Servo Following Error and Mechanical Backlash

Everyone agrees that reducing the cycle time greatly impacts the cost of production and the bottom line; time spent machining a part is the most significant contributor to part cycle time. However, simply having a high speed spindle does not result in automatic cycle time reduction. You can program faster feeds, but typically with most CNC equipment faster feed rates degrade the accuracy of the part.

The limitation of the servos directly impacts the ability of the slides to move accurately and is directly proportional to feed rate. This limitation causes the tool to undercut or overcut the programmed path, requiring slower feed rates to accurately machine the part. Adding acceleration / deceleration ramps only improves the starting and stopping, not the servo error in-between (following error).

The illustration shows one of the common shape errors that occurs with Servo Following Error. In this case, we are showing the error during servo direction reversal in the Y axis. This Flat Error is directly related to two characteristics: Servo Following Error and Mechanical Backlash. While Mechanical Backlash can be improved mechanically or by software compensation, the problem of Servo Following Error dynamically changes according to feed rates.
The Flat occurs when the Y-axis reverses direction. This is because when moving Y positive, the Following Error is a negative error, and when reversing to moving Y negative, the Following Error becomes positive.
The transition from negative to positive (Following Error) creates a servo dead-zone. The size of this dead-zone is directly proportionate to speed and is very difficult, if not impossible, to accurately correct with software, especially as feed rates increase.
With the typical machine today, the only solution operators have is to slow the feed rates to reduce the Following Error, which directly increases the part cycle time.

Performance Test
Our approach has been to design a high performance servo system that virtually eliminates the Following Error from the system. To illustrate how our performance improvements can directly reduce the part cycle time, we have three videos showing a standard Circle-Diamond-Square part being machined at various speeds with the part inspection results.
Test Specifications

1997 Fadal 4020 with the NXGEN Control.

  • Baldor/Glentek axis and spindle motors
  • Glentek amplifiers
  • Baldor H2 spindle drive
  • Ballbar test showed minor squareness error
  • .500″ diameter carbide
  • 4 flute endmill
  • RPM: 10,000
  • Material: 6060-T6 Aluminum
  • Test Features:
    OD Program Size 1.8000″
    ID Program Size 1.0728″
    Diamond Size 1.2728″
Performance Test F40
Performance Test F94
Performance Test F150
F40 Inches / Minute
F94 Inches / Minute
F150 Inches / Minute
Cycle Time 72.7 Seconds 30.9 Seconds 19.4 Seconds
Feed Rate 40 Inches/min 94 Inches/min 150 Inches/min
OD Actual Size -.0004″  -.0003″ -.0001″
OD Roundness  .0002″ .0002″ ±.0003″
ID Actual Size -.0003″ -.0005″ -.0006″
ID Roundness .0002″ .0005″ ±.0004″
Diamond Actual Size  -.0001″ -.0001″ -.0002″
Diamond Squareness ± .0003″ ± .0003″ ± .0003″
Crossover No measurable value No measurable value XY .00025″ error
Proof is in the numbers
As shown by this test, the Control dramatically reduces servo following error. No longer do you need to settle for a servo error of .005″ or greater. When compared to the Fadal Legacy control system, the NXGEN Control testing results shows it to be 40x more responsive.
Test your machine
Click below to download the program we used in these tests.  Do your own tests and see how much you could be saving by reducing your cycle times.