Re: [CAD_CAM_EDM_DRO] Safety -- servo vs. stepper

Carl Mikkelsen wrote:

>I have a choice of using stepper drive or servo drive for my next
>hexapod. My concern is how to cope with safety issues.
>
>My previous hexapod is stepper driven, with fairly slow maximum
>speed of just less than 1 inch per second. My next hexapod will be
>faster, with a designed intention of fast movements of about 3-5
>inches per second.
>
>With my hexapod design, it is very difficult to have limit switches
>on all axes. In software, it is easy to limit motions in various
>ways, such as to prevent any joint angles or actuator lengths from
>being exceeded. It is not easy to place a fixed limit switch on an
>actuator to serve as an E-stop of last resort.
>
>Under hardware fault conditions, it seems that steppers are safer.
>In particular, most failures in the high-power drive system will not
>result in actuator motion. With a servo system, it seems that an H-
>bridge failure is likely to result in uncontrolled movement.
>
>Has anyone experience with drive failures, and the reprocussions?
>

Sure. You really NEED direct encoder feedback to the computer, and a
hardware E-stop circuit that removes power from the servo amps. A faulted
transistor or two in a Gecko or Rutex drive, and it will run away. Lighting
the fault LED will not stop the motor.

> I
>was about to order servomotors, and my stomach went cold as I
>imagined the danger of an sudden, RAPID, uncontrolled movement of a
>50 pound mass moving in 3D.
>
>

A single faulted axis on a rigid hexapod will not cause movement, it will
blow fuses. That is not true on a cable hexapod, of course.

I have been using servos on a cartesian machine since 1998. I built my
own servo amps, and have had a few runaways. In general, the machine only
gets about 1/4" before the amp trips offline or the computer commands an
E-stop. In my case, I have two forms of feedback - DC tach and encoder.
The tach feeds back to the servo amp, and the encoder feeds back to the
computer. If either fails, the system is still marginally controlled.
In the
case of an encoder failure, the computer has no idea that motion is
occurring, it thinks that motion is NOT happenning, and commands more
velocity from the servo amp. This will usually trip the servo amp
offline in a very short time, maybe 50 mS or less. You can verify this
response with just the motor and no hexapod attached.

So, in about 6 years, I have NEVER had a true runaway with the machine
screaming toward the limits. I have had my share of loose wires, dirty
connector contacts and similar stuff, and a number of e-stop events when
the machine was not responding correctly, but never a servo runaway.

But, you HAVE to make sure your E-stop system really stops motion,
no ifs ands or buts! And, that means some effective way to remove
power from the servo amps and connect a braking resistor.

Jon