Can an analog input with a very low input impedance, cause the output of an input sensor to oscillate? And overheat or fail?

Question

I have a dual output sensor connected to two analog inputs on a controller.

The main output of the sensor works from 0.5 to 4.5 V DC and the other output is inverted such that it works from 4.5 to 0.5

Evidence from the field indicates one of the two sensor outputs will deviate from the other for short periods of time, that become longer during the operation period. The sensor manufacturer reports the returned sensors work correctly.

The field reports the same sensor output is usually the problem, but I have not found out if it is the normal or inverted output yet.

The controller’s data sheet says “the input impedance for the analog inputs is 200-250 kΩ”

The sensor’s datasheet says “the load resistance is 10 kΩ minimum (resistive to GND)”

I believe this is saying the input impedance for the analog input is too low for direct use with the sensor.

I admit a bit of ignorance regarding input impedance versus load resistance.

I am thinking something is triggering the sensor output to oscillate which could look like a deviation, the oscillation causes heating which causes the oscillation to get worse over the operating period.

The controller input which is reporting the deviation is filtered (somehow), so it is possible a high-speed oscillation is filtered out until the sensor output started having heating problems.

Answer 1

Clarification about impedance specifications

It looks to me like you are having some trouble with the impedance specifications, so here is a quick explanation:

The output of your sensor represents a voltage source (probably an opamp) with some series source resistance \$R_s\$. If you connect that output to the load resistance \$R_L\$ (\$200-250k\Omega\$) of the analog input card, there will be some voltage drop across \$R_s\$ which affects the reading across \$R_L\$.

In order to reduce this voltage error, the value of \$R_s\$ must be much smaller than \$R_L\$ (maybe 100x or 1000x, depending on the desired accuracy).

So if the sensor's datasheet specifies the minimal load impedance as \$10k\Omega\$, they just want to ensure that their output reading will not be affected by the internal source resistance \$R_s\$ of the output driver (\$R_s\$ migth be just \$1-10\Omega\$ in your case). There is absolutely no problem here with the input impedance of the analog input card being larger than \$10k\Omega\$.

As you probably deal with some kind of PLC system, it might interesting for you to note that the \$10k\Omega\$ load impedance specification of the sensor is no coincidence. The industrial PLC standard IEC61131-2 defines the minimal input impedance of an analog input as \$10k\Omega\$ (for voltage measurements). So this sensor will be compatible with every industrial PLC card.

Oscillations?

I doubt that the sensor (probably an off the shelf industrial product) can be brought into oscillation just by connecting it to the analog input card. The analog front-end of an input card will most likely consist of a voltage divider, some RC filter, an opamp buffer and finally the analog-to-digital converter.

However, if your wiring between the sensor and the analog input is very very bad (i.e. no twisted pair, no shielding and long cables) there could be some errors introduced by external interference.

Other problem sources

The sensor manufacturer has tested the returned units and found no issues with them. But what about the analog input card? One special thing that comes into my mind is if the two sensor signals are getting processed by two separate analog-to digital converters. If the reference voltages of these converters drift apart over time (and/or temperature) it could explain the behaviour you observe.