LIBRARY ı6ıul 0) - Cranfield University

the collection of welding data corresponding to a minimum of 10 short circuits in
stable low4 wire feed speed dip transfer welding [refs. 51,161]. Before any statistical
feature extraction can start, all the points from all the analogue-to-digital converter
channels must be transferred from the internal memory of the acquisition boards to the
memory of the main computer. This takes approximately 15 milliseconds in the
hardware used. After the transfer, the collection of a new block of data should start,
while the main computer should proceed with the calculations. Considering the
acquisition time, the data transfer time and the time needed for calculations and data
output to the user interface and to the other controllers (table controller and welding
power source), a cycle of approximately 250 milliseconds was obtained. This was
conventionally called a monitoring cycle.
4.2.3 Stand-off estimation and control
The model proposed by Ogunbiyi [ref. 51] and outlined in equations 2.21 to
2.24 was initially considered for in-process estimation of the stand-off. However, the
model was found to be imprecise when low wire feed speeds were used. An analysis
shows that the imprecision was due to the difficulty in the correct estimation of the
value of the proportionality constant, O, used in equation 2.22 and also in the
determination of the initial welding current to be used in the same equation. However,
the model was found to be useful for qualitatively indicating a change in stand-off. For
control purposes an absolute quantitative measurement was necessary. In order to
obtain a more reliable stand-off estimation, a new model was developed using as its
input the measurement of the resistance between the contact tip and the workpiece. In
dip mode metal transfer, the resistance was calculated during the short circuiting
phase, when the wire could be considered as a continuous electrical conductor
between the contact tip and the weld pool. In the spray mode of metal transfer, the
calculated resistance was in fact the combination of the wire stick-out resistance with
the arc equivalent resistance. A resistance component present in both estimations was
the wire electrical contact resistance at the contact tip, which was believed to be a
source of noise. The development of the models for estimating the stand-off from the
calculated resistance is described in section 6.2. Equation (4.8) shows the stand-off
model in a general form.
SOS = f, (WFS, V)"R+f2(WFS, V) (4.8)
where
SOefr is the estimated value for the stand-off,
R is the resistance between the points where the voltage signal is picked-up and
f, , f2 are functions of the wire feed speed and welding voltage (see section 6.2).
Considering that the stand-off estimated values were obtained from the
welding current and voltage waveforms and that process instabilities would influence
such an estimation, the implemented controller would only estimate the stand-off if a
4 Lower limit of the range used =4 m/min.
110