[tvan56]

Am Planning on fitting a knee mill with 2 and eventually 3 axis
operation. Been looking at some Ac Servo Motor/Drive combos.
Looks like most commercial knee mills w 9x42 table use motors
around 27-29 in/lb with 41 optional. Should one look for low
or medium inertia motors? Will have 1" or 32mm ballscrews.
Could one "get by" with 750 watt 2.39NM cont. 3000 rpm motors?
I only do onesy-twosy jobs. Rounding corners, pockets,
bolt patterns. Max performance not needed.
Also, in drives, I assume they work in "position" control mode
in an application like this. Centroids kits for 3rd party drives
have a +-10v control signal for the drives. Some drives only
work in position mode with step and direction.
Any advice appreciated.
tvan56

[Al_The_Man]

It all depends on the controller you are looking at and the speed of operation as to using low inertia motors.
Many hobby controllers use step/dir and require drives capable of closing the loop.
The type of controller that uses ±10vdc can use non-intelligent drives where the loop is closed back from encoder to controller.
There is a motor sizing program on the Kollmorgen site.
Al.

[tvan56]

Looking at Delta Electronics ASD-B2-0721-B Drive and ECMA-C20807
Motor from Keling. It has inputs for the encoders on the drive and pass
through encoder signals for a controller. Would that work with the Centroid
system?
Not wanting to do high speed work. I am aware of motor sizing programs. I don't know the weight of the table, the acceleration rate,
screw inertia,etc. My problem, I realize.
I have been a manual machine guy forever. I am just looking at what is out there as to what others are using. Do low inertia motors lend themselves to
quicker moves due to low inertia? If one where to have two equally sized
motors driving the same load < 10:1 inertia, and one was low inertia, the
other medium, what would the characteristics of each combo be?

[mactec54]

tvan56

If you are going with Centroid control, you will have to use what Encoders they spec for there control, it would be best just to get the whole package from them

Ballscrews 32mm would be correct for a machine that size

[tvan56]

Will check into the encoder thing. About the screws.
Hiwin has 32mm dia. jobs. Weak Yoke. Elrod offers, I
think 1" dia. and strong yoke. Just from looking at pics,
I'm guessing Anilam uses same screws as Elrod. Jena-Tec
We had a prototrak where I used to work. It had rather
small looking screw. Never measured it. I think it was
rolled also. I guess SWI maps their screws or something,
because that machine could do some good work.
I think Rockford Ballscrew may offer a 32mm because the
pics show the ball return tube protruding through yoke side,
like Hiwin. Worried about the single "preloaded" nut issue
and backlash.

[mike_Kilroy]

...........

1. Do low inertia motors lend themselves to quicker moves due to low inertia?

2. If one where to have two equally sized
motors driving the same load < 10:1 inertia, and one was low inertia, the
other medium, what would the characteristics of each combo be?

1) T=Jw/t where T is the Torque needed to make a move with J inertia to a speed of w in time t..... so of course if u r accelerating or decelerating u will make the move in less time with the same torque to or from the same speed if J is smaller (t=Jw/T) how much quicker? make J=Jmotor+Jload. u say Jload=10x Jmotor so say low J motor has inertia of "1" and medium one has J of 5.... low inertia motor total J=1+10=11; medium J=5+10=15

so t=11w/T or =15w/T hence low J motor will be accel in 11/15 or in 73% of the time of the med J motor....

for single parts u say u will do, do u think this will make a hill of beans difference? prob not.

2) see 1. other thing is if u do pick the single nut screw with backlash, the load is only 10/5 or 2x the inertia of the load. the low J is 10/1 or 10x more. the 10x ratio MAY be SIGNIFICANTLY HARDER to tune to make good accurate cuts so if u dont need to save that 100-73 or 27% of the med inertia accel time of 1/2 sec, then pick the med J motor in a heartbeat.

[Alaska Bob]

tvan56
I am running a Tree 2UVRC with Emerson MG-316 (750W) servos driven by EB-205 drives for X and Y (~32lb-in). Z axis (entire knee) is driven by an MG-208 (480W) servo and an EN-208 drive (~27lb-in). Z axis has a 10:1 gear reducer on the servo and a reduction through the belt drive as well. Not sure what the exact ratios are but I can get them for you if you are interested.

I get over 100 IPM in X/Y . Z is more like 30IPM I think. I would think 500Watt motors should get it done if you aren't looking for max performance.
--bob

[judleroy]

Why not look into the dmm tech servo systems. They may be a good fit for you.
Judleroy

[tvan56]

Thank You all for the advice and wisdom. Mike, that explanation
of yours made some light bulbs come on inside my head. Will
continue researching. Tuning ease is one thing I want to be
able to do easily. I was working at a place in the mid '80s and
was responsible for programming and interfacing a Cincinnati
Milacron T3-776 Robot. Those things were a bear to tune. Just
trimpots. Seemed like hit or miss. Throw some backlash in and
It would shake like hell in is moves, hunting.
You had to loosen it up so much it looked like the thing had
arthritus. Sluggish as hell. Replaced the worn screw and then
on could stiffen it back up. They were monitoring speed at the
motor with a tach and position elswhere on the arm for position
with a resolver.

[tvan56]

Here I am posting to my own reply, but just
thought I would put some stuff out here about
myself and my thoughts. This servo stuff is not
new. I am an old vacuum tube radio guy. It simply
amazes me what was done in the old days with
tubes. Regenerative radios in the '20s used positive
feedback to intentionally make them oscillate. They
would then just tune to the edge to get fantastic
signal gains. In the late '50s Harmon-Cardon made
an amp that is still a benchmark today. Citation II
Triple! nested feedback loops. Here is something I
found:
There are three nested feedback loops. The first is from the plates of the phase splitter tubes, to their grids. The second is from the plates of the output tubes to the grids of the phase splitter tubes. The third is from the output transformer secondary to the cathode of the input tube. The total amount of feedback is about 32 dB. This about 10 dB more feedback than a traditional Mullard or Williamson type tube amp. The multiple loops, and fantastic output iron make this possible. The time constants of each loop have been carefully calculated. 10 KHZ square waves (at about 2/3 power) look amazingly good, with just a tiny hint of ringing (one period). This amplifier is amazingly stable for any tube amplifier, let alone a tube amp with that much feedback.
One thing that we learn from history is this. We don't learn from history!
The older I get, things that seem so unrelated earlier, are
the same.
Sidetrack over...

[Al_The_Man]

I could always dig up some of my old prints on Thyratron closed loop drives, you may have a problem getting parts however.:violin:

As a guide for your retro-fit, I bought some Excello knee mills that were originally professionally retrofitted, but the controller supplier had gone out of business so they had to be re-retrofitted!
They have a DC servo 40lb-in with 2:1 reduction, ball screws are 1-1/8th" dia.
Note: The reduction increases the torque at a ratio of reduction, and the motor-load inertia ratio is decreased by the Square of the reduction.
Al.

[mike_Kilroy]

.... I was working at a place in the mid '80s and
was responsible for programming and interfacing a Cincinnati
Milacron T3-776 Robot. Those things were a bear to tune....

OMG...... I'm sorry. It sure seemed mighty advanced at its time... I was partially responsible for that robot servo design (: I supplied the servos and drives and helped in some of its servo design.... those servos sales paid for my home!

sounds like u, Al, & i could tell some cool old time stories!

[tvan56]

Actually those robots were the stuff at that time. Too
bad they sold that division. I believe ABB bought
them. A lot of the personnel went to Motoman. I went
to programming/maintenance school there and my teacher
was an old Cincinnati guy.The drives actually worked very
well. Tuning wasn't a constant issue. The only thing they
really needed was a way to determine which drive went
down if there was a fault. I think they might have had
the fault signals wired in series.I remember the Cincinnati
tech guy had built a device with some flip-flops to
latch on when the fault ocurred. It would indicate the
drive that first faulted.

[tvan56]

Found out Centroid wants a 2048 or 4096 count encoder for
their control if using ac motors. The Keling jobs have
17 bit (8192?)
Is there a way to "Dumb Down" an intelligent drive to allow
the controller to handle everything?
Do the unintelligent drives operate in Velocity or Torque
mode or what?
E-mailed Centroid to see if they had a list of third party
drive/motor combos that are known to work. No response
yet, but it's only been 1 day

[Al_The_Man]

Found out Centroid wants a 2048 or 4096 count encoder for
their control if using ac motors. The Keling jobs have
17 bit (8192?)

If differential, they may be 2048 quadrature encoders. some controls will allow the setting for x1 x2 or x4 basic quadrature resolution. 2048 x 4 = 8192.
Encoders should always be specified and referred to by the basic resolution or quadrature count. Many are under the impression that quadrature count means the x4 value.

Is there a way to "Dumb Down" an intelligent drive to allow
the controller to handle everything?

Some intelligent drives can be ran with the PID loop closed back to the controller and some also have the advantage that the encoder/resolver input is processed and a pseudo configurable count encoder output that can be made available from the drive.

Do the unintelligent drives operate in Velocity or Torque
mode or what?

Most will allow 3 or 4 methods of control, including torque or velocity.
The Galil Motion cards I use recommend the Torque mode, as does the drive manuf. A-M-C.

Al.

[tvan56]

I always thought that quadrature meant the basic count x4.
A 1000 ppr encoder, 4000 quad. Not true?

[tvan56]

Hey Mike,
I have a transformer out of a 776 Robot. It supplied
power for the DC buss for those drives. By any chance
do you know the specs for it? I realize there may some
fog to wade through. From memory it weighs 50-60lb.

[Al_The_Man]

Unfortunately this came about because of assumption.
The Quadrature encoder originally so named because of the 90° phase shift between the two pulses, and in the original applications the basic quadrature resolution count was used, it was then discovered that by utilizing the edges of the two pulses to increase the resolution this could be increased by either X2 (both leading edges) or x4 (the leading and trailing pulse edges).
Now many equate the term quadrature to this method of increasing the resolution and now when using the same term in two different ways to describe the same object can often lead to some confusion.
Page down to incremental encoder.
- Wikipedia, the free encyclopedia
Al.

[tvan56]

Most informative. I would venture to guess it started
with how many slots could be made for the optos.
then someone said, in salesmen style, We can tell
them we offer 4 times the counts as the other guys.
SOS

[mike_Kilroy]

Hey Mike,
I have a transformer out of a 776 Robot. It supplied
power for the DC buss for those drives. By any chance
do you know the specs for it? I realize there may some
fog to wade through. From memory it weighs 50-60lb.

hee hee tvan.... guess I didnt tell u I designed & supplied those Marleco brand transformers too did I I actually had to drive from CMI in cincinnati to Greenville, SC one time to help buyer explain the price I was charging to GM there......... but I digress!

I think u r off almost by factor of 10 on weight tho ok, ok, maybe not 10 but it still wieghed a good 220# IIRC....

Yes, fog.... but I believe I did bds4 drives with Goldline B20x for wrists and B40x larger models for other 3 axes... all 230v... so we combined all power requirements on the robot into 1 what we coined as a "combination" xfmr: all primaries in the world, 50/60hz, and a large 230v 3ph wye sec for the drives - fog... maybe 12kva-ish for the 776 model; more like 4kva for the 726 model...., a couple 120vac 1ph windings on (1 on each sec leg IIRC of about 1.5kva each for cnc and i/o.) with faraday shielding between each and the drive windings....

give me the M-13652 type part no and I will pm u purchase specs if u want
(we did supply a few plain 230/460v 60hz pri models to offer lowest possible cost alternative to the one size fit all model so I need ur nametag info if u want right specs)