This circuit will eliminate most of the LM134’s temperature

coefficient, and it does a good job even if the estimates of the

diode’s characteristics are not accurate (as the following ex-

ample will show). For lowest tempco with a specific diode at

the desired I

tested over temperature. If the measured tempco of I

positive, R

negative, R

for use in this circuit is the 1N457 because its tempco is cen-

tered at 11 times the tempco of the LM134, allowing R

R

with non-zero tempcos by setting the tempco component of

the tempco equation to the desired value instead of 0.

EXAMPLE: A 1mA, Zero-Tempco Current Source

First, solve for R

The values of R

sistor values (R

0.75% error.

If the forward voltage drop of the diode was 0.65V instead of

the estimate of 0.6V (an error of 8%), the actual set current

will be

an error of less than 5%.

If the estimate for the tempco of the diode’s forward voltage

drop was off, the tempco cancellation is still reasonably ef-

fective. Assume the tempco of the diode is 2.6mV/˚C instead

of 2.5mV/˚C (an error of 4%). The tempco of the circuit is

now:

A 1mA LM134 current source with no temperature compen-

sation would have a set resistor of 68

Ω and a resulting

tempco of

its estimated value, the circuit still eliminates 98% of the

LM134’s inherent tempco.

Typical Applications

Ground Referred Fahrenheit Thermometer

DS005697-15

Trim R3 to calibrate

www.national.com

7