Steppers vs. Servos - a clarification
Posted by
Mark Fraser
on 2001-08-09 11:13:16 UTC
My little Sherline mill is set up with steppers - the number of steps
per rotation x the number of threads per inch yield a ratio that gives
8,000 steps per inch. Even at 4,000 steps per inch, an error of 5% in
one step (which is really what the accuracy figure is about) will give
me a positional error of 1 part in 80,000 for THAT particular step.
Over one rotation of the leadscrew (200 or 400 steps) that error
comes back to ZERO.
I sort of suspect the leadscrew error will be greater than the gap
between steps, and the error in the position of one step relative to
the adjacent one will be so small as to be insanely minute for onlyh
the most dedicated hairsplitters.
In a much shorter form, the staircase on a 45 degree line will have
zigs that are separated by 1/8000 th of an inch. I don't think
you'll see it.
Servos DO have an ability to resolve very small amounts. Most likely,
a servo on my Sherline would be able to position the table to a
resolution
of only 1/100,000th of an inch, rather than 1/8,000th for my steppers.
Frankly, Scarlet, I really don't give a d***.
The real problem, as others have pointed out, is that I'm likely to
dig a 3/8 inch endmill into a block of 316 stainless with a .100 inch
depth cut at 10 inches per minute. The steppers will be incapable of
pushing hard enough, and the PC will think (hope?) that the steppers
have followed instructions. I'll get a positional error of 1/8,000 of
an inch for every lost step. And a loud noise.
My 150.00 Camtronics box and $15.00 worth of steppers can be upgraded
to use 30.00 worth of bigger surplus steppers, and I can upgrade my
12 volt supply to a higher voltage, which will about double or triple
the table-moving force/speed I've now got, which will be wayyyyy cheaper
than
upgrading to servos of ANY kind.
When I get ready to do production work (Sherline?) where every minute
of machine time is money, I'll go servos if it makes fiscal sense.
/mark
per rotation x the number of threads per inch yield a ratio that gives
8,000 steps per inch. Even at 4,000 steps per inch, an error of 5% in
one step (which is really what the accuracy figure is about) will give
me a positional error of 1 part in 80,000 for THAT particular step.
Over one rotation of the leadscrew (200 or 400 steps) that error
comes back to ZERO.
I sort of suspect the leadscrew error will be greater than the gap
between steps, and the error in the position of one step relative to
the adjacent one will be so small as to be insanely minute for onlyh
the most dedicated hairsplitters.
In a much shorter form, the staircase on a 45 degree line will have
zigs that are separated by 1/8000 th of an inch. I don't think
you'll see it.
Servos DO have an ability to resolve very small amounts. Most likely,
a servo on my Sherline would be able to position the table to a
resolution
of only 1/100,000th of an inch, rather than 1/8,000th for my steppers.
Frankly, Scarlet, I really don't give a d***.
The real problem, as others have pointed out, is that I'm likely to
dig a 3/8 inch endmill into a block of 316 stainless with a .100 inch
depth cut at 10 inches per minute. The steppers will be incapable of
pushing hard enough, and the PC will think (hope?) that the steppers
have followed instructions. I'll get a positional error of 1/8,000 of
an inch for every lost step. And a loud noise.
My 150.00 Camtronics box and $15.00 worth of steppers can be upgraded
to use 30.00 worth of bigger surplus steppers, and I can upgrade my
12 volt supply to a higher voltage, which will about double or triple
the table-moving force/speed I've now got, which will be wayyyyy cheaper
than
upgrading to servos of ANY kind.
When I get ready to do production work (Sherline?) where every minute
of machine time is money, I'll go servos if it makes fiscal sense.
/mark