Sequential PROPHET-600 - Page 77

^ , ;j i. ^ a ^ a
I * (IV M
t a
g' B' ? c ? ■ ■ g ?
'2' <!
4r ■ *
Ji:
a a o o —
(laiMl
-• b ^ 1 - j i. .
the*' feci tack to th» control
input pin As the fu-quency is
increased. this feed lurk voltage
will tend to sharpen the control
scale, but only at the uppe*
end. since the feedback voltage
becomes significant only at
the hghcr generator currents
The amount of loltage ‘ed
back is adjusted so the seale
is sharpered iusi enough to
compensate tor the inherent
high end flatness
The method recommended
foi trimming the control scale is
as follows The tv frequency
track ad|ust is fust set so that
rso correction voltage >s fed to
the control input. The oscilla
tor fieouency is set around
200 H/ and the scale adjust
l'immp is adjusted for the
desi'ed scale factor le.o t .000
octave/voltl. Th*n the oscillator
frequency is set to around
lOKHe. and the hi frequency
track trimmer is adjusted for the
same scale factor.
The source of error from the
precision mullipl«*r is due to the
multiplier's gain (normally
unity) changing as the control
input current changes This type
Of error causes thg frequency to
become increasingly sharp or flat
as the control Current is in¬
creased The percentage differ
ence in multiplier gams, and
hence scale factors, at any tsvo
inputs to the multiplier may he
calculated as the percentage
error given in the specifications
times the difference injjA
between the two corresponding
outputs Tor • ■ ample, suppose
tt-e scale were adjusted for
precisely I octave/volt at mid
range. At one octave above this
adjusted oeta.e. with the mu I
tiphcr output lOsiAd ffe*ent.
the scale factor could be 08 %
different worst case This
would produce a voUs/octave
error at the base of Qi ol Off .
» I8mv - 14.4yV, which would
cause this octave to be . 06 '.
II cent) sharp or flat At five
octaves above the adiustnd
octave, the scale factor could be
0 4" different worst case, pro
duemg « volts.octave e*rot o'
0.4% ■ 18mV - 72uV. Thtj fifth
octave would thus be 0.28 =
IS centsl sharp o' flat Note that
if octaves above the adjured
octaves were sharp, those
octaves below the adjusted
octave would become increas¬
ingly flat, and vice versa
Typically the error produced
by the multiplier is much less
than the above example. How¬
ever. ii maintaining a tighter
tolerance is required !o< the
particular application, the mul¬
tiplier error may be trimmed out
'O' each device. The trimming
proceduie requires that both H?
and B$ be made adjustable
*30% about the nominal value;
f>7 is first adjusted so that the
multiplies gam is constant over
the selected mpul current
range, then Rj is adjusted fos
the desired scale factor (adjust¬
ing R$ will reintroduce some
orroi. so R; may have to be
readjusted).
Should for tomo reason it bo
dented not to use the tempera
ture compensation circuitry, the
multlptier/tcmpco generator
may he bypassed simply by
leaving pin 1 . pm 2 , and the
control input pin open, and
applying the control voltaoe to
the base of Qj via the multiplier
output pin.
All waveform outputs are short-
c-rcu't protected and may be
shorted continuously to any
supply without damar^ng the
device. Each output, however,
has d 'fcrmg drive capabilities
Although the tr-angle output
can sink at least 4mA and
source over sevcrol mA. cam
must be exercised in loading
this output. Because the output
has a linitr impadance and
drives the comparator, a Change
in load will change Use frequency
of try- oscillator. Adding a I0OK
resistor to ground, for instance,
could lower the frequency by
150 100K -0.15% (2.5 centsl
worst case. A load capacitance
will act like a resistor with a
value 1/(2IC L ) and ••quires the
same considerations as above
A continuous load no greater
than 10K and/or lOOOp' to
ground is recommended
Since the sawtooth output
is buffer isolated from the
oscillator circuitry, if can sink
at least 6mA and source over
several mA with no effect on
oscillator performance, and only
negligible effect on sawtooth
waveshape Stray capacitance at
this output greater than 40pE.
however, will cause a small high
Irirouency oscillation. A 1000
resistcv between the output and
load is all that is reauired to
isolate more than 01 u r
S4V IMC
0»r T
“I
T
■ Him i aiMiMin
Waveform Outputs