After I completed a successful
feasibility study that demonstrated the efficacy of a novel idea
for an industrial valve position sensor design, I was hired to
develop the sensor product.
First, my physicist colleague, whose idea spawned the project,
developed an extremely useful theoretical model of the magnetic and
data measurement aspects of our proposed sensor system. It handled
two types of dual-sensor orientations and arbitrary dimensions,
specified by the user, for cylindrical magnets. We used this model
to determine the requirements for the magnets, the distances between
the magnets and the sensors, and the gain of the data acquisition
system. We also studied the effect of quantization error from
different A/D converter resolutions, as well as the effect of random
noise. The results of this theoretical modeling study taught us how
to build a real test fixture for our sensor system.
We then developed an automated fixture that used a stepping motor
to rotate a small shaft with two magnets embedded in the bottom. We
designed and built a prototype sensor configuration with signal
conditioning and amplification so we could capture the sensor
outputs with an A/D board in a PC. We used VisualBasic to control
the operation of the motor, the sampling of the sensor voltages, the
sensor calculations, and the display of the data.
Based on our initial encouraging results, I hired a mechanical
engineer to design and build a permanent, mechanically stable
fixture. After his superb job, we had an extremely useful test stand
that could quickly and automatically scan more than ninety degrees
of shaft rotation, collect data, calibrate the sensor system,
operate the sensor system, and display data in both graphical and
numerical formats. This test stand allowed us to exactly determine
the size and types of the magnets, the optimal geometry of the
sensors, and the mechanical design requirements for materials and
dimensions in the final valve product. I tested our final sensor
prototype in my client's environmental chamber and the accuracy and
repeatability of the system in the temperature range between -40șC
and +150șC met our expectations.
I still had more to do on this project. I designed and tested the
analog front-end, signal conditioning circuitry for the production
units, and I developed a very thin, circular, two-layer printed
circuit board that could be installed into a valve's end cap. This
board held two surface mount MR sensor chips, one on one side and
the other on the flip side. I quickly learned in testing, however,
that this approach did not work. The lower magnetic field levels at
the sensor chip on the far side of the board caused excessive
hysteresis errors in the sensor system. To solve the problem, I
researched hybrid circuit manufacturers and hired one to build my
sensor board with two sensor chips in die form on the same side of
the board. This approach solved the problem and we had an acceptable
solution to the manufacturing problem. I delivered all of my
documentation to my client, and they proceeded with the
manufacturing preparation.
As my client neared production, I modified the test stand
software to function as a production test fixture. My client
designed the mechanical fixture to connect a stepper motor to their
production sensor shaft, and I changed the software to communicate
with the sensor using a RS232 interface. This modified stand
calibrated the production unit and verified its operation.
Finally, I worked with my client to integrate the sensor's
algorithms into their sensor and controller product's C code. I was
able to minimize the run time of the code by using very efficient
straight-line code in critical areas and a simple table lookup to
determine the result.
A patent, entitled Position Detection for Rotary Control
Valves, Number 6,244,296, was issued on June 12, 2001. An
amended patent, Number 6,484,751, was issued on November 26, 2002.
During this project, my client ran into a hardware development
problem with another vendor. They needed a HART modem on a small
printed circuit board that would attach to the main board in their
valve position sensor and controller package. They asked me to look
into the problem, and I eventually developed the board for them. If
you would like to read more about that project, click
HART Modem.
