Page images
PDF
EPUB
[blocks in formation]

left dwell. The Atmel microprocessor set bits in a status word when the oscillator reached programmable percentages of the cross seam distance. The host microprocessor interrogated the Atmel microprocessor 1000 times per second to detect these status word bit transitions. When

detected, the host scheduled the appropriate tracking task, which would normally read the analog to digital converter data or start the analog to digital converter.

[blocks in formation]

Figure 4 is a block diagram of the arc voltage and current measurement daughter board. Analog values from sense wires and shunts were sampled 40,000 times per second and digitized using Hewlett Packard HP7800 isolation amplifiers and a 12 bit Analog Devices LM12454 converter. An Analog Devices ADSP2181 Digital Signal Processor processed the converted values. The board sampled the data for the commanded length of time, and returned the processed values of the sampled data when queried. One type of processing performed was excluding sampled arc voltage readings from the average when sampled arc current was below peak levels.

BLOCK DIAGRAM OF HIGH SPEED
DATA ACQUISITION BOARD

[blocks in formation]

The host microprocessor compared the center samples to the target voltage, and used the error times a scaling factor to command the torch to work motor to raise or lower the torch. The computed move limited to a programmable maximum value per sweep. The computed move command was based solely on the proportional error, without integral or derivative terms.

Studies were carried out to determine the sampling window that would optimize the comparison of the voltage drops that occur as the torch and welding electrode approach each sidewall of the joint. To carry out this research, an eight-channel analog to digital converter board from National Instruments was purchased. Isolation amplifiers were designed and built to protect the analog to digital converter board from the welding arc voltage, and to allow arc voltage, arc current, digital timing signals, and analog processes signals to be measured simultaneously.

Timing signals were provided by the Atmel microprocessor on the brushless DC servo motor board that controlled the torch movement. Arc voltage was measured at the arc, and arc current was measured close to the arc by means of a clamp-on Hall Effect current probe. Processed analog values were obtained from a digital to analog converter connected to a serial output on the ADSP2181 digital signal processor. Waveforms were stored on disk, and were printed out on a laser printer.

Trials were also conducted to determine optimum attenuation factors and maximum move factors for vertical and horizontal tracking. Tracking was given time to stabilize and then a step change was made to either the cross seam or vertical adjust motor. The response time and overshoot of the cross seam and vertical adjust axis were measured by means of linear potentiometers attached for this purpose. The optimum values for attenuation and maximum correction per sweep were selected and used on the pipeline welding project.

Results

Figure 5 shows a typical waveform obtained using the national instruments data acquisition system. The top trace is the arc voltage, the second trace down is the arc current, the third trace down is the "late indicator", the bit set by the brushless DC servo motor board Atmel chip indicating the torch has reached a programmable percentage of its sweep. The bit goes low after the host CPU detects it has gone high and has sent a command back to the Atmel to clear the bit. The next trace down is the right wall indicator, another bit set by the Atmel when the torch has reached its extreme point of oscillation in one direction. The trace below the right wall indicator

[graphic][subsumed][ocr errors][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][ocr errors][subsumed][subsumed][subsumed][subsumed][subsumed][merged small][subsumed]

is the left wall indicator. The bottom trace is the analog value of the result of the processing done by the high speed data acquisition board. The data was taken at a rate of 180 oscillations per minute. The time period between the right and left wall timing signals was 333 milliseconds. The pulse frequency was about 135 Hz. The short segments of the bottom trace represent the value of the processed arc voltage signal measured from the end of one sidewall to the start of the opposite sidewall measurement. Because these represent the torch to work measurement, they are nearly equal in value. The longer segments of the bottom trace represent the value of the processed arc voltage measured while the torch approached the sidewall. The values are lower because the contact tip to work distance decreases as the torch approaches the side wall.

Figure 6 shows the effect of a step change to the cross seam position. The top trace is arc voltage, the second, third and fourth traces are the late indicator, right wall, and left wall indicators respectively. The second trace from the bottom is the processed arc voltage measurement for the previous period. The bottom trace is oscillation motion measured by a linear potentiometer. The center of oscillation was intentionally shifted 0.010 inches after the tracking had stabilized. The time for the oscillation pattern to return is approximately seconds, or 18 oscillations. Figure 8 shows the result of excessive response to the difference in sidewall measurements. The system overcorrected in an under-damped fashion.

[graphic][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][ocr errors][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][merged small][subsumed][merged small]

The 12 bit analog to digital converter provides a resolution of arc voltage of 0.024 volts per bit. The scale factor applied to the torch to work error was equivalent to 0.125 inch of movement per

[graphic][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][ocr errors][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][subsumed][ocr errors][subsumed][subsumed][subsumed][subsumed][merged small]

2 volt difference between measured arc voltage and target arc voltage. This scale factor was attenuated to achieve the damped response shown in figure 6. The under damped response shown in Figure 7 occurred when no attenuation was used.

Prior to introducing through the arc tracking on the project site, manually guided dual torch bugs were welding 90 joints per day with a 9% repair rate. After introducing through the arc tracking, the repair rate dropped to less than 4%. The most common weld defect that required repairs without through the arc tracking was lack of sidewall fusion in the second and third fill passes. The engineering critical assessment required welds with small stacked lack of sidewall fusion defects be repaired. The cause of this defect was excessive puddle fluidity caused by excessive weld metal dilution. The excessive weld metal dilution was in turn caused by setting oscillation width wide in an attempt to insure sidewall fusion regardless of the torch not being centered in the groove. Through the arc tracking allowed the use of a width that was just wide enough to touch both sidewalls by maintaining the torch oscillation pattern centered between the sidewalls.

SUMMARY

A through the arc seam tracking system has been developed and applied to a dual torch external welding bug used for cross country pipe line girth welding. The system was successfully used to weld 122 miles (196 Km) of 42 inch (1.06 m) diameter grade X-70 pipe. Use of through the arc seam tracking reduced weld repair rate from the 9% to less than 4%.

« PreviousContinue »