What's new
What's new

Need Help with Servo Tuning - FANUC

WayneC369

Hot Rolled
Joined
Jan 4, 2014
Location
ATL, GA, USA
FANUC 18TA control
Series C amplifiers (one dual, one single)
Okuma-Howa ACT20 lathe


I have a turret indexing problem that has me dead in the water. I continually get the 414 alarm which turns out to be the HCA.

I have spent weeks on this and ruled out hydraulic problems, unclamp limit switch problems, bearing issues within the turret, cabling issues from the amplifier output to the servo motor. Replaced the alarm module in the servo amp. No joy.

The only thing I can think left to do is check the servo amp tuning parameters and be sure the max current setting (if one exists) is set properly.

The control on this machine recently smoked a power supply board. By the time it got replaced the parameter memory was corrupted. I can't help but think a parameter got missed with the reload. But, the turret was acting up before this happened. Once the power was applied to the control for about an hour everything was fine. Now it's a solid no-go. Do a manual index - bam! E-stop condition. Call a tool through MDI - bam! E-stop condition.

Any help/guidance is much appreciated.

EDIT: I have checked the power semiconductors on the amplifier and even swapped amps. Problem moved to the other amp.
 
What was the turret doing before the power supply problem?
Why did the power supply smoke?
How did you reload the parameters?
I would meg the motor and cables, or at least check with a meter on the highest ohm range.
 
Sometimes the turret would alarm on 414 when I first powered up the lathe. Most times I could simply cycle power and it would clear. Then I could run it all day without further issues. One day it simply stopped indexing and would throw no alarms. Just sit there. Now after trying to recover from the power supply failure it alarms out far more than it indexes. When it does index it's smooth and fast.

I have swapped alarm modules with the other amp and it worked great for a few indexes then after cycling power it throws the alarm as fast as you hit the index button.

Using a megger set at 1000 volts and the motor leads disconnected from the amp, of course, I read right at 200M ohms between all three phases and ground. Compared that with the other two servos and they both read right at 500M ohms. Somewhere I saw a recommendation of 500M. So I replaced the motor cable. Now the turret motor-cable combination has the highest insulation resistance of all at 5,000M ohms.

I ordered two used alarm modules and neither of those are any better than what I have already.

I tore into the turret and find no mechanical or hydraulic problems.

Parameters were loaded via computer connected to a serial port on the control.

I have yet to determine a definitive cause for the power supply failure. All the fans in the control card cage were locked down. No fuses were blown. All are sized properly and none jumpered.

EDIT: For giggles I even swapped the control cables with the X axis amp. No joy.
 
If your parameter back up was good there should be no problem there. I would look in the Fanuc manual and find the gain and speed parameters and turn them down for testing?

Meg readings sound good. I would Ohm out the motor with the cables disconnected. Some motors have such low resistance that it will require an Milli Ohm meter.
With the best meter you have check that the winding's are balanced.
 
Winding resistance as checked with a Fluke 787 reads 1.1 ohms in all three positions. And, that's after lead compensation. I find it difficult to believe the motor is bad. Either these alarm modules are lying to me or a setting needs adjustment.

Searching the manuals now. This thing has always been lightning fast with the index. I think it can do with adjustment to slow it down a bit. I'm wondering if the previous owner(s) had it screaming out of greed and now maybe wear is catching up. Maybe accelerating too hard?

There is a timer in the ladder logic that's activated by the turret unlock limit switch. Has a value of 15, but I can't figure out what time base it's using. I would assume milliseconds. BUT, this timer times out AFTER the turret unlock switch makes. So I can't figure how sluggish unlock can come into play here.

I read a little something about an auto-tuning routine. Any idea how to set that off?
 
Current turret servo tuning parameters for reference:
(PARAMETER)
FUN. BIT 00000000
LOOP GAIN 4000
TUNING ST. 0
SET PERIOD 0
INT. GAIN 151
PROP. GAIN -1355
FILTER 0
VELOC. GAIN 200

(MONITOR)
ALARM 1 00000000
ALARM 2 00001011
ALARM 3 10000000
ALARM 4 00000000
ALARM 5 00000000
LOOP GAIN 0
POS ERROR 0
CURRENT% 0
SPEED RPM 0


Any idea what fun bit #7 is all about? The parameter manual only lists it in a table with the cryptic name of "V0FST". No description to be found. The X and Z axes have this bit set.

The X and turret motors are of the same part number. The loop gain for X is set at 3000 while it's 4000 for the turret. I realize each deals with a differing mass. Was wondering if that gain was too high. So lowered it on the turret. Still no joy.
 
Auto tune can cause more trouble than it's worth.
The clamp un clamp time should have nothing to do with OVC.
The amp should wait for the switch to change state before it runs.
At this point you have looked at all the common stuff, I would try to dial back the parameters one at a time to try to find the sweet spot.
Hard to see 2 different boards with the same problem. Maybe AMP Clamp the motor leads to see if you really are OVC.
If you have 2 of the same motors, I would swap them.
Best of luck, I am bugging out for the weekend.
 
I dont know anything about 6066 series drives.
I would remove the motor from the turret and run it.Put some load on it both stopped and moving.

Unclamp the turret and move it by hand also.
 
The turret turns freely. Was looking at pulling the turret servo motor earlier. That's gonna be a PITA. The X and T servos are both 5S motors, but the X axis has a brake on it. So there's no swapping those. There's no room for the additional length.

I'm finding it difficult to accept the turret servo motor is bad as it tests good in every way. Ox suggested I swap the Alarm Modules between the two amplifiers and that worked for a minute. I bought two used modules and replaced with those units and they wouldn't clear up the problem either. I'm sure it's totally possible to have four bad alarm modules on my hands.

There has to be a configuration problem. I'm not altogether sure the turret home reference has been properly done. So, not sure if that has any bearing on the issue.
 
OK, so you can qualify that there is no resistance in the turret it'self?
I have had to replace bearings in one recently that didn't want to turn. (even with help)


If you wunna slow the turret index speed and/or acc/dec rate, look at params 1420 and 1427.
I have slowed mine down - actually, I have completely redesigned my whole index system, but those pics are down ... for now ...


The brawn side of your amp could be bad, or maybe "weak"?


You say that X and Turret have the same motors.
Does your param 2020 agree with that?



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

Think Snow Eh!
Ox
 
Last edited:
Today I figgered out how to unlock the turret and turn it by hand. With the control up and running I disconnected the PMC input from the turret unclamp limit switch. This stops the amp from being enabled by the ladder logic. Removed the hydraulic solenoid plug and applied 24 volts to the solenoid coil. (I could hear what sounded like air in the system. Haven't figgered out how to bleed that yet.) Then I could reach in and rotate the turret.

The best information I have sez there's a 20:1 total ratio betwixt the servo motor and the turret. Given the mass of the turret, turret shaft, curvic coupling half, a 100 tooth gear, a jack shaft, an 80 tooth gear, two 20 tooth gears, and the mass of the servo rotor I do believe it turns freely enough. A little hard to accelerate by hand as I am on the disadvantaged end of that ratio. I rolled it numerous complete turns and I could feel some minor gripe but nothing offering a bind.

Yes, 2020 has the same value - 48. The brawn side would hafta be weak on two amps because I swapped X and T & get same result.

I will look at params 1420 & 1427. Thank you for the direction!

Next step, I think, is to pull the turret servo and see if the amp complains with no load. I think I had rather take a beating than to do that. Gonna be a PITA.
 
......Any idea what fun bit #7 is all about? The parameter manual only lists it in a table with the cryptic name of "V0FST". No description to be found. The X and Z axes have this bit set.

Bit 7 set to 0 disables Velocity command Offset. This won't have any bearing on your problem.

Next thing we need to do is figure out how the turret servo motor is configured. There are at least three different methods that a builder can choose between depending on how they want things to work. Need some parameter info..

Is 2007.7 set to 1?
If it is, then what is in 2105?

Is 2203.4 set to 1?

Do you have an original parameter list for the machine?

When it is the alarm condition what are the contents of Diagnostic 200 through 204?
 
Bit 7 set to 0 disables Velocity command Offset. This won't have any bearing on your problem.

Next thing we need to do is figure out how the turret servo motor is configured. There are at least three different methods that a builder can choose between depending on how they want things to work. Need some parameter info..

Is 2007.7 set to 1?
If it is, then what is in 2105?

Is 2203.4 set to 1?

Do you have an original parameter list for the machine?

When it is the alarm condition what are the contents of Diagnostic 200 through 204?


Hello Vancbiker!

Thank you for responding! I will answer your questions in order below:

2007.7 for the turret axis = 0 (A axis on this mach. BTW X = Z = 0 as well)

2126 thru 3000 are not implemented on this control (FANUC 18T-A - the oldest to my knowledge). Or, are they hidden behind an access parameter?

I have a weathered printout that may or may not be original that was printed on continuous, tractor fed paper by a dot matrix printer that lists the special setting parameters (9900) and lists other things, i.e. servo data sheet, module/board part numbers, etc. One of the first things I did when I first got this lathe powered up was make a backup of the parameters, but I did not know to write down keep relay config, timer & data values and such. The power supply took a dump and in the interim of replacement and I lost that information. I have had to figure it out by parsing the ladder logic in the pmc. This turret issue makes me think I have missed something there, but can't be sure.

In alarm condition the only bit that gets set is HCA. All others in 200 thru 204 remain reset (= 0). The docs I have say this is an abnormal current alarm. Since there is already an over current alarm what would abnormal mean? Too low? Phase imbalance?
 
I made some mistakes writing the above, but I dare not attempt an edit since it will all be deleted. Perhaps you can still understand what I'm saying.
 
Thanks for getting this info. I'll have to look more closely at it later tonight and see what ideas come up. I think you are correct in suspecting an issue with the PMC parameters.

Many builders using a servo driven turret index have the system set to use two torque settings on the motor. When the turret is clamped the motor is set to a very low torque value. That allows for slight mis-alignment between the index stop position and the clamped position without causing the motor to fight the mis-alignment and get hot. When the turret is unclamped the the higher torque setting is activated to spin the turret and then set back to low torque once the turret is clamped. I have not worked on and Okuma Howa with a servo turret so don't know for sure if they are doing it this way. Parameter 2007.7 sets one method of torque control and in your case it is not activated. There may be other modes, but I'll have to look into that tonight as I'm pretty busy today.
 
BTW, the reason I'm going in this direction about torque control is that to reduce torque in a servo motor, current is reduced. If the drive is set to a low torque mode but rotation is commanded, the allowed amount of current may not be enough to rotate the motor and cause an abnormal current alarm.

Typing this made me think of another question. Does the turret unclamp and then trip the alarm or does the alarm come before the turret unclamps?
 
No, thank you for your time and willingness to lend a hand!

Wow, not knowing anything about lathes I was thinking the same thing about the torque control. That's what led me down the path of checking all that I wrote about in my original post. Good to have validation at least!

So, the current mechanical condition is the turret servo motor is removed from the turret and physically laying on the wire trough at the back of the lathe. Therefore, no mechanical attachment.

The alarm seems to trip immediately. The rotor will bump. Turns maybe 10 degrees at most. The servo motor gets hit with a good amount of current as judged by the stator reaction, but it's not getting full current I'm sure. I have it tied down as a precaution tho. (Hope that males sense.)

Based on what I see in the ladder when a turret index is commanded the hydraulic solenoid is activated. There is a stud protruding from the turret that then hits a limit switch. The switch enables the servo amp and simultaneously starts a timer with a preset of 15. I have not determined the unit of time this timer ticks off. I assume milliseconds. After time out the servo is commanded to run. At least this is how I understand it. So, this configuration would basically remove slow hydraulics and sticky clamp from the equation. Unless I have some aspect of it wrong in my mind.
 
Forgot to add, I can actuate that unclamp limit switch and that will enable the servo amp. (Mind you the motor is mechanically disconnected.) And when the amp is enabled it trips with no rotor motion at all.
 








 
Back
Top