real tests needed was Re: true closed loop vs. hobby closed loop

Mariss,

I'm glad you've posted this. Now to get some REAL info we need this
same test done with a MACHINE attached. Nothing like some added mass
to affect the numbers...

Because I'm pretty sure everyone here is accelerating more than just
the motor rotor...

Ballendo

P.S. Even better would be to add your scope to a drive that's
installed in a BP retrofit, or a large gantry table, and then run the
drive through some typical toolpaths. Repeat also on smaller
machines. THEN we'd really know how things stood up for comparison.
Which would be a good thing, IMO. As you say, nothing like actually
measuring something. (IN its intended use...)

We could perhaps get Jon E to do the same with his servo amps, and
then this whole issue would move from the theoretical to the
practical... (Perhaps someone with rutex's too?)

--- In CAD_CAM_EDM_DRO@yahoogroups.com, "Mariss Freimanis"
<mariss92705@y...> wrote:

> There seems to be a little misunderstanding about how the G320
> performs, so I ran some experiments.
>
> Signal source: G2002REV3
> Motor: Minertia TO6M (NEMA-34)
> Power Supply: Sorensen DCR 60-18B lab supply
> Encoder: 500 line (2,000 steps / rev)
> Drive: G320
> Scope: Tektronix TDS-360 (200MHz digital sampling scope)
>
> Test:
>
> The G2002 was programmed to move 25 revolutions at 2,400 RPM CW,

then

> 25 revolutions at 2,400 RPM CCW. The rate of acceleration was a

0.004

> second linear ramp to 2,400 RPM.
>
> The sampling scope was attached to the POSITION ERROR test point of
> the G320. The trigger was set to NORMAL and the vertical scale was
> 100mV differential to 5VDC (2 error counts equals 80mV).
>
> This is a vicious rate of acceleration. The motor had to be clamped
> because it jumped off of the bench on the first try.
>
> The test results showed a maximum following error of only 5 counts
> (0.9 degrees) on the 2,400 RPM CW to 2,400 RPM CCW slope. That is a
> net change of 4,800 RPM in 0.0078 seconds. More practically

seaking,

> that would be a 0.0005" following error if you were to go 480 IPM
> forward to 480 IPM backwards in 0.008 seconds on a 5 TPI leadscrew!
>
> I have some terrific 'scope pics that show this data but it seems I
> cannot post to the "Files" and "Photos" section anymore.
>
> In any case, the actual, measured following error is far far better
> than what has been mentioned on this thread. Nothing like actually
> measuring something.
>
> If anyone wants, I can e-mail the 'scope pics.
>
> Mariss
>
>
>
>
> --- In CAD_CAM_EDM_DRO@yahoogroups.com, Jon Elson <elson@p...>

wrote:

> > Well, without some interface between the Dan Mauch card and the
> Gecko,
> > it would probably be unreliable. You can't expect ground to be

the

> same at
> > the computer and the Gecko drive. An opto-coupler to repeat the
> encoder
> > signals would be a very good idea.
> >
> > You could set the following error to be quite a bit less than the
> 128
> > counts,
> > or set up the sliding scale so it would be low at lower speeds.
> You
> > might try
> > for +/- 10 counts at zero speed, and 25 counts per inch/second of
> velocity.
> > But, my experience with Gecko 320s has been that the allowable

error

> > band at idle is a lot larger than it is supposed to be.
> >
> > >On the Gecko G201/G210 if I were to ground terminal 7 (DISABLE)
> then
> > >you should be able to drive the system manually, using the

encoder

> to
> > >feed the DRO, just as you say, correct?
> > >
> > >
> > Oh, we're talking steppers here, I see. Yes, that is right. I
> can't
> > say anything
> > about loop stability with steppers and G201, and especially

G210s,

> which
> > have another control loop inside them (the pulse multiplier).
> >
> > >Sorry for the basic level questions, but I just got EMC working

on

> my
> > >laptop this weekend (WOO HOO!) and I am now going about the
> business
> > >of getting my Bridgeport Series 1 CNC retrofitted with new

drives

> > >(same old motors). Anyone have any luck fitting handle to the

old

> > >Bridgeport CNC machines? I can easily get one on the Z drive.
> The Y
> > >drive I couls maybe see how I can modify the belt cover and put

a

> > >handle in. What about the X?
> > >
> > >
> > This is a BOSS? The X leadscrew doesn't turn on a BOSS. The

nut,

> buried
> > DEEP in the knee is what turns. I can see handwheels on this

being

> almost
> > impossible! How about flexible shafts?
> >
> > J