intelligent vs dumb drives - was ac vs dc servo motors


Thread Starter

Davis Gentry

Tom Bullock wrote:
> Power blocks are available from several vendors
where they supply only the transistors and heat sinks.
> All logic, loop closures, and compensations are done
within the main controller. On the surface, this
> looks less expensive, but the additional application
effort will more than offset the lower hardware
costs. <

That would depend on the level of expertise of the integrator with the individual controller. An experienced integrator using our controller and a PWM power block type amplifier will potentially have lower integration costs than the same integrator using our controller and an expensive intelligent drive, as he may have to do loop tuning and configuration in both the drive and the controller. The other option in this situation is often to use configuration parameters in the controller which effectively turn it into a dumb drive. In which case, why pay for the expensive intelligence?

The intelligent drives are in general not capable of well coordinated motion unless you are setting up simple systems such as low speed conveyors. You therefore usually have to put an intelligent controller in front of them. Once you have paid for that intelligence, why pay for it twice?

Davis Gentry
Applications Engineer
Delta Tau Data Systems


Thomas B. Bullock

You make some good points, but the market does not seem to be jumping on power blocks. Most "main controllers" do not have the motion capabilities or expertise of a Deltas Tau controller. That is part of the reason.

It appears to me that "main controllers" are the communications gateway for the ERP host, the peer to peer networks and the I/O networks. In addition, they serve as the PLC, the data acquisition system, the HMI inteface and the servo coordination/interface. It does the coordination of the axes and transmits to the drive either the command for each axis or the positon loop error for each axis depending on whether the position servo loop is closed in the main controller or the drive.
The intelligent drive closes both the velocity and torque loops. It does the auto tuning for the loops, performs machine diagnostics/signature analyses, and returns information to the controller such as position, velocity torque, friction, work energy, etc.
Perhaps what I am describing is what you might call distributed control and what you are describing is an alternative to that with fewer bells and whistles.
With computer chip prices falling as they have, having several locations in a system where computation is being done is not a major concern. Putting the computer capability close to the experts who are writing the software will pay off in the long run. The "drive" people know a lot more about the motor and machine than the "main controller" people who are trending toward data processing expertise.
Davis has given only a partial answer. A better solution would be to put most of the intelligence down at the drive. I.e. the velocity, position, current loops as well as commutation. Furthermore interpolation is also possible down at the drive. Then the control interfaces with the drives via a deterministic communication protocol (to send explicit commands). This then frees up the controller to perform more and higher level functions. Such as one controller controlling upto 32 drives. Or being able to resolve absolute feedback into the millions of counts per revolution. Now this solution may not be cost effective with 1-4 drives perfoming simple motion, but if you need precise synchonization of 5 or more drives, then it is, because you need only on controller. Plus the wiring of the systems for the integrator is far less than a conventional digital system.

Jim Watson

Davis Gentry

Our Turbo controller CAN control up to 32 drives with a single controller
card - either PWM or Sin Commutation or Analog out or Steppers. Or any
mix thereof. And also do all of the high level calculations and
coordinate all motion in real time. Using an interpolator card you can
also get millions of counts per rev. Or you can add an aalog feedback
from a load cell and get force around position control. Or other. Or add
an SSI encoder.

Davis Gentry
Delta Tau Data Systems