Datasheet

21
Yes. Some noise energy exists beyond the 30 kHz
bandwidth of the HCPL-788J. An external RC low
pass filter can be used to improve the signal to
noise ratio. For example, a 680 Ω, 4700 pF RC filter
will cut the rms output noise roughly by a factor
of 2. This filter reduces the -3dB signal bandwidth
only by about 10%. In applications needing only a
few kHz bandwidth even better noise performance
can be obtained. The noise spectral density is
roughly 400 nV/
√√
√√
√ Hz below 15 kHz (input referred).
As an example, a 2
κκ
κκ
κHz (680 Ω, 0.1 µF) RC low pass
filter reduces output noise to a typical value of
0.08 mVrms.
At present Avago does not have a standard
product with tighter gain tolerance. A 100 Ω
variable resistor divider can be used to adjust the
input voltage at pin 1, if needed.
Op-amps are used to drive V
OUT
(pin 12) and ABSVAL
(pin 13). These op-amps can swing nearly from rail
to rail when there is no load current. The internal
V
DD2
is about 100 mV below the external V
DD2
. In
addition, the pullup and pulldown output
transistors are not identical in capability. The net
result is that the output can typically swing to
within 20 mV of GND
2
and to within 150 mV of
V
DD2
. When V
REF
is tied to V
DD2
, the output cannot
reach V
REF
exactly. This limitation has no effect on
gain — only on maximum output voltage. The output
remains linear and accurate for all inputs between
-200 mV and +200 mV. For the maximum possible
swing range, separate V
REF
and V
DD2
voltages can be
used. Since 5.0 V is normally recommended for
V
DD2
, use of 4.5 V or 4.096 V references for V
REF
allow the outputs to swing all the way up to V
REF
(and down to typically 20 mV).
No. The LED is used only to transmit a digital
pattern. Gain is determined by a bandgap voltage
reference and the user-provided V
REF
. Avago has
accounted for LED degradation in the design of the
product to ensure long life.
Ideally gain would be V
REF
/512 mV, however, due to
internal settling characteristics, the average
effective value of the internal 256 mV reference
is 252 mV.
4.2: Can the signal to noise ratio be improved?
4.3: I need 1% tolerance on gain. Does Avago
sell a more precise version?
4.4: The output doesn’t go all the way to V
REF
when the input is above full scale.
Why not?
4.5: Does the gain change if the internal
LED light output degrades with time?
4.6: Why is gain defined as V
REF
/504 mV, not
V
REF
/512 mV as expected, based on
Figure 24?