Re: New Article Posted

Dave,

Sure, maybe. If you're willing to risk it, but why not just use half
step?

(I know, because we've all--or at least enough of us-- have already
been there before and KNOW the worthwhile difference that
microstepping makes...)

It's wonderful how we can convince ourselves that "it" IS or is not
a problem by making a whole pile of assumptions.

I'm going to assume that the Allegro engineers who said, "Don't do
it" know more than me. (Mariss might have to disagree<G>) Will it
average out as you've assumed? I don't know.

I DO know I was going to use the sync mode as Phil has done UNTIL I
realised what I've shared here in the thread. Then, because I
couldn't come up with a solution to ONLY instating sync mode on
equal level phase points; I left it off our boards.

I think we should stack the deck in favor of accuracy, whenever
possible.

Ballendo

P.S. AIRC, the A3977 has this same limitation... (There's a newer
Allegro chip in an TSSOP pkg which does keep track of the microstep)

--- In CAD_CAM_EDM_DRO@yahoogroups.com, "turbulatordude"
<dave_mucha@...> wrote:

>
> --- In CAD_CAM_EDM_DRO@yahoogroups.com, "ballendo" <ballendo@>

wrote:

> >
> <SNIP>
>
> > >but does not lose any steps when in sync mode. I've been using

this

> > >setup in half-step mode for months and have not had any

problems.

> >
> > Yes, in HALF STEP mode it will work fine. BUT you WILL lose

steps if

> > you try to microstep whilst using the sync mode. (or going into

and

> > out of it).
> >
>
> I thought about that today a little. I have a 13 tpi screw on a

1:1

> pulley on a 200 step motor. that puts me at 2,600 full steps and

some

> high number of microsteps. I think the driver is 8x ? 20,400
> microsteps ?
>
> Assuming the sync pulls the motor to the nearest 1/2 step, we can
> assume that anything one microstep closer than 'center' will fall

back

> and anything one mircrostep past 'center' will fall forward.

Center

> being that value equally between half steps.
>
> in my case, 1 inch = 1/2600th of an inch or about 0.0003846"
> a microstep would be 8 times that or 0.00004807
>
> Now, assuming there are 4 microsteps between each half step, and

the

> first two will fall back and the second two will fall forward.
>
> assuming further that I have 100 pauses across a part and in every
> case, the driver was on 2 microsteps. that would mean the sync

mode

> would have pulled the position back 2 microsteps each of those 100
> times for 200 total microsteps.
>
> with a microstep being 0.000048077 (rounded), then 200 microsteps
> would be 0.01" or 10 thou.
>
> the assumptions are;
> #1) that the drive actually does pull to the nearest half step

when in

> sync mode.
> #2) that every sync mode would drive from the greatest distance

from a

> half step
> #3) that every sync mode would drive the same way.
> #4) that the driver does not automatically recover.
>
> If the sync mode were split 50/50 all errors would effectively

cancel

> themselves out.
>
> As for my PCB etcher, I move to 1.000" or 4.500" or 0.01" some

other

> whole number or simple decimal that will be divided evenly by half
> stepping. (10 tpi screw on that) in those cases the sync mode

would

> never have any effect.
>
>
> Probability would favor that the plus errors and minus errors would
> cancel each other out to a large degree.
>
> And, even if one had 1/3 one way, and 2/3 the other, my 200 would

drop

> to 66. And, if half were one micorstep away and half were 2
> microsteps away, then those 66 times is only 99 microsteps or
> 0.0048077. or just under 5 thou.
>
> 5 thou is certainly enough to measure, but in so many cases, I

would

> never even fall on a microstep that it might be a rare occurance.
>
> Also, if one were drilling a bunch of 80 pin DIP chips or such, the
> first hole would be off by a max of 2 microsteps, but the driver

would

> have the axis pulled to the half step. and I assume that the driver
> would never know it was not on that axis, so it would not
> automatically add (or subtract) those microsteps. So, if the axis
> were to be pulled to the nearest 1/2 step, and the driver didn't

know

> it moved, then the other 79 pins would fall on the half step.
>
> I am sure there are applications where 5 thou is huge, and others
> where is it nothing. I would also assume that some might have many
> hundreds of axis stops.
>
> Interesting problem. I would be interested in any empirical data.
>
> Dave
>