INA4181-Q1: Regarding High-Side Current Sensing for Negative voltages

Hello Sunny,

There are several options you can take when measuring current off a negative voltage rail, but tradeoffs between them will need to be considered.

If you need a digital option you should consider reading over this reference design:

http://www.ti.com/lit/ug/tidu361a/tidu361a.pdf

The INA181 or its quad package INA4181 cannot sense current when common mode voltages (VCM) are below -0.1V. There are a couple methods you could take to float the device’s power rail in order to shift the acceptable VCM range. Much of what is covered can be referenced with this document and reference design:

http://www.ti.com/lit/an/sboa198/sboa198.pdf

http://www.ti.com/tool/TIDA-00332

Method 1 – Power the INA4181 or INA2181 with a 3.3V Zener diode off the negative bus rail

This option should present the smallest form factor since you only need to add the Zener diode and the current-limiting resistor Rz. You would probably need two INA2181 devices. One (U1) powered normally with separate power supply that measures the +9V and +4.25V raila and the other INA2181 (U2) that is powered with floating Zener that measures the -9V and -4.25V rails. The only issue here is that you will have to level shift the output of U2 since it will be referenced to a negative voltage.

Another issue to consider here is the power consumption of the Zener diode off your -9V rail. The Zener will need some current in order to be reverse-biased and generate the 3.3V drop which powers U2. This current is determined by the value of Rz. Over temperature and with 0V Vsense, the INA2181 has an IQ of 520µA. This will over time consume current from your power rails and may pose a problem if you are trying to measure milliamps of current or know that the negative voltage rails will over time droop below the Zener diode reverse breakdown voltage if you are measuring battery current.

Note that the Zener diode is necessary because the differential supply voltage for INA181 cannot exceed 5.5V, thus powering it from the -9V rail directly would break the part. Here is schematic where I model INA181 discretely and reference it to its ground pin. I modified the Zener diode to have a breakdown of 5V here so this is not meant to represent an actual Zener device in market.

Method 2 – Use a current sense amplifier with >9V differential power supply range

This allows you to remove the need for a Zener diode providing <5.5V drop and will save some power consumption needed to reverse bias said Zener diode. You will still need to level shift the output of the current sensor. Automotive qualified devices with this criterion include INA282—Q1, LMP8640-Q1, INA193A-Q1, INA270A-Q1, INA203-Q1, INA200-Q1, INA210-Q1, INA199-Q1, INA225-Q1, INA250-Q1, LMP8481-Q1.

Method 3 – Same as Method 1 and 2, but convert amplifier into a Howland Current Pump

This is done by feeding the output pins of U2 into the reference pins with a resistor (RF). A voltage drop will be generated  across RFand the output will be converted into a current which is much easier to level shift to absolute ground via a load resistor RL. If the negative voltage rails will be changing overtime then I would recommend buffering the feedback path with an op amp because the INA181 will lose CMRR in this topology when you load the reference pin with a resistor.

The downside here is that you need two extra resistors and possibly one op amp per INA181 channel. Here is more information and theory of this topology. http://www.ti.com/lit/an/snoa474a/snoa474a.pdf

Method 3 – Use a current sense amplifier with negative VCM range.

These include INA282-Q1, INA200-Q1, INA203-Q1, INA193A-Q1, and LMP860x-Q1. These could be powered with low voltage power rails such 3V and thus would not need level shifting on the output. Unfortunately, they do not come in quad package.

Method 4 – Use operational amplifiers that are referenced to the negative power rails

You can configure OPAs into difference amplifiers that can measure small shunt voltage almost exactly like a current sense amplifier, but many OPAs will have larger acceptable voltage supply ranges so you could potentially power the OPA from the -9V rail directly. You would still need to level shift the output of the OPA since it will be referenced to -9V. Additionally you will take up more space with the precision resistors needed for OPA feedback network.

Hope this helps.

Sincerely,

Peter Iliya

Current Sensing Applications