Re: [CAD_CAM_EDM_DRO] Re: stepper vs. servo

Tony Jeffree wrote:

>At 22:43 10/02/2004, you wrote:
>
>
>>Les,
>>
>>Well then let's just confine ourselves to the second half of my
>>statement. Steppers, once tuned--and I disagree with your assertion
>>that "tuning" is required--need no further attention. (steppers DO
>>need proper "selection". But not tuning.)
>>
>>Now let's address your second assertion, in a fun way:
>>
>>No matter how carefully you set up SERVOS, one day you will hit a
>>particular set of circumstances that will make them lose steps. It is
>>most likely to happen if you are demonstrating the machine or just
>>finishing a very expensive and time consuming job..."
>>
>>But you won't know it! Only when the part is rejected because it
>>doesn't meet spec will you find that the following error ALWAYS
>>PRESENT with servos, has caused the actual toolpath to be other than
>>intended... Because the g320 and similar can be off by up to 128 steps
>>(encoder counts, really) before a fault is issued...
>>
>>If you're going to defend things by using worst case scenarios, let's
>>apply them across BOTH systems being considered...
>>
>>
>
>It is worth remembering that with a servo system, if the motor is moving,
>it is doing so *because* there is a discrepancy between the current
>position and the position that it is required to be in. So with a servo
>system, the only time you know that there is no positional error is when
>the motor is stopped. As you increase the load that the motor has to shift
>(e.g., by increasing depth of cut), the size of any positional error will
>tend to increase.
>
>

This is not actually true. With feedforward, the control laws are such that
the motors move when velocity is needed, and the velocity command is
modulated whenever the machine is not at the commanded position to
minimize error. This way, the error does not have to build up before
motion starts. EMC has the capability to graph actual vs. commanded
position, and you can adjust the servo tuning until the lines totally
overlap.

>In contrast, with a stepper-based system, as long as the system doesn't
>lose steps (i.e., as long as the motors are sized correctly for the applied
>load), there is no change in the accuracy of the system as the load is
>increased.
>
>

Well, within a full step, that is true. But, a full step can be a
significant amount
of error in some systems. If the error reaches two full steps, you lose
4 steps
suddenly, or the motor just stalls. But, if you are running at, say, 75
% of the
speed at which it stalls, that will likely be lagging about one full
step behind
commanded position. Also, you can't assume that any stepper drive is making
the motor follow the step pulses instantly. For instance, the Gecko 210
drive has a PLL step multiplier that adds some delay between steps in and
steps out. Many other microstepping drives have internal delays, especially
if they have PIC or similar microcontrollers in them. So, you can't know
that the motor steps instantly when the computer sends the step pulse,
unless
you have measured it.

Jon