Gearbox selection for mechatronic project

I am using a servo motor to make my own rotary indexing table which will have 6 nests. My servo motor has 65536 counts per revolution. I am thinking of ordering a 10:1 gearbox, so I would need 655360 counts per full revolution of the table but of course I am only indexing incrementally to the next nest thus 109226 counts. The motor in position limit is 50 counts (can be adjusted).

I have my motor setup at the minute with 999 accel and decel params and an rpm of 2000. It takes about 1 second to index which is what I was hoping for.

I am wanting to achieve repeatability of a wire loaded in the nests which has a diameter of 3mm.

Am I on the right tracks, or would you sacrifice index speed for greater resolution by choosing a higher ratio gearbox?

If you do as you say, you will be off by 0.6667 counts after each index. After 10,000 indexes you'll be off by 6,666.6667 counts.
If you could make an 8 stop rather than 6 stop index table, then your numbers are evenly divisible and you won't suffer that issue.
Or, you could fall back and solve it at the mechanical level by using a ring gear to not only support your table, but then would allow you to rim-drive the gear and between the gear ratio and the reducer arrive at a suitable motor index period so you don't end up missing fractional parts of counts.

Many servo drives allow to alter the number of feedback pulses sent to the motion controller just to be an even multiple of your gear ratio/number of stops. Of course the altered value cannot exceed the physical max, so in your case, if you can change your servodrive parameters to send 60000 pulses per motor revolution instead of 65536, you should be able to eliminate the totalizing error with a negligeable loss of positioning accuracy.

What is the backlash of the gearbox?, and at what radius are you trying to achieve this positioning accuracy?

The uneven pulse distribution shouldn't cause too much of a problem. As mentioned, you would lose pulses. But if you make two moves of 109226 pulses and four moves of 109227 pulses you will stay within your move tolerance and still end up with the total number of pulses you need in a single table revolution.

However, I would use a motor with a multiturn absolute encoder. Set it up in modulo mode and have it roll over at 10 motor revolutions. Then command your moves in absolute positions. You won't need to worry about lost counts if you do that and you only have to establish a reference once as part of commissioning or if you take the system apart (decouple the motor from the load).

Keith

Thanks for all the replies. This is for my final year project for my Beng in Mechatronics, so not industrial it does not have to be accurate as long as it is repeatable on demo day.

I picked up the gearbox this week; it's a fenner 742d0371d at 10:1. Backlash I will need to ask the supplier tomorrow. Padees I don't know the answer to your second question, for being a mechatronics course there sure wasn't a lot of mecha in it. I don't have all the stations designed yet so don't know the size that the turntable plate is going to be yet.

I originally was going for 8 nests, but went to 6 as 2 of the functions would be hard to achieve in the timescale I have (highly dexterious work for those 2 stations). However you have given me the idea of going to 4, which I think could be do-able with the benefit of keeping the footprint down.

I don't have choice in the servo motor, I am using one borrowed from work it is a Bautz AC servo controlled by a Pacific Scientific P800 drive.