Help! Consistent Gecko Fault or Servo Stall

[Noah]

Arrg!! My machine is working well, but has the repeatability of a blind squirrel. I can't run more than 40-50 lines of code before an axis will fault for some reason. I can't seem to find any possible explanation for this. I don't know if this is the servos stalling or what.

I am running G320's with 1000 PPR Renco differential encoders and Electrocraft 60v servos. I am a little unsure here. I called Mariss and he said a diff. encoder could be used as a TTL so long as only channels A and B were connected. He also said I could use 10 ft cables no problem as the geckos can filter noise, but at 45,000 Hz isn't it more than likely that the servos will fall 128 counts behind?

The only other cause that I can come up with is my ballscrews are too tight and I am stalling the servos. I set about 100 lbs of preload so it takes a fair amount of force to turn them over, but I can still do it by hand so a 720 oz-in should have no problems right? The machine snapped one of my cutters like a twig today so I am also skeptical about this cause. Please help.

[Bubba]

Noah,
If you have differential encoders, they generally take about 125 ma of current. The Gecko can only supply 50:{(
You have to run an external power supply for the encoder and put a resistor on the drive (forget which pins), but it is in the instruction sheet.

Had the same problem!

[Noah]

I am currently powering the encoders off my computer's USB port since they are powered at 5 volts. I will try another supply today though as I am not sure if Windows is capable of shutting off power to those ports every once in a while.

[HomeCNC]

Are you using Mach 2 as your controller? If so when I first setup the new version on my router my servo system did the same thing. The code would run for about 20 to 45 seconds and then with a very fast shake, one drive or two would fault.

I had to play with the speed and acceleration in Mach 2 before it would stay running smoothly.

I did not have this problem with an older version of Mach 2. When I upgraded to the new release I had all my setting the same. Art must have made some changes that made the system more sensitive.

[Noah]

Thanks guys. I switched to a separate +5v supply and that seemed to fix everything. Thats too bad though. I had really wanted to use the USB supply. I am still thinking of resetting the preload on my screws anyway as 100 lbs seems a bit excessive. How much preload do you normally have to apply to a rolled ballscrew to eliminate backlash?

[HuFlungDung]

100 inch lbs? I don't know, but that sounds a bit much. It may cause flaking of the raceways. The backlash is eliminated when there is zero longitudinal shift. The screw should feel quite solid in the nut, but should not feel gritty as you roll it. If there are any tight spots in the screw's length, then back off the preload until it will go over the tight spot without much force. That is about as tight as you can run it, IMO

[Noah]

Yeah, I thought 100 lbs. was too much. I will back it off. One last question. Mariss said it would be possible to use the other set of outputs from the encoder (A& and B&) as computer inputs to get a fully closed loop system. I noticed Mach 2 accepts quadrature inputs for 3 encoders. Does this mean mach 2 is capable of error correcting closing the servo loop?

[ger21]

No, Mach2 can't close the loop. I think the inputs are for jogging wheels.? You'd need to find software that does close the loop. $$$

[Noah]

Ok, thanks for all your help guys.

[sendkeys]

ger21 so is there really closed loop software that works with gecko's via the printer port.? I wouldnt might looking into it if you know the name of some.

anything that runs as fast as the mach2 ?

thanks

[ynneb]

When people talk about "closed loop", what do they mean?
Could some one explain it to me in very simple terms?
Thanks

[HuFlungDung]

Benny,
In an "intelligent system", not only does the controller have to issue a movement command, it also needs to receive verification that the position was reached as commanded. This is the so called servo "loop".

In a typical stepper system, the movement commands are issued, but no check is carried out to see if the movement succeeded in execution. This is "open loop".

In a servo system, the controller monitors the encoder position. There is logic built in that compares the current commanded position to the actual position of the encoder. As soon as there is a discrepancy, this generates an error signal in the controller. This error signal is amplified by the servo amps into a useful voltage and current to drive the servo motor. The motor turns, and all the while the controller is looking at the feedback from the encoder.

As the current encoder position is fed back and compared to the commanded position, the controller decreases the error signal, because the positional error is decreasing as the motor moves the machine. The servo amp also cuts back on the voltage applied to the motor. So an equilibrium is reached when the encoder shows zero error between the present and the commanded position. Then motion stops.

So if there is a large inertial load on the servo motor during the movement, the controller is still looking at the encoder feedback, the error signal still exists for as long as it takes for the encoder to reach the equilibrium position. If the motor turns, the encoder counts, and thus there is some "intelligence" to the circuit because the feedback loop is always reporting, and the motor will receive voltage until such time as the fuse blows, or it reaches position.