Philips AN1651 Stereo Amplifier User Manual


 
Philips Semiconductors Application note
AN1651Using the NE/SA5234 amplifier
1991 Oct
5
4
11
+
47k
5234
100
x10
HP
3585
SPECTRUM
ANALYZER
10
600
+2.5V
–2.5V
SL00634
Figure 6. Noise Test Circuit
IV. NOISE REFERRED TO THE INPUT
The typical spectral voltage noise referred to each of the op amps in
the NE/SA5234 is specified to be 25nV/Hz. Current noise is not
specified. In the interest of providing a balance of information on the
device parameters, a small sample of the standard NE5234s, were
tested for input noise current. While this data does not represent a
specification, it will give the designer a ball park figure to work with
when beginning a particular design with the device. For
completeness I have provided the corresponding spectral noise
voltage data for the same sample. The data was taken using an
HP3585A spectrum analyzer which has the capability of reading
noise in nV/Hz.
The test circuit is shown in Figure 6. As is typical for such
measurements the amplifier under test is terminated at its input first
with a very low resistance, for the voltage noise reading, followed by
the same test with a high value of resistance to register the effect of
current noise. The amplifier is set to a non-inverting
closed-loop gain of 20dB. Dual supply operation was chosen to
allow direct termination of the input resistors to ground.
The measurements were made over the range from 200Hz to 2kHz.
Each sample is measured at 200Hz, 500Hz, 1kHz and 2kHz. The
data is averaged for each frequency and then the small sample
distribution is derived statistically giving the standard deviation
relative to the mean.
Referring to the graph in Figure 7a, the equivalent voltage noise is
seen to average 18 nV/Hz. The 95% confidence interval is
determined to be approximately one nV/Hz. The majority of the
errors which contribute to this measurement are due to the thermal
noise of the parallel combination of the feedback resistor network, in
addition to the 10 termination resistor on the non-inverting input.
At 300° Kelvin a 10 resistor generates 0.4 nV/Hz and the
feedback network’s equivalent resistance of 90 generates
1.2nV/Hz. Their order-of-magnitude difference from the main noise
sources allows them to be neglected in the overall calculation of
total stage noise.
Noise current is measured across a 47k resistor and averaged in
the same manner. The thermal noise generated by this large
resistance is not insignificant. At room temperature it is 28nV/Hz
and must be subtracted from the total noise as measured at the
output of the op amp in order to arrive at the equivalent current
generated noise voltage. Figure 7b shows the derived current
noise distribution for the small sample of 10 NE5234 devices. The
result shows that noise current in the 200Hz to 2kHz frequency is
typically 0.2pA/Hz. The 1/f region was not determined for either
current or voltage noise.
95%
INT.
E
n
for R
S
= 10 -nV/Hz
22
19
18
17
16
100 200 2000 10000
nV
Hz
Ǹ
a.
pA
ń
Hz
Ǹ
0.5
12
0.1
100
200
1k
2k
10000Hz
f
P
in 10
P
b.
SL00635
Figure 7. Typical Noise Current and Voltage vs Frequency
V. GUIDE LINES FOR MINIMIZING NOISE
When designing a circuit where noise must be kept to a minimum,
the source resistances should be kept low to limit thermally
generated degradation in the overall output response.
Orders-of-magnitude should be kept in mind when evaluating noise
performance of a particular circuit or in planning a new design. For
instance, a transducer with a 10k source resistance will generate
2µV of RMS noise over a 20kHz bandwidth. Using the graphical
data above, total noise from a gain stage may be calculated.-
25nVń Hz
Ǹ
@ BW
Ǹ
+ 3.5mV
RMS
(EQ. 1.)
Amplifier Noise Voltage
BW + 10kHz
Noise from source 10k Resistance –
14nVń Hz
Ǹ
@ BW
Ǹ
+ 20mV
RMS
(EQ. 2.)
Noise Voltage from source resistance
0.2pAń Hz
Ǹ
@ 10
3
@ BW
Ǹ
+ 0.28mV
RMS
(EQ. 3.)
Current generated noise
The total noise is the root-to-sum-of-the-squares of the individual
noise voltages –
En
+
(3.5)
2
)
(2.0)
2
)
(0.28)
2
Ǹ
(EQ. 4.)
+ 4.04mV
RMS
To determine the signal-to-noise ratio of the stage we must first
choose a stage gain, make it 40dB, and a signal voltage magnitude
from the transducer which we will set at 10mV
RMS
. The resulting