20203346

FIGURE 7. Sensor Output with No Load

Referring to the simplified diagram in Figure 6, and assuming

that required full scale at the output of the amplifier is 2.5V, a

gain of 23.2 is needed for U1. It is clear from the Thevenin

equivalent circuit in Figure 8 that a sensor Thevenin equiva-

lent source resistance, R

both the inverting and non-inverting inputs of the LMV641.

Therefore, the required gain is:

Choosing R

580 k

Ω. The actual values chosen will depend on the full-

scale needs of the succeeding circuitry as well as bandwidth

requirements. The values shown here provide a −3 dB band-

width of approximately 431 kHz, and are found as follows.

20203347

FIGURE 8. Thevenin Equivalent Showing Required Gain

By choosing input resistor values for R

to ten times the bridge element resistance, the bridge is min-

imally loaded and the offset errors induced by the op amp

stages are minimized. These resistors should have 1% toler-

ance, or better, for the best noise rejection and offset mini-

mization.

Referring once again to Figure 4, U2 is an additional gain

stage with a thermistor element, R

performs a temperature compensation function for the bridge

so that it will have greater accuracy over a wide range of op-

erational temperatures. With mangetoresistive sensors, tem-

perature drift of the bridge sensitivity is negative and linear,

and in the case of the sensor used here, is nominally −3000

PP/M. Thus the gain of U2 needs to increase proportionally

with increasing temperature, suggesting a thermistor with a

positive temperature coefficient. Selection of the temperature

compensation resistor, R

required, on the thermistor chosen, and is dependent on the

thermistor’s %/°C shift in resistance. For best op amp com-

patibility, the thermistor resistance should be greater than

1000

Ω. R

resistor. Because the temperature coefficient of the AMR

bridge is largely linear, R

fashion with temperature, thus R

R

Gain Error and Bandwidth Consideration if Using an

Analog to Digital Converter

The bandwidth available from Figure 4 is dependent on the

system closed loop gain required and the maximum gain-er-

ror allowed if driving an analog to digital converter (ADC). If

the output from the sensor is intended to drive an ADC, the

bandwidth will be considerably reduced from the closed-loop

corner frequency. This is because the gain error of the pre-

amplifier stage needs to be taken into account when calcu-

lating total error budget. Good practice dictates that the gain

error of the amplifier be less than or equal to half LSB (prefer-

ably less in order to allow for other system errors that will eat

up a portion of the available error budget) of the ADC. How-

ever, at the −3 dB corner frequency the gain error for any

amplifier is 29.3%. In reality, the gain starts rolling off long

before the −3 dB corner is reached. For example, if the am-

plifier is driving an 8-bit ADC, the minimum gain error allowed

for half LSB would be approximately 0.2%. To achieve this

gain error with the op amp, the maximum frequency of interest

can be no higher than

where n is the bit resolution of the ADC and f

loop corner frequency.

Given that the LMV641 has a GBW of 10 MHz, and is oper-

ating with a closed loop gain of 26.3, its closed loop bandwidth

is 380 kHZ, therefore

which is the highest frequency that can be measured with re-

quired accuracy.

www.national.com

14