I've got 3 ametek 50V servos hooked up to a 48 volt power supply via gecko 320s. The peak torque value is 600oz/in. It's a chinese mill, about equivalent in size to a Bridgeport I. I've replaced the acme screws with 5mm (0.2") lead ballscrews, and have counterbalanced the knee with pneumatic cylinders.
The knee, when counterbalanced correctly, stalls lifting about 300lbs. I haven't done the math, but it seems like something is wrong. I've used a Bridgeport II with 40 in/lb servos(about the same), for years without ever stalling the knee. All of the mechanical stuff is clean, new paired angular contact bearings, oiled ways, etc. I have a 2:1 pulley reduction on the knee, so it's about 2.5 mm/revolution travel.
The X and Y are better, but a good shove can fault them as well.
I'm using Mach 2, and have the motor acceleration and speed turned way down. Any suggestions?
You may be hitting the tight +/- 128 count following error, this can happen very quickly if your encoder it a higher count one.
Also you're current may be set a little low. Where is the current limiting pot on the Gecko Set? What are the motors rated at?
Or more tuning may be in order.
Lifting a heavy bridgeport knee is always going to be a challenge. 2:1 reduction may not be enough.
It's so much easier to have the z axis on the quill. albeit a fiddle on a manual b/port. The being so much lower in mass, makes for a more dynamic z axis, which becomes quite critical for 3d work.
I have the same set up as you. Amtek motors, with 320 drives. BP style retro fit. I had 1000 count encoders, it would fault when you turned up the speed. After reading the post on the gecko forum, I have figured out, need around 250 count encoders.
How many count are the encoders?
The tune on the geckoout of the box should be really close. I turned both pots to around the 10 oclock postions, and it works good.
DRD, that is why I replied asking what his encoder count is.
1000 line encoder would be 4000 in quadrature, which would mean an error greater than 128/4000 = 3.2% of a revolution would cause a fault. If you had 5tpi ballscrews, and a further 2:1 belt reduction, then a position error greater than about 0.0032 inches of table travel would cause a fault. It was much worse on my machine with 10000 count encoders Getting the tuning exactly right helps a bit.
Good info!
So how many count encoders did you run? I have a 5tpi ball screw and 2.1 reduction. I was thinking around 250 counts, but don't know if 400 or 500 counts would be better?I am only looking for .0005 resoultion.
Oh, I still run the 10000 count encoders...I just developed my own servo controller that had an adjustable position error to deal with the high count encoder issue.
Encoders are 500cpr quadrature 2000 pulses per revolution.
The X and Y drive mechanisms are the same: 19 groove HTD belt on motor, 28 groove on ballscrew. Ballscrew lead is 5mm (0.2") single start
The Z mechanism is the same as the X and Y except that the pulley on the screw is 38 grooves, for a 2:1 reduction resulting in 2.5mm(0.2") travel per revolution of the motor.
The motor is a 40 VDC, I said 50 in the earlier post, sorry. The resistance is 5 ohm, so if I understand it correctly, its 8 amps.
I realize the knee is very heavy, hence the pnuematic cylinders. With the pressure set right, the pulleys turn by hand pretty easily. I wanted to quantify this, it takes 464 oz/in of torque to open a jar of apple sauce, I'd estimate the torque for the knee at about half a jar of apple sauce. My wife could probably manage lifting the knee, but would complain about the grease.
1) It sounds like the system should work, just try adjusting the drives, and perhaps switch to 250cpr encoders?
2) If that doesn't work get a bigger pulley for the z screw?
3) Finally, I've seen 72VDC Ameteks on ebay. I could build a 80VDC power supply and run them instead?
One more question, to verify the voltage from the drives can I disconnect the servo power leads from the motor, plug in a multimeter, and displace the encoder? I should see about +-40 volts when it's displaced before faulting, correct?
LinkBack URL
About LinkBacks
