Need help diagnosing a humming servo on my mill

I bought a 2001 Kitamura Mycenter 1xi in 2013. Of course it has a few issues and one of them is the Y axis servo motor hums when it stops moving. I have talked with a Kitamura national tech about this and he thinks it may be caused by weakened insulation on the servo, after ruling some other things out. To test this I need to get a megger since I don't have one. Before I do this I would like to run what I know about this mill past anyone interested and show some videos where you can hear the humming when I jog it manually, which is what make it hum the most. I also just got a free smart phone a month ago so I can take videos.

What is known:
Control is a Yasnac J300
I believe the machine has around 80,000 on hours, I think a lot of it was just sitting.
The machine ran straight oil for that first 80,000 hours.
If I screw the gibs in all they way they move .0005", so less than .0005" clearance.
Thrust bearings let the screw move a total of .0003" with a 300 lb load. There is .0002" of lost motion when I remove the load.
There is no measurable backlash in the nut with a 300 lb load. I have the backlash set to 0 in the control.
When it hums servo error does not exceed .001, resolution is 00.000. Hum lasts no more than 1-2 seconds at most.
Servo is connected to the screw with a solid steel connector. Connection to the screw is a key held by a set screw through the connector.
The hum is the same the entire axis travel.
All lube lines are flowing way oil.
When I indicate how much movement is lost from stick slip first thing upon startup I can get around .001" of commanded movement with the MPG in tenths mode before the table moves. After that first movement then only one or two tenths before the table moves.

I don't know how to embed a video yet so here is the link to the noisiest and a link to the video I did next that the machine quieted down some but better visual references.

Anyway I am about 80 miles from Portland Oregon and only know one machine tech but he is a sworn Fanuc guy so if possible I would like to figure this out myself with some help from this forum.

Thanks in advance for any help.

This machine wont have scales on the axis? All it's feed back position and velocity comes off an encoder mounted on the motor.

Short of a stiff or over loaded axis, that cant pull into position, you can discount back lash & stick slip. It's not getting feed back from the position of the saddle, because it has no scales. Thats all coming off the motor encoder.

I'd expect excess current error's for an insulation breakdown.

I'd expect bad feed back. Dirty connections. On rare occasions connection between motor and encoder. Its past brush tacho vintage? Or drive tuning, excess gain will make an axis oscillate and hum. Fact that it pulls into position and you're only seeing 0.001" lag. That's not a positioning error due to backlash.

Do you have an inposition parameter on that control?

Regards Phil.

Thanks for your help Phil. It looks like parameter 4015 bit 4 is "0 Sets to IN-POSITION WAIT with G00 and G01"arrow right"G00 commands", 1 does not set. When I look this up the decimal value is 32, which converts to 00100000, so it looks like it is set to on. I will double check this when I turn the machine on this morning.

Dirty connections, thank you. What is the best way to clean these? This would be easy and free of risk.

I am having another problem that hopefully is a dirty connection. When booting up the control it is getting a new alarm that "INTEX RTC UNREADY". So far I just power down then up and it boots but is getting worse fast. When I search this it looks like it is an Ethernet card, which my machine has. With the machine off can I remove the card, clean the contacts, and replace or is there more to it?


There is always something missed, sorry.

No scales and servos have absolute encoders.
All servo settings such as gain etc are the same as when sold new. I compared the current parameters against the original floppy of parameters in the cabinet.
I think servos are brushless but will confirm.

Its not uncommon for servos to need a retune gain and damping wise with age, when new there full of stiffer grease, over time this gets pushed out the way, things free up and if it was set - tuned tight at the factory it will now need dialling back a little to counter act the reduced mechanical damping. There should be a procedure.

I too secound the really doubt its insulation related. Insulations good or bad, if the coils are moving enough in the motor to make noise they would have worn through the insulation and shorted by now!

adama said:

Its not uncommon for servos to need a retune gain and damping wise with age, when new there full of stiffer grease, over time this gets pushed out the way, things free up and if it was set - tuned tight at the factory it will now need dialling back a little to counter act the reduced mechanical damping. There should be a procedure.

Click to expand...

I've seen similar to this also. System drag acts as a damper to the servo. If you can find your servo error display, check to see if it is jittering around a lot on the Y axis. I suspect it will be.

Another spot to check for play is the coupling between the servo and encoder. That is a common cause of servo hum on Fanuc DC servos. They used an Oldham type coupler between the servo and encoder and with some wear, would get a tiny bit of slop. Not sure what your Yaskawa servos use for coupling.

Vancbiker said:

I've seen similar to this also. System drag acts as a damper to the servo. If you can find your servo error display, check to see if it is jittering around a lot on the Y axis. I suspect it will be.

Another spot to check for play is the coupling between the servo and encoder. That is a common cause of servo hum on Fanuc DC servos. They used an Oldham type coupler between the servo and encoder and with some wear, would get a tiny bit of slop. Not sure what your Yaskawa servos use for coupling.

Click to expand...

I actually ran a simple rectangle with large radius profile program yesterday to watch how it worked just cutting air without the spindle on. I did watch the servo error screen and the Y axis did not jump around at all, there was no noticeable difference between the Y axis and the X axis error. I also enabled high speed machining, HON, to see if that made any difference and it did not. The error remained the same and there was no hum from the servo. The high speed machining option makes an amazing difference in how accurately the mill maintains it's programed path, it will go from rounded corners with it off to sharp, crisp, chamfers with it enabled. It does so by tightening the servos up and driving them harder, you can really hear it. The hum is by far the worse when jogging manually with the MPG in smaller moves. When I run a program with sharp direction changes, no corner rounding G2 or G3, then the servo will hum some.

I will run a different program tomorrow with hard corners to get the servo to hum and see if the error jumps around, and try high speed machining again to see if it makes a difference when the servo is doing moves that make it act up. I will also try to get the servo numbers and see if I can get any info on it from Yasnac's web site.

Thanks again for everyone's help with this.

David

Can you post a picture of the servo drives please?

The hum you are hearing most likely will only occur when stopped and you will see no problems when moving.
With a scope you would probably see some ripple on a constant velocity move but you won't see it watching the error as it is only one or two counts tall.
The D term is an unhappy camper, if an analog input to the amp one can also get this from bad grounding and just a touch of electrical noise.
One can remove it by changing the loop parameters or inserting/changing a notch filter by why that would be necessary is a mystery.
During high speed machining the system may or may not tighten the loops during the run, acc/dec will certainly change, but it would normally shift back to normal mode just before the stopped position.
I don't think you will find much here to help you.

I'm not so much in favor of the "weakened insulation, bad servo" deal although a bad bearing in the motor can cause this problem.
Yes I know, all this is of no help but maybe sends the search in different directions.
Is there a "motion guru" near you?
Bob

I played with it some more today and figured something new. If I push on the table as hard as I can with my feet wedged against the wall, around 150-200 lbs, when running the program the hum goes away completely. Back off on the pressure just a little bit and the hum returns. If I set the feed to 200 ipm the hum is far less, almost gone. High speed machining doesn't affect the hum. The program is one pass profiling a 1" square with a 3/8" mill with the spindle off cutting air. It only hums when the Y axis stops. Could this be the thrust bearings??? The servo error pulse does not exceed .001 when humming, the pulse error only goes 3 decimal places. The thrust bearings have .0002" of lost motion and .0003" of total movement with a 300 lb load from both directions. Pushing and pulling on the table to see how much movement isn't the same on this machine as linear way machines, you need a 2x4 to pry with to really test for movement.

Here are some photos showing the machine, servo drives, and the data screen info on the control.






CarbideBob - I really haven't seen the hum affect the parts. Other than not liking the hum I am worried that it is detrimental to the servo and screw. The mill is in pretty good shape so it is worth fixing these little problems. As far as there being "motion gurus" near me I am sure there are a bunch of them working for the different drone companies that are local, I just don't know of any.

Swap motors and see if the problem follows the motor. If the cabling from the motor to the drive can't be swapped, at least you'll have a better idea how the motor behaves on a different axis. You can swap the drives and see if the hum follows the drive. If it does, it's probably a tuning issue.

A couple other things to try...

Hold the screw tightly and see if hum is reduced. Twist both directions and note if it changes relative to direction of twist.

Swap the X and Y servo drives and see if the hum follows the drive.

Swapping drives sounds a lot easier than motors and cables would be a real PITA. How do I swap drives? I know enough about electronics to be very careful and not enough to really have a clue. I am far more comfortable with mechanical things. Should I pull the way cover back to test holding the screw first? Here is a photo with the covers opened at the top. Drives from left to right are X, Y, Z just in case there was any doubt.

could be slop between the servo and the encoder. if that's the case then swapping the servo motors will translate the problem to the x axis and it will probably sound different as well because there may be different pid constants between the two axis. but that doesn't really explain the hum going away with force exerted on the table by OP's feet.

theoretically i would expect different tuning parameters between the x and y axis, so swapping the servo drives may or may not fix anything. where are those parameters stored? if they are stored on the drive then swapping the drive may seem to fix the problem, until you discover it again under some different conditions.

if there's a problem with the connector to the encoder, then simply reconnecting it may make it go away.

Look the servo is dithering when stationary because the gains too high - not enough damping, when you load it you increase the damping through mechanical means and the dither reduces. You have to retune it, if you don't know how you need a tech that does.

Basically no servo knows its stationary, so what it does is go clock wise till it sees a pulse, then reverses direction till it sees a pulse. Fine enough encoders this is so small a movement its not audible. When the gains too high - under damped it over shoots and then goes the other way, this movements a lot more and hence is very audible. Most servo drives after x amount of time stationary power down to some degree unless they sense movement and will then fight it. Go read the WIKI on how servos work and you will make a lot more sense of this. google the phrase "critically damped", yours servo is not! It needs to be.

Most encoders on mill servos will have pulse widths under the screen resolution so you probably won't see it on the display and your talking movement errors of probaly only a few microns tops, again its not going to be visible in the typical part.

I think the tuning parameters are held in the control. Here is what I think is the page of relevant parameters and if so they are the same for X, Y, Z. They start at 1580.




Yeah, I could use some help from a tech for a few things. I would like to make sure I have all of my parameters backup up on both Yasnac mills correctly, and I need to replace the batteries in the controls. I have never done this and would like to make sure it is done right, one machine is 20 years old and I have had it for 12. I would also love to know what I have as options on my Kitamura. The high speed machining card and Ethernet were nice surprises. I know it was set up for tool setting and am wondering if it had a probe on it since there were two cables located in the machining area.

If this matters 1580 and 1581 are 4000, 1582 is 500, 1586 is 50, and everything else is 0.

Yep, compare the Kv values for the X and Y. The Kv is the velocity loop, the servo response. The Kp is the outer loop, encoder feedback positioning loop (positional error).

Try reducing the Y axis Kv and see if the hum dies down. You want to only go as far as needed to squelch the hum.

A highly responsive servo can produce better parts by not overshooting. They usually run them on the more "jackrabbit on speed" side of the aisle because it allows them to run higher feeds with accuracy.

The loop frequency is much higher on the Kv than Kp. Somewhere in the range of 300-3000Hz for Kv, and 250-1000Hz for Kp.

At 1000Hz, that's 1000 updates per second on the servo position. When you're moving 1181ipm, that's one update for every 0.019684 inches. With that loop, you'd max out at 60ipm for 0.001" updates.

I know that one of Fadal's claims to fame was their 1Khz servo loop. They were using analog resolver based drives, which have a servo loop of around 8-15Khz, probably closer to 8Khz, based on the obnoxious whine. Their 1Khz is the position loop frequency.

They have you do the servo balancing at 100ipm, which means to me that they targeted 100ipm as the feed which produces the most accurate moves. When you balance the drives, it's doing a 3 axis simultaneous move at 100ipm and the goal is to set the velocity command to 590-600 counts. Having all 3 axes move at the same velocity, when commanded at the same velocity, means less trajectory errors during contouring operations. The control doesn't have to speed-up or slow-down axes to keep them contouring at the same rate. It's probably very crucial to their (Fadal) overall design.

Thank you Perry! The Kv and Kp values are 4000 each for X, Y, and Z. These values have not changed since the machine left the factory. I will give it a try right now.

Got it powered up and playing with it. It still hums some jogging. Do I need to power down and back up for the changes to take effect? I will try running the same programs again to see what difference it makes.

It seems odd to me that all of the values are at 4000. I would expect those to be factory baseline values, then the inspectors would run the machine through a procedure to set each of the values to something specific to that machine.

I'm gonna bet the parameters were erased and those are the baseline values. You may want to inquire what procedure was used to calibrate those settings, or try to get the original parameters.

I have what I think was the original 3.5" floppy of parameters, the parameter backup I had done when I had the machine inspected, the printed sheets of parameters, options, setting data, pitch error, inverter settings, and the changes made when the 4th axis was installed in the field. They look to be the originals with some of them dated June of 01. The machine came out of a BIG shop doing medial parts like hip joints and the like. I know that these were the parameters it had when it came out of service.

So it still hums when I set the Kv to 1000 and jog it around, pretty much no difference. When I run the same test program I can set the Kv to 3300 and it doesn't make a peep at 40 ipm. At 3500 and 40 ipm it starts to hum, if I bump the feeds to 100 ipm is no longer hums. If I run the feeds at 200 ipm I can set Kv at 3800 and it barely hums.

Please explain what reducing the Kv value does to the cut moves vs programed. I watched the pulse error dispaly and it did not change with the different Kv settings.