Application Hints (Continued)
NOISE
Current noise generated by the LM134 is approximately 4
times the shot noise of a transistor. If the LM134 is used as
an active load for a transistor amplifier, input referred noise
will be increased by about 12dB. In many cases, this is ac-
ceptable and a single stage amplifier can be built with a volt-
age gain exceeding 2000.
LEAD RESISTANCE
The sense voltage which determines operating current of the
LM134 is less than 100mV. At this level, thermocouple or
lead resistance effects should be minimized by locating the
current setting resistor physically close to the device. Sock-
ets should be avoided if possible. It takes only 0.7Ωcontact
resistance to reduce output current by 1% at the 1 mA level.
SENSING TEMPERATURE
The LM134 makes an ideal remote temperature sensor be-
cause its current mode operation does not lose accuracy
over long wire runs. Output current is directly proportional to
absolute temperature in degrees Kelvin, according to the fol-
lowing formula:
Calibration of the LM134 is greatly simplified because of the
fact that most of the initial inaccuracy is due to a gain term
(slope error) and not an offset. This means that a calibration
consisting of a gain adjustment only will trim both slope and
zero at the same time. In addition, gain adjustment is a one
point trim because the output of the LM134 extrapolates to
zero at 0˚K, independent of R
SET
or any initial inaccuracy.
This property of the LM134 is illustrated in the accompanying
graph. Line abc is the sensor current before trimming. Line
a'b'c' is the desired output. A gain trim done at T2 will move
the output from b to b' and will simultaneously correct the
slope so that the output at T1 and T3 will be correct. This
gain trim can be done on R
SET
or on the load resistor used
to terminate the LM134. Slope error after trim will normally
be less than ±1%. To maintain this accuracy, however, a low
temperature coefficient resistor must be used for R
SET
.
A 33 ppm/˚C drift of R
SET
will give a 1% slope error because
the resistor will normally see about the same temperature
variations as the LM134. Separating R
SET
from the LM134
requires 3 wires and has lead resistance problems, so is not
normally recommended. Metal film resistors with less than
20 ppm/˚C drift are readily available. Wire wound resistors
may also be used where best stability is required.
APPLICATION AS A ZERO TEMPERATURE
COEFFICENT CURRENT SOURCE
Adding a diode and a resistor to the standard LM134 con-
figuration can cancel the temperature-dependent character-
istic of the LM134. The circuit shown in
Figure 3
balances
the positive tempco of the LM134 (about +0.23 mV/˚C) with
the negative tempco of a forward-biased silicon diode (about
−2.5 mV/˚C).
The set current (I
SET
) is the sum of I
1
and I
2
, each contribut-
ing approximately 50% of the set current, and I
BIAS
.I
BIAS
is
usually included in the I
1
term by increasing the V
R
value
used for calculations by 5.9%. (See CALCULATING R
SET
.)
The first step is to minimize the tempco of the circuit, using
the following equations.An example is given using a value of
+227µV/˚C as the tempco of the LM134 (which includes the
I
BIAS
component), and −2.5 mV/˚C as the tempco of the di-
ode (for best results, this value should be directly measured
or obtained from the manufacturer of the diode).
With the R
1
to R
2
ratio determined, values for R
1
and R
2
should be determined to give the desired set current. The
formula for calculating the set current at T = 25˚C is shown
below, followed by an example that assumes the forward
voltage drop across the diode (V
D
) is 0.6V, the voltage
across R
1
is 67.7mV (64 mV + 5.9% to account for I
BIAS
),
and R
2
/R
1
= 10 (from the previous calculations).
DS005697-4
FIGURE 2. Gain Adjustment
DS005697-28
FIGURE 3. Zero Tempco Current Source
LM134/LM234/LM334
www.national.com 6