The HEPI Pump Servo has been running on one of the pump stations at LHO for the last week controlling the motor speed to maintain pressure for the HEPI actuator fluid. The servo box contains a single board computer running CentOS for control via EPICS. We cabled everything up, booted and started the IOC with no issues.

Initially, voltages to each adc channel input were stepped from 0 until the adc counts were maxed at 32767cts. The specifications from the manufacturer of the pressure sensors (Honeywell) gives a calibration of 3000PSI/100mV for the readout. With a +10V input, the adc maxed out at 32768cts. With a sensor measuring attempting to measure 80PSI, this gives:

(80PSI) x (.1V/3000PSI) = .00267V x (32768cts/10000 V) =~ 9cts for a pressure of 80PSI.

As it turned out, their was not enough gain on the inputs to readout the output voltage above the adc noise, even after adding a “SMOO” field to the database file. Richard McCarthy added a gain of 300 to the pressure input channels so that now a voltage of 0.034V will max the adc at 32768cts.

Next, it was noticed the sample rate was affecting the CPU load when all sensors were attached, so the EPICS database file was modified to sample at 2Hz as opposed to the initial 10Hz.

The motor controller box was modified to allow remote control and voltage input.

The EPICS database file was modified to read out PSI units using a slope calculated from stepping through the range of voltages to adc counts and the manufacturer's calibration. The output voltage was stepped in counts via medm and recorded. Maxing out at 1000cts outputs ~5.11V for a PSI of ~100. The PID servo is now maintaining the PSI at ~60PSI without deviations larger than 1PSI.

The next step is to implement a way of writing EPICS data and monitoring the PSI levels over several-day periods.