I've got a new circuit idea I'm playing with, but I realized there's a potential problem I hadn't thought about before. In the circuit below, C2 will charge to somewhere around 2.5V and stay in that general range most of the time. When power to the circuit is disconnected, it will continue to hold the input to op amp U2 at that voltage.


(I apologize for the messiness of the schematic. I was starting to clean it up and annotate it when I realized I might have a problem with it. It's a work in progress.)

I know there are abs max input limits, and they're referenced to the supply voltage. When the supply voltage drops, is it critical to bleed that cap immediately? Do I need a diode to the supply voltage maybe? Do op amps usually have input protection diodes like GPIO pins often do? If so, can they handle discharging a 10uF tantalum cap?

It seems like filters in the midst op amp circuits are pretty common, so I assume issues like this are a possibility pretty regularly, but I don't remember seeing protection diodes sprinkled liberally through every op amp circuit I look at. Am I just being paranoid, or is this a real issue that I need to address?

For anyone who's interested, the circuit above is a new idea for a silly one-off project where I'm amplifying the output of a Hall Effect sensor to try to read very, very small magnetic fields with it. It was inspired by this thread:
https://forum.allaboutcircuits.com/threads/hall-effect-sensor-circuit-recommendations.143813/

I realized pretty quickly that I could manually trim my reference for the null point and get very good results in an given moment, but that sensor null point drift made my readings useless in a fairly short order. So, the idea in this circuit is to use a few op amp stages and a very slow RC filter to get a long term average (10 second time constant as drawn, will probably increase to 47 second) of the idle sensor null value and the amplify the difference between that and the current value.

Obviously I could get better low Gauss measurements with a dedicated digital magnetometer IC, but I just thought it would be a fun exercise in analog design to see how far I could push the limits with a simple analog sensor.

 

Either put a diode from the capacitor to the pos supply rail, or replace the tant for an electrolytic cap.

 

I know there are abs max input limits, and they're referenced to the supply voltage.

If you're actually using an LT1490A, they aren't. The LT1490A is one of Linear Tech's "Over-The-Top" series of op amps, and these can handle input voltages up to ≈ +40 volts regardless of supply voltage.

When the supply voltage drops, is it critical to bleed that cap immediately? Do I need a diode to the supply voltage maybe?

Maybe, maybe not. At the low voltages you're working with, and with only a 10 μF cap, I wouldn't worry about it.

Some op amps are susceptible to catastrophic failure via latch-up if the inputs exceed the supply; but AFAIK the LT1490A does not, and I've never had any trouble with these parts.

Do op amps usually have input protection diodes like GPIO pins often do? If so, can they handle discharging a 10uF tantalum cap?

I don't know whether they do or not generally, although I would expect protection diodes on CMOS op amps.

 

If you're actually using an LT1490A, they aren't. The LT1490A is one of Linear Tech's "Over-The-Top" series of op amps, and these can handle input voltages up to ≈ +40 volts regardless of supply voltage.


Maybe, maybe not. At the low voltages you're working with, and with only a 10 μF, I wouldn't worry about it.

Some op amps are susceptible to catastrophic failure via latch-up if the inputs exceed the supply; but AFAIK the LT1490A does not, and I've never had any trouble with these parts.


I don't know whether they do or not generally, although I would expect protection diodes on CMOS op amps.

No, I usually sim with whatever I've got in LTSpice unless I think specific traits of the part will change things dramatically.

At the moment, I've got a rough, breadboarded circuit using NTE859 (which I believe is a TL074 equivalent) with some extra +/-9V supply rails to power it separately from the 5V Hall effect circuitry (sharing a common ground.) I've found in my test circuit that input offset on the op amps is a pretty big factor by the time it's amplified x100, and I'm planning to use the Advanced Linear Devices ALD1722PAL in the final build (cheapest op amp I found with rail-to-rail I/O, 5V single-supply capability, and very low input offset voltage.)

 

...or replace the tant for an electrolytic cap.

I'm curious, what would be the advantage there? I know there are a lot of behavioral differences between tantalum and electrolytic caps, but I can't think of how one or the other would be more likely to damage an op amp input.

 

The datasheet for the LT1490A says its maximum input current is 12mA and its schematic shows 1k series input resistors. Since your supply is only 5V then it cannot produce an input current from the capacitor of more than 5mA so it will be fine.

 

So, I'm planning to use the Advanced Linear Devices ALD1722:

https://www.digikey.com/product-det...r-devices-inc/ALD1722PAL/1014-1095-ND/2414371

http://aldinc.com/pdf/ALD1722.pdf

I don't see anything in the datasheet showing internal schematic details, nor input current limits comparable to the LT1490 specs.

I'm guessing this means I should just play it safe and use a diode from the input to the supply voltage.

Thanks for all the insights. I'll have a better idea of some things to look for in the datasheets next time. Cheers!

 

And by the sound of it the input voltages must be less than 300mV outside the power rails, which would be quite difficult to achieve.

5. ALD1722/ALD1722G operational amplifier has been designed to provide full static discharge protection. Internally, the design has been carefully implemented to minimize latch up. However, care must be exercised when handling the device to avoid strong static fields that may degrade a diode junction, causing increased input leakage currents. In using the operational amplifier, the user is advised to power up the circuit before, or simultaneously with, any input voltages applied and to limit input voltages to not exceed 0.3V of the power supply voltage levels.

 

And by the sound of it the input voltages must be less than 300mV outside the power rails, which would be quite difficult to achieve.

5. ALD1722/ALD1722G operational amplifier has been designed to provide full static discharge protection. Internally, the design has been carefully implemented to minimize latch up. However, care must be exercised when handling the device to avoid strong static fields that may degrade a diode junction, causing increased input leakage currents. In using the operational amplifier, the user is advised to power up the circuit before, or simultaneously with, any input voltages applied and to limit input voltages to not exceed 0.3V of the power supply voltage levels.

Hmmm, maybe I need to use a Schottky? They usually have Vf around that level, right?

Also, thanks for calling my attention to that portion of the datasheet. I had somehow overlooked that bit.