INA828: How to measure / sense currents of +/-5A with 40uA resolution

Hi Niels,

I would go for a low-side measurement. This will eliminate nearly all of the common mode voltage issues. Low ohmic source resistance (10mR shunt) would allow the use of a chopper-stabilized OPAmp. Here are some nice ones:

What is your signal bandwidth?

Can you increase the shunt resistance? This would enormeously help here :-)

Kai

Hello Niels,

If you are ok with digital output current sensors, then you may want to look into the INA228-q1 (I2C) or the INA229-q1 (SPI), which may be a good fit for your design.

Hello Kai,

increasing the shunt resistance would increase power loss and thermal drift of the shunt resistor. That's why I selected 10m Ohm.
With 100m Ohm, I would have a power loss of 2.5 Watt, only for current-measurment. That's too much.
Perhaps I could increase the sense resistor to 20m Ohm, do you think this would help?

My signal bandwidth is up to 20 kHz.

Regards, Niels

Mitch,

INA229 is a very interesting device, but currently not available.
Do you have more information about the availability of this device?

Due to the datasheet, INA229 can run at 20 ksps, but this reduces ENOB to 9.9.
That's not that much for a 20 Bit ADC. ;-)
(I know that the ENOB value includes the amplifier noise.)

I was looking for a solution sampling at app. 100 ksps. (Actual idea was to use the ADS8699 for analog-to-digital conversion.)
20 ksps may be sufficient, have to check this with my customer. But ENOB should be better than 9.9 at 20 ksps.

Regards, Niels

Hello Niels,

I have sent you a friend request so that I can send you private messages to discuss availability of the INA229. Let me know if 20ksps ends up being enough for your customer. We do have parts that sample faster if needed, but may or may not be what you're looking for.  Do you prefer digital or analog output?

Hello Niels,

One other option you have would be to use the INA190 (analog output), and use your own ADC.

Hi Niels,

every increase of shunt will help :-)

Can you have more than one current measuring range and more than one shunt?

Kai

Hi Kai,

switching between measuring ranges is something that should be avoided.

If it is possible to implement this without increased costs, this may be an option.
(As the digital interface has to cross an isolation, the number of digital signals also affects overall cost.)

Maybe I could increase the shunt resistor to 40mR.  But this will increase the power loss to a full 1 Watt, only for measuring the current. Quite a lot, isn't it?

Niels

Hello Niels,

One more side note, the ENOB you mentioned above for the INA229 at 20ksps is when using the ±40.96 mV range.  If you use ±163.84 mV range, then you get 12.8 (with the fastest conversion time and no averaging).

Hi Niels,

whether 1W heat dissipation is much, depends on the application. But 1W is not really that unusual for a shunt.

What is the precision you need? 40µA?

Kai

My requirements are to sense a current range of +/-5 A with

• 100 uA resolution
• +/-1 mA accuracy

+/-5 A with 100 ua resolution requires an ADC with more than 16 bits.
For the accuracy, 14 "noise free" bits would be sufficient.

I intended to use the ADS8699 as ADC. It has 18 bits resolution. In addition, SINAD and Histograms indicate that my accuracy requirements could be achieved.
Due to its 100 ksps, I could do some averaging to further reduce the noise.
(In addition, I already use it to measure a voltage up to 10V on the same board.)

As it has a "de facto" single-ended bipolar analog input, I thought of using an instrumentation amplifier for the current to voltage conversion.
+12 V and -12 V (as well as 5 V) supplies are available.

For current-to-voltage conversion, I thought of using the the INA214.
The idea is to power INA214 from my +/-12V supplies and connect the REF pin to the same ground as the IN- of the ADS8699.

The other idea is to use an Instrumentation amplifier like the INA828, INA826 or similar.

Do you agree that INA214 would be the better choise?