When you go to a hospital or clinic, you expect everything to be not just clean, but spotless. Therefore, you might think we use tap water in the hydraulic system of treadmills that operate in magnetic resonance imaging (MRI) facilities because tap water is clean, so any incidental spills or leaks would be of little consequence. That’s true; when it comes to environmental friendliness, it’s hard to beat plain old water. But that’s not why we use it.

We use fluid power because MRIs in clinical use generate powerful magnetic fields — typically 1.5 to 3 Tesla. Consequently, anything attracted by magnetic fields could be pulled toward the MRI’s magnets. Furthermore, anything that can produce magnetic fields — such as electrical current flow — can interfere with the MRI itself. Therefore, the treadmill cannot contain any ferromagnetic materials or electrical components and must use non-conductive hoses and other components. 

Obviously, we couldn’t use electric motors or solenoids in the treadmill. That left us with a choice between hydraulics or pneumatics. Pneumatics was briefly considered, but issues with compliance (compressibility) and noise seemed too significant to overcome within application parameters. Therefore, hydraulics became the obvious choice. But hydraulic motors are usually made of ferrous metals, so those were ruled out. We then focused on stainless steel hydraulic motors and found that those that fit our application parameters were designed to use water as the hydraulic fluid.

Hydraulic drive
Typical exercise treadmills are powered by a variable-speed electric motor for the belt drive and a second electric motor for elevation control. Standardized multistage cardiac stress tests typically call for treadmill belt speeds starting at about 1.7 mph, with increases every 3 min to a maximum speed of 6 mph. Elevation begins at a 10% grade and eventually increases to a maximum of 22%.

The motor’s operating speed ranges from about 500 rpm to more than 3000 rpm. As it turns out, torque requirements for this application are highest at the lowest speed and decrease steadily as speed and elevation increase.

This is a perfect application for a variable-displacement motor. However, we could not find any in the size range and the specific materials required for the MRI-compatible treadmill. The elevation actuation system for the treadmill was straightforward to design and source because stainless steel cylinders are readily available. However, finding a source for a stainless steel hydraulic motor to drive the treadmill belt was more difficult, and we could not find any with a variable-speed option. We chose a fixed-displacement, axial-piston motor constructed of stainless steel and high performance polymers. Supplied by Water Hydraulics Co., East Yorkshire, England, the motor is designed to operate with tap water.

The hydraulic system’s power unit is located in an equipment room adjacent to the MRI room, shielded from the MRI critical environment. Choosing a prime mover and pump was the next challenge. Our fluid limited the pump choices to materials and components tuned to water, even though the equipment room has no limitations on ferromagnetic materials. A constant-speed electric motor coupled to a variable-displacement pump was considered, but again, we could not find a variable displacement pump scaled for our application.

We selected a fixed-displacement axial-piston pump similar in construction to the treadmill’s hydraulic motor. Because the pump and motor have fixed displacement, this dictated a variable-speed prime mover with high torque capability at low speed and low torque demands at high speed. A 3-phase ac motor — rated at 3 kW — was selected. The speed of the motor is regulated by a variable-frequency vector control.

Variable-speed controls
Radio frequency (RF) interference limitations were established because MRI devices depend on a highly shielded RF environment. Control components located within the MRI room itself must be completely free of electrical components and free of electrical conductors between the MRI room and the adjacent power unit. However, the treadmill control system must be linked to other diagnostic software. System status monitors are required to display in the MRI room and log system performance data.

Therefore, we chose fiber optic devices to meet the needs of speed feedback, position sensing, and emergency stop control. The fiber optic cables selected are non-conductive, and all electrical devices are located in the power unit, isolated from the critical environment. Optical feedback from the treadmill is converted to electric signals and communicated to the motor controller and data acquisition (DAC) board.

The hydraulic pump and motor are plumbed for open-circuit, unidirectional operation. Therefore, except for the classic variations in the pump and motor’s volumetric efficiency, belt speed is proportional to the prime mover speed. The treadmill’s belt speed control proved the most challenging control issue. We measure hydraulic motor output speed using an optical rotary encoder, which provides a fiber optic feedback signal that becomes an analog input to the electric motor drive. The variable-frequency vector control is programmed to compensate for changes in load and volumetric efficiency of the pump and motor.

The treadmill’s incline is achieved by controlling fluid flow to an elevation cylinder. The cylinder was designed and manufactured by Lehigh Fluid Power Inc., Lambertville, N. J. Made of 316 stainless steel, it has a 1½-in. bore and 4½-in. stroke. Fluid is supplied to the cylinder by an accumulator, which stores enough pressurized fluid to raise the front of the deck to its maximum incline. The accumulator is precharged with nitrogen and is pressurized to maximum system pressure at the beginning of each stress test by diverting pump output to the accumulator circuit.

Cylinder displacement is controlled by two, 2-way directional control valves (DCVs) and fiber-optic position feedback. The DCVs are not proportional, but they are subjected to pulsed input signals when the error signal is small. This scheme was developed to reduce the overshoot encountered in early testing.