Here's the latest version:
Hot set The old ones used a battery of cells totalling 630 volts. The new ones use an 18 volt rechargable gell cell battery (three 6-volt in series) and a transistorized voltage multiplier, supplying 630 volts at the test leads.
While not designed to weld PIC, as Ken has said, in practice we used them on PIC as well as paper cables. Welding is only one of several functions that they can perform.
In the case of an already solid short circuit, the trace tone can be used to pinpoint the short, using the mouse (induction coil or probe).
In the case of a crackling corroded open, usually caused by a Picabond connector failing, the "strap and zap" method can be used to weld the open together at least for the short term. In this method, the end of the pair is intentionally shorted, and at some other point the breakdown voltage is applied across the pair. One of two results will occur. Either the corroded open will weld itself, and the trouble ticket can be closed out, or it will blow open, allowing the fault to be found using the open meter.
Getting back to its other uses, the trace tone, which will only be transmitted into a solid short circuit, can be used to identify a particular buried service wire (BSW) in a bunch of BSW's that have been spliced into a main buried cable.
The convention for houses in a buried development where I worked was to splice the first three pairs of a BSW into the distribution cable, allowing three dial tones to be energized at every house in a development. The 4th pair was left open in the splice, and the 5th pair was left shorted in the splice. The open pair could be measured using an open meter (accurate to within a foot) so that the route and length of a BSW could be determined easily. The shorted pair would allow the hotset tone, applied at the house, to be traceable at the buried splice, so that the splicer would know which BSW was the one going to the house in question, without having to tear apart an entire splice.
The hotset's ohmeter is accurate down to one ohm. Using the known feet per ohm value of various gauges of copper wire, a very accurate estimate of a fault can be determined.
For instance, if the meter reads 4 ohms, and the wire gauge is known to be 22 gauge, which has a foot per ohm value of 62, the length of the loop is 248 feet (the entire loop resistance) which is then divided by 2 for a distance of 124 feet to the short circuit.
With experience, accurate plats (maps) of the cable plant, and experience, a good splicer can tell the gauge of the wire, given the resistance, or the distance to the fault, given the resistance. Did I mention experience? That helps too.