Hi,

my math:

32mm ball screw diameter / 1um of CNC machine accuracy = 32,000 ppr,

encoder resolution = 32,000 * 4 = 128,000 cpr (below delta encoder spec)

No, not quite.....the diameter of the screw is 32mm, the

**pitch** is 5mm per rev. Thus for a resolution of 1um

I require an effective pulse rate of 5000 cpr, the numerator/denominator could be:

160,000=5000 x N/D

so N/D =32

Thus the choice of N and D is very wide, for instance if D=1 then N=32 OR

D=10 then N=320 OR

D=25 then N=800 and so on.

At 3000 rpm (50 rev per sec) the pulse rate is:

pulse rate=50 x 5000

=250kHz.

As you point out the ESS has a max rate of 4Mhz, so at 250kHz its cruising. 250kHz is a little faster than the max recommended

single ended (open collector) signaling rate but very comfortably within the 'low speed' differential signaling spec of the servo

drive of 500kHz.

I have some experience with an Allen Bradley servo which I wanted to signal at 466.66kHz. Its servo drive has a recommended

max differential signal rate of 500kHz and a single ended rate of 200kHz. I can assure you that while 466.66kHz

**is possible**its much easier to signal at a lower rate. Thus I reduced the angular resolution to 2000 cpr (from the encoder max of 8000cpr)

and got the signaling rate down to 116.66kHz which can be done comfortably with a single ended transistor.

Note that this servo is used as a direct drive spindle which gives a resolution of 10.8 arc min ( with 2000 cpr) from a max

resolution of 2.7 arc min (with 8000 cpr). I practice even 10.8 arc min is much greater than I require. Reducing the resolution

has made the whole thing easier to do and

**still exceed** my actual requirements.

My existing mini-mill has a linear resolution of 1um and is very adequate. Thus I am quite happy to aim for 1um resolution

in this new build. I could, given the capacity of the servo encoder, have a resolution of 0.03125um ( 31nm!!!) but at the

expense of complexity of the signaling side of the design. I struggle to even measure 1um......let alone 30nm so why

bother?

The rated speed of the 750W B2 series servos is 3000 rpm. Direct coupled to a 5mm pitch ballscrew results in a max axis

speed of 15m/min which is a marked increase from my existing mill of 1.2m/min. I have experimented with pushing the

max speed parameter out to 5000rpm, the servos max speed (cf rated of 3000rpm) and they still work a treat.

Thus I could have rapids (G0's) of 25m/min in field weakened mode and full rated thrust for machining moves (G1's) of

15m/min while still maintaining 1um resolution and a max signaling rate of 416.66kHz, well within the 'low speed'

differential spec of the drives. As far as I'm concerned that is speed

**AND** resolution beyond what I ever anticipated

I could achieve with a hobby machine. If I want to use the full acceleration potential of my existing mini-mill (I can

tune it to 1g accel easily enough) I have to tie the machine to the wall or the machine tries to dance all over the floor.

The practical limit for the machine is about 0.05 g without undue machine motion and still provides excellent tool path

following (at modest axis speeds).

These 750W servos offer at least a 50 fold improvement in potential over the existing stepper/reduction drives I currently

use. Thus with the X axis bed/ballscrew/rails/vice/workpiece weighing over 200kg if I tune the axis to 1g acceleration

(of which the servos are more than capable of) then I'm going to have to bolt this machine to a concrete floor just to stop

it lurching around threatening to crush any-one or any-thing in its path!

The speed, torque and resolution of these servos continues to exceed my expectations. 15 m/min G1's sounds fine on paper

but when you see the axis machining at 15m/min you realize just how much power and even how much potential damage

can be done.....its scary!! Fun scary.....but still scary.

Craig