Browse Prior Art Database

Closed Loop Acceleration Control Algorithm for Disk or Tape Drives

IP.com Disclosure Number: IPCOM000108071D
Original Publication Date: 1992-Apr-01
Included in the Prior Art Database: 2005-Mar-22
Document File: 4 page(s) / 135K

Publishing Venue

IBM

Related People

Bahr, AA: AUTHOR [+2]

Abstract

This algorithm enables a servo motor control loop to directly measure and control load acceleration using an incremental position encoder and a time counter. It can be optimized for use by a Digital Signal Processor (DSP), in that an acceleration measurement can be performed at a cost of only one long-division operation, the remaining arithmetic operations being add, subtract, and multiply which are performed quickly.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Closed Loop Acceleration Control Algorithm for Disk or Tape Drives

       This algorithm enables a servo motor control loop to
directly measure and control load acceleration using an incremental
position encoder and a time counter.  It can be optimized for use by
a Digital Signal Processor (DSP), in that an acceleration measurement
can be performed at a cost of only one long-division operation, the
remaining arithmetic operations being add, subtract, and multiply
which are performed quickly.

      Existing tape and disk drives that use incremental encoders as
position and velocity sensors still perform acceleration as a blind
operation.  Acceleration is performed in open-loop mode, where the
control system drives the servo amplifier with a step or ramp
function, waits a preset time interval, then attempts to close its
velocity control loop when the load is predicted to have reached the
correct speed.

      This article demonstrates an affordable method of maintaining
closed-loop control of the load during acceleration or deceleration,
especially if the control system is implemented in a DSP, with single
clock-cycle multiplication.  This article now describes this method
being applied to a tape drive which is instrumented with a rotary
encoder, which has lines spaced D radians apart.

      A plot of load position versus time during acceleration is a
parabola (Fig. 1) of the form:
Y = AT2 + BT + C

      In equation 1, T is time, A is 1/2 the load acceleration in
rad/ second-squared, B is the initial load velocity in rad/sec, C is
initial load position in radians, and Y is the load displacement
which is also in radians.

      Acceleration data may be collected as a sequence of the paired
numbers, as shown in Fig. 1 and Table 1.  In Table 1, the integers 0,
1, 2, 3 . . . represent an index count of sequential encoder pulses.
These paired numbers (Y, T) are the displacement Y of the encoder and
the total elapsed time T (seconds) at each encoder pulse.  The
quantity D in Fig. 1 and Table 1 is the incremental radian angle
between successive lines of the encoder.

                            (Image Omitted)

      The objective now is to solve for the coefficients of equation
1 that will fit the three most recent acceleration data points of
Table 1.  Fitting the parabola (equation 1) through these data points
yields:
Y(0) = C
   D = AT(1)2 + BT(1)
  2D = AT(2)2 + BT(2)

      From this information, a system of two equations and two
unknowns can be written:

      Solving equation 5 with Cramer's Rule, we find unknown A via
equation 6 and unknown B via equ...