TLV522: UNSTABLE AMPLIFIER

I forgot to mention. There is a bug in the second stage of the circuit. The -inB and +inB inputs need to be swapped. We have corrected this by cutting traces and jumper wires.

More info: This circuit is used for sensing shock. The piezo can generate an appreciable voltage with a hard shock. Could the piezo be damaging the inputs. I read there is a +0.3V and -0.3V above and below rails limits. However this device has 10mA protection diodes on the inputs. As mentioned in my OP we also notice that when one opamp is replaced the opamp next to it gets compromised. Piezo generate voltage with heat and this could also be above or below the above limits. Again one would think the protection diodes will kick in and protect the device.

Hello Manjula,

This is a high impedance application (1Megohm), so it will be affected by layout and cleanliness. You may wish to consider applying a layer of beeswax (paraffin) or other conformal coating over the circuit to seal out dust and humidity. You MUST fully clean the board each time a component is replaced. Burnt solder flux is resistive and can create leakage paths between pins. Be sure to clean between and under the IC pins and flush UNDER the package.

Also - because it is a high impedance input, it will be affected by movement near the sensor wires (and traces). Is the board in a shielded environment?

There are protection diodes from the inputs (and output) to each supply.

Any transients in the inputs will be passed (clamped) to the power supply. It is possible to lift the power supply - but the bypass caps should clamp the transient if they are large enough (and ESR low enough). If these are tiny SMT resistors and caps, it is possible they are breaking down under the high voltage.

However, you have 75k in series with the sensor, so the sensor would need to generate over 750V to get 10mA.it is not uncommon for these sensors to generate >1kV when struck directly.

10mA is abs-max for the diodes - you really want to keep the diode current as low as possible (<1mA recommended, <100uA even better). Repeated pounding on the diodes at 10mA can damage them over time as they are NOT rated for continuous use (just occasional events like an ESD event!).

If the sensor does generate a high voltage transient, it will generate it between the 1.5V bus and pins 1&2.

If the transient risetime is fast enough, it will pass straight through C4 and hit the pins. C6 and R9 will dampen the transient a little. If the risetime is slow enough, it will probably charge C6 and C4 to a high voltage.

Do you know the amplitude and rise time of the piezo transients?

Ideally, you should have a clamp diodes across the sensor - preferably between the junction of R3/R4/C4 and the 1.5V line. You can put to diodes in series for a 1.6V clamp if 0.8V is too low. Like this (at 0.8V per diode).

Are you really seeing the offset of the amplifier shift? Read the voltage between pins 2 and 3 with a good (4-5 digit) DMM with the sensor shorted. That will be the amplifiers offset voltage. If it is more than the datasheet max spec (±4mV) - it could be damaged. For accuracy, first place both probes on pin 3 and null the meter to remove any meter/probe thermal offsets (or memorize the zero measurement and subtract it from the between-pins measurement).