for the learning experience), you will need to redo it for the loaded case.

Too much integral gain can often cause oscillation as the integral term over-corrects. A simple test is to set the "speed int gain" term to 0 and see if the oscillations disappear. Then re-introduce it gradually.

For the purpose of evaluating the drive's loops, it is essential that you open any outer loops (it looks like you are already planning on doing this). It is very difficult to disentangle the performance of inner loops from a closed outer loop.

Bad current-loop gains seldom cause oscillation. Also, these are only dependent on the electrical properties of the motor, so they are much more likely to be correct out of the box.

By the way, I don't know if these drives have any automatic temperature compensation, but if they don't and you end up tuning the current loop and especially the slip, make sure you do it with the motor hot. Slip can change significantly with rotor temperature because the resistance increases with temperature. If tuned hot, the motor will run a little inefficiently when cold, which means it will use extra current -- this will heat the motor quite quickly to the optimal state.

Curt Wilson
Delta Tau Data Systems

 

[itsmoked]

There now, see?

Curt opens his mouth.. My learning begins in earnest!

I think I learned about 8 things just now.

Thanks Mr. Wilson!

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[robertoesc2001]

Just Amazin!!, yes Keith, i now know why you said so.
Ok, I got your point about those two systems, it sounds so posible. Actually I really hope it is not the cause, since as i explained above in my very first post here, the integrator/designer guy left the company who sold the machine, and it would mean to get into the MMC program, and we dont have any software, program, cables to do that, not to mention to learn this specific control from GL controls.
Hard but not imposible!, so things are now in our court, let me see all the above and will come with news tomorrow.
Thank you all guys, really!
Roberto