IBM 727, Manual Of Instruction

Page: 44 / 163

Specifically, it is necessary to minimize skew so that the closely spaced characters
on tape can be correctly interpreted and distinguished from each other. Figure 16
illustrates three of the causes of skew in an exaggérated manner. The effects of skew
are magnified when the skew written into a tape by one tape drive is opposite to the
skew associated with reading on another tape drive. The factors involved in the skew
problem are:
1. Head skew. This is inherent in the manufacture of the head and cannot be ad­
justed in the field. It is derived from the fact that mechanically and magnetically the
individual tracks (pole tips) are not in perfect alignment nor are they completely iden­
tical in shape.
2. E lectrical skew. The wave form s produced by different tracks in conjunction
with their associated circuitry, e. g. in reading identical tape signals, are not exactly
alike. (See item 4 b elow .)
3. Angular alignment of the head and tape. This can be adjusted in the field
(section 1 5.2 .6) and is of great importance. While the average alignment is subject
to adjustment, the rapid fluctuations in alignment called tape "flutter" are difficult to
rem ove completely. The angular alignment of tape, as well as the lateral position of
tape (tape tracking), is determined by the accurate location of the vacuum columns,
split idlers and the read-w rite head (guide posts). This is explained further in se c­
tion 15. 2. 5.
4. Signal amplitude— tracking. Two tracks have an effective skew between them
if one signal is weaker than the other even though their wave form s are sim ilar and
peak at the same time. This is true because the stronger signal r 2 .5 volts
on the leading edge of each pulse before the weaker signal does, and this is the time
that the corresponding read register trigger is set by the final am plifier. Tracking,
in a strict sense, is the lateral agreement of the individual tracks of the read write
head with the signal tracks on tape. This depends on the lateral position of the tape
with respect to the read-w rite head and also the lateral position of each individual track
with respect to the reference edge. The latter is held to close tolerances at the fa c­
tory. Inasmuch as signal amplitude is dependent on tracking, skew is also dependent
on tracking to some extent. For this reason, the pre-am ps are tem porarily adjusted,
if necessary, to obtain 20v peak-to-peak output from the master tape when checking
the skew adjustment.
5. Speed of tape. To distinguish skewed successive characters coming from
tape, it is necessary to use a character gate in the control unit. The character gate
is turned on by the first bit in the character, and turned off after a fixed interval
such as 33. 6 usee. Speed, therefore, is a factor in determining the effective skew
because, when reading, the slower the tape moves the greater is the time interval
between bits as a result of the geom etric skew already present. Furthermore, in
writing, the speed determines the distance between characters. While the average
speed of the tape will probably not deviate far from the prescribed 75 inches/second
(unless the belt is slipping, or there is an appreciable bind), the fluctuations of speed
called "jitter" add to the skew problem . Jitter is aggravated by loose and worn parts
in the moving capstan drive system 0
Skew can even be observed when reading forward a tape that has just been written
on the same head. This occu rs with no apparent change in tape alignment and with
all signal amplitudes at 20v peak-to-peak. This can be attributed to: (1) head skew,
and (2) electrical skew» A probable contributing factor is that the location of the bit
on tape as it is written forward is determined by the right-hand pole tip (effective),
while the leading edge of the signal pulse in reading forward is determined by the
44