Re: [CAD_CAM_EDM_DRO] SLA7062 Sync Mode (was New Article bla bla)

Phil Mattison wrote:

> The other Dave's analysis assumes that stepping errors will be cumulative
> when microstepping with the SLA7062 and switching to sync mode when paused.
> I suspect that assumption is incorrect, for the following reasons. For one
> thing, the SLA7062 data sheet does not say WHY it is not recommended. I
> would bet the reason they say that is because they must assume that if you
> are microstepping you need the extra position accuracy when holding.

I've always heard that microstepping (anything higher than
half-stepping) does not improve positional accuracy, it only serves to
reduce mid-band resonance by smoothing out the step impulses. The
holding torque at microstep positions is very low, and the actual
position is not linear, so even in very lightly loaded conditions it is
necessary to carefully profile each stepper to map the position of the
motor at microstep positions.

If the board is running in microstep mode and the machine is not aware
of this fact, then it is likely to stop and attempt to hold at weak
microstep positions. When this happens the motor is likely to slip
forward or back to a full- or half-step position. When the machine
starts again it may lose steps depending on where the motor syncs back up.

This behavior doesn't have anything to do with sync mode, its just a
result of the step generator not knowing where it is appropriate to stop.

> It seems very unlikely to me that
> entering sync mode would get the step logic out of phase. More likely is
> that the PWM current regulator is not accurate enough to hold the motor in a
> position beyond a half step precision. Keeping track of steps is a logic
> function, and is (or at least should be) unaffected by the PWM operation.
>

Agreed. It would be interesting to see some more analysis.

I considered doing some experiments with a pulse multiplier that would
sit between my HCNC board and computer such that I could run the 7062 in
8x step mode and set the computer up so that it would only ever stop at
half- or full-step positions. Never got 'round to it though, as I never
had a problem with lost steps, at least, not so's you'd notice.

On my next machine I plan to put encoders on the steppers and run it all
with a GREX with the 'unstallable stepper' feature, so it won't be an
issue anyway :) I don't really need that kind of performance, but since
it's a hobby the 'coolness' factor weighs fairly heavily in the decision :)

> An easy way to put this question to rest would be to setup a mill with a
> dial gage against a block, zero it, and generate series of moves and stops
> in one direction with random distances, and then the sum of those moves at
> once (taking into account backlash) in the other.

If anyone performs such a test I think it would be important to plan the
moves such that the machine would stop at a microstep position that is
nearest to a full step position, maximizing the likely-hood that the
motor will slip to the full step position. The system should then
reverse direction so that the driver steps the logical rotor position
away from the actual position, resulting in the rotor having to jump
forward to 'catch up'. Of course the farthest it would have to catch up
would be a half step, which it should be able to handle, but I suppose
under some conditions it might not be able to, resulting in a lost full
step (?).

I would give it a try, but alas, I am currently without machine, having
sold my Taig system to raise funds for a new house (with more space for
machines :) ).

Dave K