The Direct Answer on Instrumentation Op Amp Gain
Unlike standard operational amplifiers that require three or four precision-matched resistors to set differential gain, the instrumentation op amp gain is typically programmed by a single external resistor ($R_G$) connected between two dedicated pins. This internal architecture—usually consisting of two input buffering op-amps and one difference amplifier—forces the gain equation to rely on internal laser-trimmed resistors, guaranteeing high Common-Mode Rejection Ratio (CMRR) even at high gains.
For industry-standard parts like the Texas Instruments INA128 or Analog Devices AD620, the gain formula is:
$G = 1 + \frac{50k\Omega}{R_G}$
If you need a gain of 100, you simply solve for $R_G$: $R_G = 50,000 / (100 - 1) = 505.05\Omega$. You then place a standard 1% 505Ω resistor across the $R_G$ pins. This single-resistor paradigm eliminates the CMRR degradation caused by external resistor mismatch, which is the fatal flaw of discrete difference amplifier circuits.
Pinout, Symbol, and Operating Regions
Before wiring a circuit, you must understand the standard 8-pin instrumentation amplifier symbol and pinout. The symbol features a non-inverting input (+), an inverting input (-), an output, positive and negative supply rails (V+, V-), a Reference (Ref) pin, and two Gain-setting pins ($R_G$).
The Reference pin is critical: it sets the zero-point (output baseline) of the amplifier. If you tie Ref to ground, an input differential of 0V yields an output of 0V. If you tie Ref to 2.5V, a 0V differential input yields a 2.5V output, allowing you to measure bipolar signals on a single-supply microcontroller ADC.
Core Specifications and Operating Regions (Safe Defaults)
| Part Number | Supply Voltage Range | Quiescent Current ($I_q$) | Input Offset Voltage ($V_{os}$) | CMRR (at G=10) | Best Application |
|---|---|---|---|---|---|
| INA128 (TI) | ±2.25V to ±18V | 700 µA | 50 µV (max) | 120 dB | General purpose, ±15V bench setups, bridge sensors |
| INA333 (TI) | 1.8V to 5.5V (Single) | 50 µA | 10 µV (max) | 106 dB | Battery-powered 3.3V/5V MCU systems, zero-drift needs |
| AD620 (ADI) | ±2.3V to ±18V | 900 µA | 125 µV (max) | 100 dB | Medical front-ends (ECG), high-accuracy industrial |
| INA828 (TI) | 2.7V to 36V (Single) | 150 µA | 35 µV (max) | 104 dB | High-side current sensing, 24V industrial loops |
Sources: TI INA128 Datasheet, Analog Devices AD620 Datasheet.
Setting the Gain: A Complete Load Cell Application Circuit
Let us move from theory to the workbench. A common use case for setting instrumentation op amp gain is amplifying a Wheatstone bridge load cell for a microcontroller ADC.
The Design Scenario
- Sensor: 350Ω load cell with a sensitivity of 2 mV/V.
- Excitation Voltage: 5V (yielding a maximum full-scale differential output of 10 mV).
- Target Output: 0.5V to 4.5V (to fit a 5V ADC with headroom, utilizing a 2.5V Ref for bidirectional load measurement).
- Required Output Swing: 4.0V peak-to-peak for a 10 mV input swing.
Calculating the Gain and $R_G$
We need an output swing of 4.0V from an input swing of 0.01V.
Required Gain ($G$) = $4.0V / 0.01V = 400$.
Using the INA128 formula: $400 = 1 + (50,000 / R_G)$.
$R_G = 50,000 / 399 = 125.31\Omega$.
We will use a standard 1% 124Ω resistor, which gives an actual gain of 404.2. This is perfectly acceptable, as we can calibrate the final offset in software.
Complete Circuit Component Values
- U1: INA128P (8-pin DIP).
- $R_G$: 124Ω (1% metal film) placed directly between Pin 1 and Pin 8.
- Decoupling: 0.1 µF ceramic capacitors from Pin 4 (V-) to Ground, and Pin 7 (V+) to Ground. Place these within 2mm of the IC pins.
- Input Filtering: To prevent RF rectification (a common failure mode where high-frequency noise is demodulated into DC offset), place 100Ω resistors in series with both inputs, and a 100 nF X7R capacitor directly across the two input pins. This creates a differential low-pass filter with a cutoff of roughly 15.9 kHz.
- Reference Pin (Pin 5): Do not tie this to a simple resistive voltage divider. The Ref pin draws dynamic current; any impedance on this pin will unbalance the internal difference amplifier and destroy your CMRR. Drive Pin 5 with a buffered 2.5V reference (e.g., an LM4040-2.5 buffered by a TLV2371 op-amp voltage follower).
Bench Tip: Keep the physical trace length between the $R_G$ pins and the gain resistor as short as possible. Long traces add parasitic inductance and capacitance, which can cause high-frequency peaking or oscillation at gains above 100.
Biasing, Selection, and the 'Diamond Plot'
Selecting the right part and biasing it correctly requires understanding the Input Common-Mode Range (ICMR). Beginners often assume that if the supply is ±15V, the inputs can swing anywhere from -15V to +15V. This is false.
Older bipolar architectures like the INA128 require the inputs to stay at least 1.5V to 2V away from the supply rails. Furthermore, you must consider the internal nodes. Even if the input voltage is within the supply rails, a high differential voltage combined with a high common-mode voltage can saturate the internal input stage op-amps. Manufacturers visualize this limitation using a Common-Mode vs. Differential Input 'Diamond Plot' in the datasheet. If your operating point falls outside the diamond, the output will rail out or exhibit severe non-linearity.
How to Choose Your Safe Default
- Choose the INA128 when: You have dual supplies (e.g., ±12V or ±15V), need a robust, well-documented workhorse, and are measuring standard strain gauges or thermocouples.
- Choose the INA333 when: You are running off a single 3.3V or 5V battery, need rail-to-rail inputs/outputs, and require zero-drift (chopper-stabilized) performance to eliminate low-frequency 1/f noise.
- Choose the INA828 when: You are doing high-side current sensing on a 12V or 24V industrial bus and need a wide common-mode range that includes the positive supply rail.
Failure Modes and Multimeter Testing
Instrumentation amplifiers rarely fail spontaneously; they fail due to electrical overstress. The most common failure modes include input overvoltage (exceeding the absolute maximum rating of V+ + 0.5V), ESD damage to the high-impedance input gates, and latch-up caused by driving the inputs beyond the supply rails while the IC is powered.
If your circuit is outputting a railed voltage or a noisy, drifting signal, follow these numbered steps to test the IC with a digital multimeter (DMM):
- Verify Supply Rails: Power the circuit. Set your DMM to DC voltage. Measure Pin 7 to Ground (should be exactly +Vcc) and Pin 4 to Ground (should be exactly -Vee or 0V for single supply). If these are drooping, your decoupling is inadequate or the IC has an internal short.
- Check the Reference Voltage: Measure Pin 5 to Ground. It must exactly match your intended baseline (e.g., 2.500V). If it is off by even 20mV, your buffer op-amp is failing or your voltage divider impedance is too high.
- Validate Input Common-Mode Voltage: Measure the DC voltage at Pin 3 (+) and Pin 2 (-) relative to Ground. Ensure both voltages fall strictly within the 'Diamond Plot' limits of your specific IC. If they are pegged to the supply rails, your sensor bridge is likely open or disconnected.
- The 'Shorted Input' Test: Power down, and use a jumper wire to short Pin 2 and Pin 3 together. Power the circuit back on. Measure the Output (Pin 6). The output voltage should now read exactly the voltage present on the Ref pin (Pin 5), plus or minus the IC's input offset voltage (typically < 100 µV). If the output is railed to V+ or V-, the internal difference amplifier stage is blown.
- Check for Parasitic Oscillation: Switch your DMM to AC voltage mode and measure the output. If you read more than a few millivolts of AC on a DC load cell circuit, your IC is oscillating. This is usually caused by capacitive loading on the output pin or missing $R_G$ bypassing. Add a 100Ω series resistor at the output followed by a 1nF capacitor to ground to isolate the capacitive load.
By treating instrumentation op amp gain not just as a math equation, but as a system constrained by common-mode limits, reference impedance, and RF filtering, you will eliminate the vast majority of bench-level debugging headaches.
