Secondary random errors affecting sensor accuracy are caused by friction, degree of 

 resolution, potentiometer wiper noise, hysteresis, shock and vibration. The errors due to 

 friction are greatly reduced due to the ever-present vibrations associated with buoy towing. 



The potentiometer sensor accuracies shown in Table 1 represent the cumulative system- 

 atic and random errors as specified for each type of sensor by the manufacturer. Empirical 

 calibration data taken at DTNSRDC show a combined error of less than 0.5 percent of full 

 scale when a vibrator is used to reduce static friction and air temperature is held constant to 

 within plus or minus 1 -degree Fahrenheit, during the calibration. 



The cable tension sensor used aboard the submarine consists of a custom made strain- 

 gaged flexure designed as an integral part of the winching system and for this reason calibra- 

 tion of this sensor must be performed aboard the submarine. The strain-gaged flexure is 

 mounted between the winch bearing housing and the winch base so that a tension applied to 

 the towcable must be transferred to the winch drum and the winch bearing housing before 

 being sensed by the flexure. There is a large cable sheave external to the winch over which 

 the towcable must travel during the winching operation. A calibration is performed by 

 placing a calibrated Baldwin-Lima-Hamilton load cell in line with the towcable outboard of 

 the sheave. The cable is incrementally tensioned using a ratchet hoist. The accuracy of the 

 tension sensor was determined to be approximately 5 percent of full scale based on the 

 empirical data of several calibrations, which include the errors due to the static friction of the 

 winch bearings and sheave. 



During the buoy evaluation in the towing basin at DTNSRDC, a tension sensor accuracy 

 of 0.5 percent of full scale was realized using a Baldwin-Lima-Hamilton load cell. Towing 

 carriage speed is measured to an accuracy of better than 0.01 knot. The submarine speed 

 sensor, an electromagnetic log, is part of the ship's equipment and its accuracy is unknown. 



The calibration network is made up of resistors which are subject to variation as a func- 

 tion of temperature. A variation of temperature would affect an error in output data only if 

 each resistor value did not change proportionately with the remaining resistors. The problem 

 is minimized by using resistors where each is made of the same material and specifying temper 

 ature compensated resistors which vary only 50 parts per million per degree Centigrade. 

 Another component of the calibration circuitry includes a one-pole four-position switch for 

 each sensor. The contact resistance of these switches is specified to be no greater than 0.02 

 ohm; this resistance is in series with an amplifier input impedance of 10,000 ohms and may 

 be neglected. The overall worst case error introduced by the calibration circuitry is calculated 

 to be less than 0.1 percent of full scale for a temperature variation of 20-degree Fahrenheit. 



The telemetry electronics, including the transmitting and receiving unit, have a combined 

 error of 0.3 percent of full scale attributable to nonlinearity. Variations due to temperature 

 will be shown to have no affect on final data accuracy. 



21 



