INA849: Current noise specification datasheet query

H S1

Part Number: INA849
Other Parts Discussed in Thread: TINA-TI

Tool/software:

Hello,
I am trying to determine an accurate value for the 0.1Hz-10Hz current noise on INA849.
On Page6 of its datasheet this is quoted at 100pApp. However, in Figure 7-28 it looks more like 25pApp - a factor of ~4 less.

As another comparison, looking at the datasheet of AD8421 the equivalent figure (Fig 44) shows that amplifier has an input current noise of very close to 1/4 that of INA849 across the relevant frequencies - and the AD8421 has an 18pApp spec and this looks consistent with the time trace in Fig 45 of that datasheet. In other word, all three measurements on the AD8421 datasheet seem consistent with one another and consistently 4x smaller than the INA849 datasheet values (except Figure 7-28)

Taking this information together - and assuming the calculations and characterisation experiments of both devices are broadly similar - it seems the most likely error is in Fig 7-28 of the INA849 datasheet and in fact this data should be scaled (amplified) by approximately a factor of 4 so that it matches the 100pApp spec in the table. If that were the case all three entries in the INA849 datasheet would then be approximately 4x larger than the corresponding entries in the AD8421 datasheet, and would be consistent with one another.

I should note this is an important question for me as I am trying to decide between INA849 and AD8421 for an application in which I want to minimise noise in 0.1Hz to 100Hz range when amplifying a 1k resistance bridge (strain gauge) with 8V across it by a gain factor of 100 to 200.

Thank you in advance for any advice on either clarifying the datasheet or indeed better amplifier choices for my case.

4 months ago

0 Luis Chioye 4 months ago

TI__Guru 54376 points

Hi H S1,

Before we discuss the current noise density in detail, please note that the INA849 offers lower voltage noise than the device you mentioned above. At Gain of 100V/V, the INA849 voltage noise density is around ~1.5nV/sqrt(Hz) at 1kHz, and the INA849 0.1 to 10-Hz noise is also lower 0.06uVpp. The other device above offers higher noise voltage at ~3.5nV/sqrt(Hz) at 1kHz and 0.1 to 10-Hz noise is 0.09uVpp.

Also, an important point for this particular application, the bridge sensor thermal resistor noise at 1kΩ will likely dominate over the voltage noise and current noise contribution of the instrumentation amplifier, contributing a thermal resistor noise at 4nV/sqrt(Hz). To put things into perspective, the noise contribution due to the the INA849 input current noise density interacting with the bridge sensor at 1kHz is a significantly smaller, 1.1pA/(sqrt(Hz) * 1kΩ = 1.1nV/sqrt(Hz).

The current noise spec on the INA849 table are specified at 1.1pA/sqrt(Hz) at 1kHz with an unbalanced input source impedance Rs on the Electrical Characteristics Table. This is in agreement with plot 7-25 that provides two plots. The curve in black is for the most conservative, using an unbalanced source agreeing with the spec table, and in red, a second curve with the balanced source impedance showing lower current at around 0.8pA/sqrt(Hz).

The noise on Figure 7-28, at 0.1Hz to 10Hz shows the typical current noise bench measured data, notice that the noise is below the conservative 100pA peak-to-peak specification. The Electrical Characteristic Table offers the most conservative spec for peak-to-peak current noise at 100pA peak-to-peak. I will need to seek clarification if the 0.1Hz to 10Hz of figure 7-28 uses an unbalanced or balanced source to obtain the figure 7-28 plot that may explain some of the difference. Also, most instrumentation amplifier devices typically only provide a less conservative spec for balanced input sources, where some of the current noise contribution tends to cancel. Another key concept, noise is a random event, and when performing an oscilloscope measurement in the time domain, such as Figure 7-28, you will likely capture the RMS value or standard deviation, but not necessarily capture the worst case peak-to-peak noise. The probability of measuring noise events close to the center of the distribution is very high, but the probability of measuring noise at the skirts, or peak-to-peak value is relatively low. The spec of peak-to-peak current noise of 100pA peak-to-peak on the Electrical Characteristics table is conservative.

In summary, the noise performance will likely be dominated by the thermal noise of the 1kΩ bridge at 4nV/sqrt(Hz) depending on the bandwidth requirement of the circuit. Also, the application uses a relatively low impedance of 1kΩ, where the instrumentation amplifier current noise plays a relative small role on the overall noise of the circuit (depending on the bandwidth requirement). The INA849 offers the trade-off of lower voltage noise density against the device you have mentioned above. Since this is a bridge sensor, the input source impedance is balanced, where the INA849 current noise spec is the lower (red plot) on figure 7-25.

What is the frequency bandwidth requirement of the application? I can provide an estimate on RMS and peak-to-peak assuming G=100V/V and 1kΩ bridge sensor.

Thank you and Regards,

Luis

0 Luis Chioye 4 months ago in reply to Luis Chioye

TI__Guru 54376 points

HI H S1,

A hand analysis noise calculation based on the INA849 data sheet specifications, assuming G=100V/V, Bridge Sensor = 1kΩ, and BW to 100Hz (ENBW = 157Hz) yields to an input referred noise (RTI) of 0.057μVRMS (0.377μVpp). This corresponds to an output referred noise (RTO) of 5.7μVRMS (37.7μVpp).

The dominant source of noise is the thermal noise of the bridge, the current noise is the smallest contributor.

When using a bridge of 1kΩ impedance, the equivalent resistance (REQ) on each INA input is 500Ω. See figure below.

This analysis does not account for any other extrinsic noise sources to the circuit. TINA-TI noise simulation results are in very close agreement to the hand calculations (simulation results at the bottom).

The noise calculation process are very similar to the equations published in this application note. Although this application note discussed a different programmable gain amplifier, the general equations are the same.

Application Note

Achieve High SNR with the PGA855, Fully Differential Programmable-Gain Amplifier

See the INA849 hand analysis calculations below: