[Charles2]

Hi Tom,

I don't understand why I can't get higher a 'P' value. no matter what i try, it seems that in order to have stability with larger moves, P is going to be around .12 and D is going to be around 90. And I still don't have a good plot. I can move P and D up proportionally above P=1.0, but lose stability on larger moves.

This drive has a feature called Torque Hold, which is activated by jumping terminals 6 & 7 on the drive terminal strip. My understanding is that it overrides the tach signal and puts the drive in torque mode. There is also a feature called External Current Limit. My understanding is that it limits drive current for testing purposes. It is activated by jumping from the drive terminal strip common to terminal 9. Per advise from Mike, the drive specialist, External Current Limit has been active since I had the first problems with the drives. He advised to leave it on until ready to cut chips as a precaution. I ran some tests, all with the same parameters, with various combinations of tach on/off, torque hold on/off, external current limit on/off, and tach terminals jumpered. As expected, it doesn't seem to matter much if the tach is on or off, if the torque hold is on. see notes in plot window for what is on/off.

Plot and Data 28

Attachment 447990

Attachment 447992

Plot and Data 29

Attachment 448004

Attachment 447994

Data 30, Plot in next post, Exceeded file # limit

Attachment 447996

Data 31, Plot in next post, Exceeded file # limit

Attachment 447998

[Charles2]

I hope my attempt to post too many files doesn't make a mess. Note that Plot and Data files share a common number (28-33).

Plot and Data 30

Attachment 448006

Attachment 448008

Plot and Data 31

Attachment 448012

Attachment 448010

Plot and Data 32

Attachment 448014

Attachment 448016

Plot and Data 33

Attachment 448020

Attachment 448018

[Charles2]

Hi Tom,

I pulled the tach cable out of the factory routing and looped it over and away from other wiring. Ran a plot with same parameters as above.

Plot and Data 34

Attachment 448024

Data34.zip

[TomKerekes]

Hi Charles,

Looking at the schematics I think shorting Torque hold may reduce the input gain and also effectively defeat some Lag compensation. This seems to be corroborated in the plots with smoother motion and larger errors as one would expect by turning down the P D Gains. I'd recommend trying shorting the Torque Hold (not sure why they call it Hold?). But then optimizing the tuning (increasing gains).

I'm assuming these are all small size moves that don't ever have any problem with higher gains. I would have tested large moves and tune higher gains to see which performs better.

You might check if the current limit Output ever pulses while making bigger moves.

The Amplifier has some non-linear "Gain changing" components which reduce the gain at higher currents? Its not clear how to defeat that or if it is in use.

I see there is a current monitor out. I suppose we could connect that to a Kanalog ADC. A program could be written to capture data and analyze what the Amplifier is doing. Maybe ramp the DAC up and down and monitor the actual current to see how linear and with what lag.

Thes amplifiers are from 1973? I found a related patent listed in the manual I found. Do you have any idea how well the original control ever worked? Modern drives switch using FETs at like 20KHz vs these with SCRs switching at 120Hz. So it isn't clear how well it could ever work. You might consider trying a modern Drive such as an AMC to see if the performance is greatly improved. They can often be found 2nd hand fairly inexpensively.

[Charles2]

Hi Tom,

I will test as you suggested. I did run this machine years ago when I first got it. It had a very early GE2000 control with a few short programs in memory. I think they were demo or test programs. As I said before, this machine came from a community college. I don't think it ever ran very much at all. Probably sat powered up most of its 400 hours. The programs were contouring and threading on 1" round stock. I'm pretty sure it was running CSS when it cut the ball end on the part. It ran fine, good surface finish, good threads.

I didn't have the source code for the control, so I couldn't upgrade the software. After I let it set for several years, the control would not power up. That's when I started looking at Kflop/Kanalog.

I was hoping to use the original drives in order to keep the cost down. I'm not opposed to considering other options, if it can be done at a reasonable cost. I don't know how to select a replacement drive. According to the existing drive manual, the axis drive is 40 amp, controlled by a +/- 10V analog signal. Motors are DC brush with tachs and encoders. I don't know the torque rating of the motors.
The spindle motor is 7.5 hp and the spindle drive is similar to the axis drive but 85 amps. I hope it is usable. I need to test it before I get too deep into all this. I assume it is to be tuned about the same as the axis drive.
As has been mentioned several times, this machine is in such excellent condition mechanically that it would be a shame to not get some use out of it.
Please excuse the rant. I'm just getting frustrated with my lack of progress, mainly due to not knowing what I'm doing.

Thanks,
CJS

[TomKerekes]

I was hoping to use the original drives in order to keep the cost down. I'm not opposed to considering other options, if it can be done at a reasonable cost. I don't know how to select a replacement drive. According to the existing drive manual, the axis drive is 40 amp, controlled by a +/- 10V analog signal. Motors are DC brush with tachs and encoders. I don't know the torque rating of the motors.

I just mentioned this if you are in a rush to find a solution it would probably be worth while. Used AMC Drives can often be found on Ebay for less than $100. You can go with less voltage and less current from the original drives (especially for a test). But the max speed and max acceleration will be limited respectively. I doubt if this would be an issue for the small/slow test moves that we've been looking at. One thing that just occurred to me is that this system doesn't have a DC Motor Supply as it works by applying portions of the AC cycles. Most AMC drives will require a DC Supply.

The spindle motor is 7.5 hp and the spindle drive is similar to the axis drive but 85 amps. I hope it is usable. I need to test it before I get too deep into all this. I assume it is to be tuned about the same as the axis drive.

Yes I would test this ASAP. But Spindles are not usually very demanding as compared to Axes Drives. Gradual speed changes are normally sufficient.

As has been mentioned several times, this machine is in such excellent condition mechanically that it would be a shame to not get some use out of it.
Please excuse the rant. I'm just getting frustrated with my lack of progress, mainly due to not knowing what I'm doing.

Hang in there

[Charles2]

Hi Tom, and anyone who has ideas on affordable newer drives. I've been looking at AMC drives on ebay. Plenty of what appears to be 12 amp drives. Mostly for brushless or PWM motors. With my existing drives, just how low do you think I could go on drive amperage and get reasonable performance? I saw somewhere that brushless drives could be used by connecting only 2 motor wires. Do you or anyone know about this? I guess I could try a 12 amp drive just to test everything else. The spindle seems to work pretty well with DAC commands. Will post results of spindle runs. Do I need to the spindle as an axis for closed loop results?

Thanks,
CJS

[TomKerekes]

Hi CJS,

With my existing drives, just how low do you think I could go on drive amperage and get reasonable performance?

80V @ 12A is 960Watts of Power which is more than 1HP. I would suspect it would be fine for tests.

I saw somewhere that brushless drives could be used by connecting only 2 motor wires. Do you or anyone know about this?

Attached is an App Note from AMC. Basically set 60 degree phasing, no Hall sensor connections and connect motor between Motor 1 and Motor 22.

[Charles2]

Hi Tom,

I think I will find a 12-? Amp AMC drive for at least testing. Looks like I can try it in velocity mode with the tach or torque mode without the tach by setting a switch. I will also have to work out a dc power supply.

My spindle motor drives the spindle through a toothed belt with the motor turning 1 turn for 2 turns of the spindle and the belt driven encoder. Looks backwards to me, but I guess it's overdriving for a reason. Using DAC commands, I get the following:
DAC2, rpm spindle, rpm motor
100,. 235, 118
200, 467,. 234
300,. 702,. 351
400,. 936,. 468
500,. 1172, 587
600,. 1408,. 704
700, 1644,. 822
725,. 1701,. 851
750,. 1741,. 871
775,. 1742,. 872
800,. 1742,. 872

I expected higher spindle speeds. But I do still have current limit active on the spindle drive.

Negative DAC commands produce the same rpms in opposite rotation.

Thanks,
CJS

[TomKerekes]

Hi CJS,

Motor RPM max of 872 RPM seems low. I wouldn't expect a current limit to limit unloaded speed. BTW are you applying 240V 3 phase line power? What kind of isolation transformer does this have?

It seems the TPA/2-15640 means the internal AC voltage should be 156V AC RMS? Have you ever check that?

[Charles2]

Hi Tom,

This machine originally of course was 460 volt three phase. I'm using a 10 hp phase converter to produce about 230 volts three phase with a wild leg. With the machine powered up, hydraulics running, input to the transformer are 123, 168, 123, VAC line to neutral and 219, 213, 248, VAC line to line. Transformer output to drives are labeled to be 275 VAC L-L. They are 278, 231, 281 VAC. Line to neutral are 141, 169, 142. Looking at the drive tag again, I now think it is 15840 instead of 15640, as the manual lists 120, 138, or 158 as "Secondary Line to Neutral Voltage" options. The spindle is 15885. The 40 and 85 are "Continuous Current Rating." I can post the transformer details later. I did switch the transformer jumpers from 460 to 230 per the tag on the transformer. There is a separate transformer for the 120 VAC and 39 VAC, which feeds the 24 VDC power supply.
I suspected the phase converter imbalance may be a problem, but I was told by someone who should know that it was not a problem. The drive chops it up and converts it to DC anyway.

One wrinkle that just occurred to me, the machine is labeled as 40 amps at 460 volts. Changing it to 230 volts means it now requires 80 amps. I'm not sure it's getting enough current.

[TomKerekes]

Hi CJS,

One wrinkle that just occurred to me, the machine is labeled as 40 amps at 460 volts. Changing it to 230 volts means it now requires 80 amps. I'm not sure it's getting enough current.

230V @ 40A is over 10HP. Unless more power than this is needed it shouldn't be a problem. I don't think the small relatively slow moves we are testing should take much power.

[Charles2]

Hi Tom,

The transformer is labeled as follows:
KVA 24:3
Primary 230/460 Delta
Amp 54.9/27.5
Secondary 275Y/159
Amp 56.12 AC

This transformer only serves the spindle and axis drives.

There is a second transformer for 120 VAC and 39 VAC which feeds the 24 VDC power supply. Also a hydraulic pump motor and coolant pump motor.

Found an AMC 25 amp peak PWM brushless drive with built-in power supply, 120VAC input. Most of what I see require a separate power supply.

Thanks,
CJS

[Charles2]

Hi Tom,

I am looking at an AMC 30A8 drive that is rated at 30 amps peak, 15 continuous, and 80 volts. The price is very affordable, but I'm guessing it will operate the motor at only about half the motor's capacity, which is fine for testing, and may be ok on this machine.
Am I correct in assuming the power supply should be rated at the same amperage and voltage as the drive output? Also, does the power supply need a transformer for isolation, or can the 120vac line current be rectified directly to ~84vdc?

Thanks,
CJS

[TomKerekes]

Hi CJS,

I am looking at an AMC 30A8 drive that is rated at 30 amps peak, 15 continuous, and 80 volts. The price is very affordable, but I'm guessing it will operate the motor at only about half the motor's capacity, which is fine for testing, and may be ok on this machine.

I agree. Voltage limits motor RPM. So expect about half maximum velocity (assuming the original supplied 160V?)

Am I correct in assuming the power supply should be rated at the same amperage and voltage as the drive output?

That would be best but not absolutely necessary. Higher would be ok but not useful. Lower will simply limit the speed and accelerations that you will be able to do.

Also, does the power supply need a transformer for isolation

Yes

or can the 120vac line current be rectified directly to ~84vdc?

No. Rectified 120VAC RMS would be ~ 170VDC. If the amplifier is limited to 80VDC never apply more than 80VDC. Note some non-regulated linear supply's voltage increase when unloaded so you should be careful with this. AnTek is a commonly used supplier for linear supplies.

[Charles2]

Took a break to work on another project while waiting for the AMC drive to come in. I got a USPS tracking number, but the package never showed as accepted by the USPS. Ebay eventually refunded my money. Guess I have to order again and hope for better results.

[Charles2]

I'm finally back on this project. I bought a used Advanced Motion Controls 30A8V (30 amp peak,15 amp continuous, 80 VDC) brush type PWM servo amp to try out on Z axis because my original amps were not working out. I bought a 77VDC, 13A, 1000 Watt power supply from Antek. It is half the original voltage of 160 and much less than the original 40 amps, but should be ok for testing. Actually, I can't see why this small machine came with such large motors and drives. Connected both and started all over with testing. It seems to work pretty well so far. I can get pretty good plots. Will upload later. Problem is the output is spiking. The higher the P gain is raised, the higher the frequency of the spikes. The higher the frequency, the smoother the axis.

[Charles2]

Plots and data

[Charles2]

Resolution=80,000 counts/inch.
Z axis travel= 11 inches
X axis travel= 8 inches

[TomKerekes]

Hi CJS,

That doesn't look too bad.


You might add a 2nd order low pass filter to reduce the 160 count spikes due to tje D Gain. Without any filter whenever the encoder suddenly changes by 1 count the output will spike 160 counts. If you zoom in you can see this:

Attachment 459044

Usually a 500Hz Q=1.4 2nd order low pass works well.

To move at max velocity the amplifier is requiring around 600 DAC counts. Without Feed Forward or significant I Gain there has to be enough position error to create this output. Try increasing I Gain or VFF to reduce the constant error while in motion.

After that this should be what is remaining:

Attachment 459046

Is this the 'spikes" you were referring to? This looks to me some sort of repeating pattern. Something like lead screw wobble or something. But I believe the motion is 1.25 revs? This looks to be ~5 cycles. You might try different move sizes and velocities to try to determine if its period is based on time or position.