RE: [CAD_CAM_EDM_DRO] Re: DC motor controller

If you go to the Rutex site they show a graph of output position against
time to a step response input.

In control systems a step response is when you tell the servo to go x number
of pulses in zero time. Obviously this is impossible it will take some time
to accelerate, and then decelerate to position.

(To those control systems engineers out there, I am going to try and keep
this simple and avoid PID, nyquist, poles and zero's etc)

There are several things that can happen to a step response.

In a slow system you see it move to position, and slow down before it gets
there taking a long time to move the last little bit to position.

In a fast system, overshoot (go past it a little), and then pull itself back
to position. But it may get to position, or it may oscillate above and below
for a while.



The servo drive has a gain figure which controls how quick it gets there
when there is an error, more error and it must get there quicker, but too
fast and it will overshoot, and too much gain causes this oscillation which
you don't want.

You normally up the gain until you get a little oscillation, you then add
some damping to stop the oscillation.



Depending on the servo control there are then other adjustments that work
slightly differently with each manufacturers servo drive.



Remember the step response is not a real situation, it's a standard test
method, software like mach3 would control the acceleration within the limits
of the machine, and then the servo's work is much easier.

The thing is to allow the servo to know when there is a slight error so it
can correct it within a few encoder pulses, you give it room to play, but so
fine you never see it.



If you watch commercial machines with controls of the 1980's and early
1990's fitted with hand wheels perform a fast G0 move to position before
slowing to cut, you will actually see the index on the hand wheel stop and
reverse a bit. As a control systems engineer, the control is rather sloppy,
but a normal cut is very much slower so the system is then much more
precise. Most of these machines typically use 250 cpr encoders, if you put a
2500 CPR encoder on and run mach3, you won't see any over or undershoot.
It's real impressing to CNC operators that think our DIY stuff might be
inferior.

This is also why we tend to rough cut, and then finish cut with a slower
feed which is more accurate as well as having a better surface finish, the
older machines suffered from a lot more positional lag during normal
operation.



To get a fast machine running well might take several hours to set up the
servo drives. I never really know when to stop. I like to see a fast
response with very very little overshoot, perhaps 1 to 2 counts max. But I
also see machines especially some of the old ones with analogue controls
that are very sloppy and they still work quite well, it depends how hard you
push them.



Hope this gives you a little insight

It will all make sense when you start to play.



Regs Mark



_____

From: CAD_CAM_EDM_DRO@yahoogroups.com
[mailto:CAD_CAM_EDM_DRO@yahoogroups.com] On Behalf Of John Hansford
Sent: 12 November 2006 11:44
To: CAD_CAM_EDM_DRO@yahoogroups.com
Subject: [CAD_CAM_EDM_DRO] Re: DC motor controller



Thanks Mark,

I'm trying to follow along, but I've never seen a servo setup
run. I feel like I did 4 years ago when I got my first kit
from stepperworld with 80 oz motors. The vids I've seen show
them(servos) zipping around with real authority.

I am greatly appreciative of all this help and advice.

I've got some stuff ordered to get into "Trial & Error" phase
as soon as possible. I guess a great deal will become clear
when Hook-up time comes.

I'll refer back to these posts many times.

I usually build a machine 2-3 times before I'm satisfied with it,
so I'm not counting on getting it right the first time.
The main reason the machine's 80% MDF.

For anyone interested, I'm going with the treadmill motors &
cheap encoders from surpluscenter. Gecko G320's.
Mainly because something similar has worked for Les for years.
I need to run ANY servo to get an idea of what to expect.
I can't go wrong having these motors on hand.
The parts I need(for now) are basic geometric shapes from MDF.
I won't know what I want to improve til I see something running.
The Gecko's are akin to currency...If I can't use them somebody
else can.

I want to do this in a way that the next guy can learn from it too.

We're all very fortunate to have such helpful and informed people
from all around the world giving their input.

Thanks Mark,

John

--- In CAD_CAM_EDM_ <mailto:CAD_CAM_EDM_DRO%40yahoogroups.com>
DRO@yahoogroups.com, "Mark Vaughan" <mark@...> wrote:

>
> Les wrote >>. Higher CPR encoders make it easier for the drive to
> compensate.
>
>
>
> This is a very good point for diy'ers. Many diy'ers look at the

output speed

> of programs like mach3 together with acceptable accuracy/resolution to
> decide how many count they will select for their encoder. But

resolution and

> accuracy are not necessarily the same.
>
> Many modern servo controllers have a divider to allow you to set one

step

> input pulse to correspond to several encoder output pulses. This

allows you

> to use a hi count encoder so the servo drive has very good accuracy and
> several counts per step so it can see errors within each step and

keep those

> error small to much less than one step resolution. The divider then

lets you

> set the step resolution, and maintain speed with your step

resolution being

> pretty much your maintained accuracy even with a little over and

undershoot

> error. (cogging problems excluded in poor servo motors). Try selecting a
> pulse count resolution of at least 4 times your step resolution.
>
>
>
> Regs Mark
>
>
>
> _____
>
> From: CAD_CAM_EDM_ <mailto:CAD_CAM_EDM_DRO%40yahoogroups.com>

DRO@yahoogroups.com

> [mailto:CAD_CAM_EDM_ <mailto:CAD_CAM_EDM_DRO%40yahoogroups.com>

DRO@yahoogroups.com] On Behalf Of Leslie Newell

> Sent: 12 November 2006 09:29
> To: CAD_CAM_EDM_ <mailto:CAD_CAM_EDM_DRO%40yahoogroups.com>

DRO@yahoogroups.com

> Subject: Re: [CAD_CAM_EDM_DRO] Re: DC motor controller
>
>
>
> Hi John,
>
> The servo drive will have to work to reduce the effect of the cogging.
> Modern drives are pretty good and can mask most of it. On my mill, if I
> watch the pulleys very closely I can just see the motor speed varying
> slightly at certain feed rates. It appears to have no discernible
> effect on accuracy or surface finish so I don't worry about it. Higher
> CPR encoders make it easier for the drive to compensate.
>
> Les
>
> John Hansford wrote:
> > Hi Dave,Ron,
> >
> > Pictures...
> > http://www.flickr <http://www.flickr

<http://www.flickr.com/photos/26361144@N00/> com/photos/26361144@N00/>

> com/photos/26361144@N00/
> >
> > I took the CW to mean the nut would probably fly off if it ran
> > the other way. I opened it ...Nothing flew apart.
> >
> > What's the Bad news...How _BAD_ could they be as servos?
> >
> > I wouldn't have noticed the very slight cogging if I weren't
> > trying to find it.
> >
> >
> > John
> >
>
>
>
>
>
> [Non-text portions of this message have been removed]
>

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