encoders on ballscrew ends

[jguillen08]

Does any one know if there is problem with mounting the encoder directly on end of the ball screw? I plan on using a 3:1 ratio on the pulleys. The drivers will be either Geko or Rutex, I haven't decided.

Thanks,
Joe

[CNCadmin]

I believe that is not a good idea, I can't really explain it, but encoders are precise, and running them off of a object that has play in relation to the drive system to me will make set-up and tunning very difficult.

[sendkeys]

i seen people say it's fine and others say dont do it :P

reasons i seen are.

do it so it will take care of any backlash in the belts

reason i seen not to do it are the backlash in the belts will make the servo vibrate .

So guess if you have good belts and a smaller encoder i say go for it. but low cost pulleys and large encoders i wouldnt

[Bubba]

Joe,
I have been running my encoders off the ball screws (and microfeed for Z) and am quite happy with the results. You need to have backlash free transmission between the servo and the leadscrew or the system will "dither". The looser the belts, the more the dither.
I did it this way as I thought I might have (want) to play with different reduction ratios and by mounting on the screw, would not have to reconfigure my software each time as I would if they were mounted on the servo itself. It was also easier (I'm lazy) to make the mount for the screw rather than trying to make a mount for the encoder on the servo. Also remember when deciding on the line count that if mounted on the servo you need to take the ratio into account as well. Line count will affect your resolution AND possibly top speeds. This is an area of tradeoffs and only you can make the decision of resolution/speed that you want.

[jguillen08]

Thanks for all the input.

[HomeCNC]

I wish I had the explanation that Mariss of Gecko drives gave me. Mariss is the smartest person I know of in the area of servos and servo drives. He is against doing this. His reasons are technical and like I said I wish I had his answer that he gave me so I could copy it here.

[Mariss Freimanis]

Thanks, I'll save the previous message and show it to my kids frequently. They often question my intelligence.

I think there is some mixing of apples and oranges here.

1) It is not a good idea to use a linear encoder with a single-loop PID servo like a G320. The problem is backlash. Assume the motor is always in motion, even while "stopped". With an encoder mounted on the motor shaft, the motor "ping-pongs" between two adjacent encoder edges. This normally is not noticable because (a) the angle is small (0.18 degrees for a 500-line encoder) and (b) the servo response is made intentionally sluggish over that angle.

Things change, and not for the better either, when backlash gets included in the mix. The servo still "ping-pongs" between adjacent encoder edges as before, but now the motor must take-up the backlash distance before motion can be detected by the linear encoder. In the meantime, the motor may build up a good head of steam (speed) if backlash is say 90 degrees of motor motion. Motor speed would be 90 / 0.18 or 500 times higher when it's time for the next ping or pong!

The result would be a continuous and vigourous thrashing between the extents of the machine's backlash. It would be noisy, place bad wear on the screw nut and heat the motor with all that activity. Not a pretty picture.

2) Encoder on the screw. This is OK and actually better than having it on the motor. The proviso is there must be no backlash in the motor to screw reduction gearing. Toothed belts and pulleys easily meet that criterea.

One "better" part is it eliminates cyclic errors with the reduction gearing. Imagine you have a pulley 1" in diameter and a chunk of something 0.01" thick lands on one of the teeth. That increases the radius of the pulley 2% while the chunk is between the belt and pulley. This would result in a "wobble" in your expected position (cyclic error).

The other "better" part is the encoder line-count can remain the same even if you have to change your reduction gearing ratio. A 500-line encoder on a 5 TPI screw would give a 0.0001" resolution regardless of the reduction ratio, be it 2:1 or 20:1.

The downside is now it is impossible to bench-run the motor. The first time you run the motor you run the machine as well. That can be very exciting if something major (like a miss-phased encoder) isn't just right. :-)

Mariss

[HomeCNC]

Oh I had it backwards! Thanks for posting this Mariss.