RE: [CAD_CAM_EDM_DRO] Re: Bridgeport Ballscrew Backlash

Hi Scott

Looks very nice, especially the shiny paint, and polished belt casing.



Just a couple of queries re your problem which may not have been considered.

1 The bed play if mechanical could also be the bearings at the end of the
screws, this may be more probable to have play than the ball nut and screw
itself. I have seen many ballscrews fitted that just have a basic ball race
without any thrust race which by design needs some play. There are also
different axial float tolerances between bearings of the same size, some so
called higher spec bearings are designed for higher speed use and have more
play to account for expansion. It's would be easy to go to the bearing shop
for some new ones, and be sold what looks better but in our case isn't. On
long ball screws you control all the end play at one end of the screw and
leave the other to float again due to thermal expansion, one of the reason
more accurate machines use glass slides on the table rather than encoders on
the spindle. Before someone buts in, I haven't a clue what's on your
Bridgeport. You may be able to get a DTI from the bed onto the end of your
screws to check this.

2 Are the servo drives set up correct. When you set a servo up, it's
response will look nice with P and D parameters, but it will stop before it
gets to position or just after, there will always or near always be an
error, it needs a little more oomph on the last bit. As you add I overshoot
will increase and the error will become zero. Then you play with the figures
to get a crisp response without any error. The difficulty is knowing whether
there is an error or not, easy with a Rutex drive, you have a computer set
up graphically showing response position error etc, I'm not sure how easy it
is with a gecko drive unless you add a counter to the encoder signal. I have
set up many baldor drives, with pots, a pulse generator and a CRO, we had no
way to know if there was an error, only the shape of the response. We would
manually measure backlash with a handle on a screw, then use a DTI to set
the drive so the error after backlash was zero.

If there is a servo error it would play havoc with the manner you use to
check your back lash.



As I said just an idea, but something easy to miss in the depth of it all,
would be nice if it's something simple you can tweak.



Regs Mark





Dr. Mark Vaughan Ph'D. B.Eng. M0VAU

Managing Director

Vaughan Industries Ltd, reg in UK no 2561068

Water Care Technology Ltd, reg in UK no 4129351

Addr Unit3, Sydney House, Blackwater, Truro, Cornwall, TR4 8HH, UK.
Phone/Fax 44 1872 561288

_____

From: CAD_CAM_EDM_DRO@yahoogroups.com
[mailto:CAD_CAM_EDM_DRO@yahoogroups.com] On Behalf Of Polaraligned
Sent: 24 January 2007 20:08
To: CAD_CAM_EDM_DRO@yahoogroups.com
Subject: [CAD_CAM_EDM_DRO] Re: Bridgeport Ballscrew Backlash



Thanks Jon, I appreciate any help here.

The ways were not rescraped. They looked to be
in decent condition. Automatic oil pumps helped
save a lot of these machines.
I put a precision level on the table and shimmed
it level at one extreme of travel. At the other
extreme of travel, (x-axis), I needed to shim the
level up .006" in 10" to be level again.

I snugged down on the X and Y gibs again and was able
to rock the table .002", pulling the table towards me
and measuring about 13" out from the centerline of the
table. Bridgeport's spec for this measurement is .0005".
Weird thing was that after the first pull that moved the
table .002", I was not able to move the table more than
the .0005" spec. The gibs seem fairly tight and I am
hesitant to tighten them more.

As for ballscrew alignment, I used the same factory housing
for the motors and pulleys. The leadscrews on these
machines has only one end restrained so I would think
if the screw was not in alignment it would show as extra
backlash only on one end. I would think that the "unrestrained"
end would be the best, and it would get worse as you
approached the restrained end. Of course, I could be
wrong here.

I replaced the original steppers with Baldor 12.5 oz-in
servos with 1000 line encoders. They are reduced 4:1 with
Gates GT2 timing belts. Using Gecko drives and Mach.

I test the backlash by mounting a .0001" graduated dial
test indicator to the spindle and using a 1-2-3 block
mounted to the table for a straight edge to read off of.
I set the MPG for single step mode with a resolution of
.001" and bring the block edge into the test indicator.
I step the block so the test indicator is reading in it's
mid range, then count how many single steps are required
to get the indicator moving again.

I don't even need the indicator to count the .001" increments
needed to overcome the backlash. As I hit the arrow key on
the keyboard I can hear and feel the ballscrew kick in.
The servo makes a slightly different noise and you can hear
when the ballnut "grabs". You can also feel the vibration
from the step pulse when the ballnut "grabs".

I redid the backlash measurements after tightening the gibs
down:
X- Center of travel: .009"
Right extreme: .013"
Left extreme: .011"

Y- Center of travel: .005"
Right extreme: .009:
Left extreme: .009"

I am testing the backlash over the course of less than
.020" of table movement. It is hard for me to comprehend
that the backlash is caused by anything other than the
ballscrew.

Help!!
A few pictures of my many hours of work:

http://smg.photobuc
<http://smg.photobucket.com/albums/v223/polaraligned/Bridgeport/>
ket.com/albums/v223/polaraligned/Bridgeport/

Scott

> Something is seriously wrong, here. I can think of a couple of possible
> explanations. Obviously, wear is not likely, unless the machine was

used

> in some very odd way that caused massive wear only at the ends of the
> screws.
>
> Did you scrape or grind the ways? If not, the ways may not be fitting
> well, as they have wear in the middle. You may be able to detect this
> by leaning on the table with a dial indicator showing deflections

against

> the spindle. Generally, the top of the saddle wears convex, and the

bottom

> of the table wears concave, allowing the table to rock when the

direction

> the screw is driving it reverses. This LOOKS like backlash, but is a
> totally
> different phenomonon. If you have a master precision level, you can

lay

> that
> on the table and reverse the travel direction. If it is rocking, you
> will see a
> sudden shift in the level right at the reversal.
>
> Another possibility is the screws are not aligned well with the table
> travel,
> and the bow in the screws is both binding it, and causing a

deflection of

> the movement. This would be expected at the ends of travel. If you
> reground
> the ways, the screw and nut will no longer line up, and some adjustment
> will have to be made.
>
> Jon
>

[Non-text portions of this message have been removed]