Ref: Hurco Servos

I have never worked on a Hurco, however it would seem there would be parameters that allow you to alter the position loop gain.

Do you have parameters for backlash etc ?

Dave B

Dave;

The amps are Servomate Randtronics. They have the following Pot adjustments: GAN =3D Gain, TAC =3D tachometer input, AUX=3Dauxillary, (not used), BAL=3DBalance, SIG=3DSignal and CLM=3Dcurrent limit. Standard preliminary setup is to MAX out, (20 turns CW) the CLM and the TAC, set the BAL in the middle, adjust SIG to get the A/D out from the servo control board to the Servo amps around 0.65V during a 20 in/min feed, and use the BAL to get the voltage the sam in both directions. Also the gain gets adjusted up to instability and backed off a few turns. The problem is with the base settings the servos are vibrating like crazy. Backing off the Sig kills the vibration, but with the sig backed off to where the vibration stops, the servos are very slow to respond, and the A/D voltage gets to only .45 volts. Things checked so far:

1.)Tach inputs to servo amp are okay and correct polarity. 2.)Encoder pulses look good on A, B and Z channels on scope at the servo control board. 3.)Backlash on the ballscrews is less than 0.0002" 4.)Main cap on servo amp 80Vdc supply checks out okay. 5.) +12V and -12V Servo amp supply looks good. 6.) Position display is reading in the correct direction and is accurate. 7.) A/D signal from servo amp looks noisy, but I don't know what normal looks like.

At this point I'm not sure what to check next. Vendor thinks maybe problem is the new encoders, but I'm not convinced.

top posted to annoy those who hate top posting

if the servo is vibrating, it is unstable. You can do a root locus plot, or you can fake it the most likely cause is lag in the feedback system - you can accommodate additional lag by reducing the forward path gain (that's what your gain pot is for) - you might try playing with the TAC input, you may not have enough rate feedback - it sounds like your servo has an inner rate loop and an outer position loop, so the rate is primarily responsible for damping and responsiveness, and the position loop gets you where you want to go.

Just a little reading on feedback systems will go a long way towards giving you some understanding of what is going on.

Given that this happened when you changed the position sensor, I am guessing that it is a phase lag type phenomena because the new sensor is slower (maybe only by a few milliseconds) than the old one. Digital sensors introduce pure lag which is the worst thing for a standard feedback control system.

Dave;

The amps are Servomate Randtronics. They have the following Pot adjustments: GAN = Gain, TAC = tachometer input, AUX=auxillary, (not used), BAL=Balance, SIG=Signal and CLM=current limit. Standard preliminary setup is to MAX out, (20 turns CW) the CLM and the TAC, set the BAL in the middle, adjust SIG to get the A/D out from the servo control board to the Servo amps around 0.65V during a 20 in/min feed, and use the BAL to get the voltage the sam in both directions. Also the gain gets adjusted up to instability and backed off a few turns. The problem is with the base settings the servos are vibrating like crazy. Backing off the Sig kills the vibration, but with the sig backed off to where the vibration stops, the servos are very slow to respond, and the A/D voltage gets to only .45 volts. Things checked so far:

1.)Tach inputs to servo amp are okay and correct polarity. 2.)Encoder pulses look good on A, B and Z channels on scope at the servo control board. 3.)Backlash on the ballscrews is less than 0.0002" 4.)Main cap on servo amp 80Vdc supply checks out okay. 5.) +12V and -12V Servo amp supply looks good. 6.) Position display is reading in the correct direction and is accurate. 7.) A/D signal from servo amp looks noisy, but I don't know what normal looks like.

At this point I'm not sure what to check next. Vendor thinks maybe problem is the new encoders, but I'm not convinced.

** Posted from **

Presumably the encoders connect to the motion (CNC) controller. You've changed the encoder resolution, so to get best performance you need to change the gains in the PID loop in the controller. As you've found, fussing with the amp's gain will stabilize the loop at the expense of performance, but I don't think you'll be able to get the stiffness back without tuning the PID loop in the controller. Do you have access to the proportional, integral, and derivative gains in the controller?

I tend to agree with the vendor, I looked all over the net for some information on the MPU unit you replaced and didn't find much. The symptons are related to a resoultion problem, however the fact the move amount is correct confuses the issues.

Would be nice to get in an open loop condition and run only the velocity loop from a battery box. If the D/A puts out 9-10 volts at max then you could use a battery box and put lets say a 1 volt speed command and monitor the tach to ensure the scaling is correct for proper speed.

If the rapid rate is 200 ipm and the tach is 3 volts per 1000 rpm (screw) or whatever you could at least determine how close you are.

Once this is correct in your mind then you know the problem is in the position loop.

If you disable the servo and command a rapid move what is the output voltage to the drive?

Regards

Dave B

What do you mean with the term A/D as "A/D signal" and "A/D voltage"? You sound, from context, as though the term D/A would be appropriate. On another angle, you might look for a y in the path from the encoders; one pathway to the displays and another to the control boards. Possibly the pathway to the controller has it's own adjustments to accomodate the change.

Hul

oldjag wrote:

.

Dave;

I separated the A/D, (analog output), from the Servo control board that goes to the Servo amp input. I see +/- 0.45 volts max when I generate a position error by moving the servo motor shaft by hand back and forth, the same min./max as with the Amps connected. There is only a 1 count deadband between a reading of +0.45 volts and -0.45 volts. I have a dual channel scope monitoring the encoder inputs and the Servo control board output. Inputting a rapid move or a very small move makes no difference, ie I instantly get the same 0.45 volts max either way. This is well below the specified 0.65-0.80vdc @ 20 IPM the tuning specs call for. I have probably had power on and off to this thing a 100 times, and once in a while, say every 10 startups, I get either no A/D output or I get 1.75vdc., rebooting the MPU board returns the A/D to the +/- 0.45 scenario. The first time this occurred I thought a probe got miss-placed, but it has happened enough times now that I know that is not the case. So now it seems I have an intermittent problem, possibly related to or in addition to the tuning problem. When the 1.75 vdc happens, it seems real, ie. it goes +/- as error is dialed in, or if I move the motor. I have a signal generator which I could hookup to feed a fake varying A/D to the Servo amp to check it out somewhat. Should a control A/D output from a control board be normally be proportional to the position error?

Yeah, sorry I mean D/A output from the Servo control card. The backlash comp. is still set to the same values as before the board upgrade when everything was working. The backlash is set on the interface board on this machine. I also set the backlash comps to zero with not change in the vibration issue.

Bill;

TAC does not have much effect. With a dual trace scope I measured the delay time from an encoder pulse input state change to the Servo control board output changing state, and it's 9.6 millisec, which seems a bit long. The new encoders are good for 100Khz, or 6000 rpm, much faster than this machine can go. I measured the encoder rise time at less than 0.4 micro seconds. By the way this "oscillation" problem I am having begins as a "ticking" noise that occurs several seconds apart, and increases in intensity and frequency off occurance as the SIG pot is increased.

Each "Tick" is preceded by a gradual motor rotation of about 1/8 turn and then a "snap" back to starting position, which makes the ticking noise. All axis are doing the same thing, but for testing I've disconnected all except the Z axis for now. I thinking about trying an open loop run with a signal generator feeding the servo amp a slow sine wave, say to start maybe 0.2 Hz. and increasing the frequency from there. just to verify correct amp operation on the velocity loop. The tach signal looks okay at the amp input.

Now I have noticed another issue, every so many bootups of the MPU, instead of getting +0.45/-0.45 Vdc D/A output from the servo board it gives +1.75/-1.75vdc on encoder state change. On some boots I get nothing...nada. Maybe this is related, maybe not but it's very weird. I double checked my connections when this happened, checked the encoder inputs, and everything else was normal. Even the 0.45vdc output is suspect, as the D/A is supposed to put out at least 0.65 -

0.8vdc at 20 IPM feed. This aspect seems like a board issue, but the boards were sent back last week to the vendor and they checked out okay on their test rig....

Ned;

There are no pots for Prop, Integral, or Derivative on these amps.

Not on the amp, but in the controller itself, i.e., the device that's generating the command signal. I don't know how old your machine is, but for most any motion controller designed in the last 20 years or so, the PID gains will be set in software. Adjusting the gain on the amp has the effect of tweaking all the PID gains simultaneously by a common multiplier, which I'm not convinced is what you want to do when changing encoder resolution.

Following your doubt Ned, perhaps a couple of ttl decimal dividers soldered to a prepoked pc board to divide both the quadrature signals by 5 might work. The ease with which such a board is conceived and probably to make might indicate such a creature already exists on the market. Might be worth a look.

Hul

After another minutes thought, decimal up/down counters (dividers) and some logic to determine direction would be required. Hopefully though, still worth a look.

Hul

snipped-for-privacy@kbrx.com wrote:

Well it seems the weird boot to boot and day to day variation I was seeing in the D/A output signal was a flaky 15+ power supply that feeds the servo control board in the card rack. It would power up okay to 15v, but then after a short period of time it's output voltage would drop down to 3.5 volts. On some starts it would hold up okay for a while, (that's when I was seeing the 1.75 vdc DA output). The old TO-5 can regulator for this 1.5 amp 15vdc supply was shot. I made a new supply from a 1.5 amp LM317T adjustable regulator, (2 bucks at Radio Shack), installed it and now the Servos tune up rock steady. The vibration or ticking on all of the axis is gone.

One clue there was a PS issue was that as I tried to adjust the SIG pots up, I had to keep moving the BAL pot further and further from center to get a balanced D/A voltage in both directions from the Servo board. The only issue now is getting the machine to home correctly. This may be an related to getting the marker or index pulse in the correct location relative to the end of travel limit switches.

Thanks for all of the feedback and comments.

Congratulations - Hul

oldjag wrote:

One last point, Kudos to Buddy at AMTS, he mentioned checking the plus & minus 15V supplies for the Servo Board when I told him the symptoms. I had checked all of the Supplies earlier, (there are 13 DC supplies in this machine), and when checked, it must have been on it's best behavior.

Did you test them with a voltmeter or a scope?