where RL is the load resistor in the drains of Q3 and Q8, and
RS is the resistor connected between the sources of the input
transistors Q4 and Q7. The connections for various RS com-
binations are brought out to device pins LNPGS1, LNPGS2,
and LNPGS3 (pins 13-15 for channel A, 22-24 for channel B).
These Gain Strap pins allow the user to establish one of four
fixed LNP gain options as shown in Table I.
NOISE (nV/√Hz)
LNP GAIN (dB)
Input-Referred
Output-Referred
25
22
17
5
1.35
1.41
1.63
4.28
2260
1650
1060
597
TABLE II. Equivalent Noise Performance for MGS = 111 and
VCACNTL = 3.0V with 50Ω source impedance.
LNP PIN STRAPPING
LNP GAIN (dB)
LNPGS1, LNPGS2, LNPGS3 Connected Together
LNPGS1 Connected to LNPGS3
LNPGS1 Connected to LNPGS2
All Pins Open
25
22
17
5
The LNP is capable of generating a 2VPP differential signal.
The maximum signal at the LNP input is therefore 2VPP
divided by the LNP gain. An input signal greater than this
would exceed the linear range of the LNP, an especially
important consideration at low LNP gain settings.
TABLE I. Pin Strappings of the LNP for Various Gains.
The VCA2611 is an upgraded version of the VCA2616. The
only difference between the VCA2616 and the VCA2611 is the
input structure to the LNP. The VCA2616 is limited to –0.3V
negative-going input spikes; the VCA2611 is limited to –2.0V
negative-going input spikes. This change allows the user to
use slower and less expensive input clamping diodes prior to
the LNA input. In some designs, input clamping may not be
required.
It is also possible to create other gain settings by connecting
an external resistor between LNPGS1 on one side, and
LNPGS2 and/or LNPGS3 on the other. In that case, the
internal resistor values (see Figure 4) should be combined
with the external resistor to calculate the effective value of RS
for use in Equation 1. The resulting expression for external
resistor value is given in Equation 2:
2RS1RL + 2RFIXRL – Gain ×RS1RFIX
Gain× RS1 – 2RL
(2)
REXT
=
ACTIVE FEEDBACK WITH THE LNP
One of the key features of the LNP architecture is the ability
to employ active-feedback termination to achieve superior
noise performance. Active-feedback termination achieves a
lower noise figure than conventional shunt termination, es-
sentially because no signal current is wasted in the termina-
tion resistor itself. Another way to understand this is to
consider first that the input source, at the far end of the signal
cable, has a cable-matching source resistance of RS. Using
conventional shunt termination at the LNP input, a second
terminating resistor of value RS is connected to ground.
Therefore, the signal loss is 6dB due to the voltage divider
action of the series and shunt RS resistors. The effective
source resistance has been reduced by the same factor of 2,
where REXT is the externally selected resistor value needed
to achieve the desired gain setting, RS1 is the fixed parallel
resistor in Figure 4, and RFIX is the effective fixed value of the
remaining internal resistors: RS2, RS3, or (RS2 || RS3), de-
pending on the pin connections.
Note that the best process and temperature stability will be
achieved by using the pre-programmed fixed-gain options of
Table I, since the gain is then set entirely by internal resistor
ratios, which are typically accurate to ±0.5%, and track quite
well over process and temperature. When combining exter-
nal resistors with the internal values to create an effective RS
value, note that the internal resistors have a typical tempera-
ture coefficient of +700ppm/°C and an absolute value toler-
ance of approximately ±5%, yielding somewhat less predict-
able and stable gain settings. With or without external resis-
tors, the board layout should use short Gain Strap connec-
tions to minimize parasitic resistance and inductance effects.
but the noise contribution has been reduced by only the √2
,
only a 3dB reduction. Therefore, the net theoretical SNR
degradation is 3dB, assuming a noise-free amplifier input. (In
practice, the amplifier noise contribution will degrade both
the unterminated and the terminated noise figures, some-
what reducing the distinction between them.)
The overall noise performance of the VCA2616 and VCA2611
will vary as a function of gain. Table II shows the typical input-
and-output-referred noise densities of the entire VCA2616 and
VCA2611 for maximum VCA and PGA gain; that is, VCACNTL
set to 3.0V and all MGS bits set to 1. Note that the input-
referred noise values include the contribution of a 50Ω fixed
source impedance, and are therefore somewhat larger than
the intrinsic input noise. As the LNP gain is reduced, the noise
contribution from the VCA/PGA portion becomes more signifi-
cant, resulting in higher input-referred noise. However, the
output-referred noise, which is indicative of the overall SNR at
that gain setting, is reduced.
See Figure 5 for an amplifier using active feedback. This
diagram appears very similar to a traditional inverting ampli-
fier. However, the analysis is somewhat different because
the gain A in this case is not a very large open-loop op amp
gain; rather, it is the relatively low and controlled gain of the
LNP itself. Thus, the impedance at the inverting amplifier
terminal will be reduced by a finite amount, as given in the
familiar relationship of Equation 3:
RF
RIN
=
(3)
1+ A
(
)
where RF is the feedback resistor (supplied externally be-
tween the LNPINP and FB terminals for each channel), A is
To preserve the low-noise performance of the LNP, the user
should take care to minimize resistance in the input lead. A
parasitic resistance of only 10Ω will contribute 0.4nV/√Hz
.
VCA2616, VCA2611
11
SBOS234E
www.ti.com
TAOS [ TEXAS ADVANCED OPTOELECTRONIC SOLUTIONS ]
