LOG114: What limits input voltage range

Hi Adam,

The primary LOG114 input limitation is the absolute maximum current limit of 10 mA. The absolute input voltage limits are (V−) −0.5V to (V+) + 0.5V, but that is not a linear operating range and the current must be limited to 10 mA max over that voltage range.

An input current flows through the logging transistor that is connected between the input and output connections of an input transimpedance stage, A1 or A2. The logging transistor is designed to have minimal junction capacitance. It has limited current carrying capacity and power dissipation. The recommended maximum input current is 3.5 mA, but may be used to as high as 10 mA. However, currents from 3.5 to 10 mA results in increasingly higher transistor power dissipation which degrades the log conformity, gain and other electrical performances. Many of the Typical Characteristics graphs show an increase in error levels when the input current exceeds 1 mA.

Datasheet page 18 describes using the LOG114 with a voltage input. The limitations of using a voltage input are described in that section. The range limitations are related to the high resistance required for the high end of the voltage and corresponding higher-end current, vs using that same resistor to set up the current when the input voltage is low. Keep in mind that when the input current to the LOG114 will be very low when current through the shunt resistor is low. In practical setups it is often difficult to keep induced noise from corrupting the intended input current. 

One common mistake made when applying the LOG114 is the that the exposed thermal die pad on the package underside isn't connected to V−. For dual supply setups the pad is connected to V-, and for a single supply setup that is ground. Make sure you have observed that requirement, or the LOG114 performance may not be as expected.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hello Smitty,

I am attempting better understand the LOG114 information presented in the first paragraph of the Voltage Input section, on datasheet page 18. An example is provided in which the input voltage is 10 V and the input current level is 10 nA, requiring a series resistance of 1 Gigaohm. If the voltage is decreased by a decade to 1 V, then for that resistance value the current drops to 1 nA and drops further to 100 pA with a 100 mV source. 100 pA is the minimum recommended input current for LOG114. Hence, the mention of "dynamic input range is limited to approximately three decades of input voltage" is valid for that example.

Certainly that is a different set of conditions than what you are proposing. Two-hundred millivolts developed across the 500 uOhm results in 10 mA flowing through the 20 Ohm series input resistor and the logging transistor in the LOG114 feedback path. That would be the high-end condition. I've already pointed out the issues of exceeding the recommended maximum of 3.5 mA.

On the low end when 4 uA is flowing through the 500 uOhm resistor and the voltage develped across the resistor would be 2 nV. Then, the ideal current through the 20 Ohm resistor would be 100 pA. I am very skeptical how noise free that 2 nV through the 500 uOhm resistor will be. The transimpedance gain of the LOG114 is extremely high at low current levels and the logging transistor is the equivalent of an extremely high feedback resistance. The external and internal noise become significant factors in the measured results. When we have set the LOG114 up in lab conditions and operated it with 100 pA level currents it is quite difficult to keep line and other noises from corrupting the measurment. The circuit is usually encased in a metal enclosure, power by an internal bettery supply and we use coaxial cable connections for signals going into, and out of the metal box. Even then, it is difficult to get the noise to acceptable levels.

Regards, Thomas

Precision Amplifiers Applications Engineering

I agree with the challenges of measuring the really low values, and in practice will be unlikely that we care to measure that low of a value. However, I'd still like to measure with a log amp, and would be happy to have 112dB of bandwidth (400A down to 1mA). With respect to the issues above 3.5mA, we don't need to necessarily ensure linearity for currents beyond 100A and won't happen all that often if at all.