INA219: Bus Voltage Accuracy

Oren Ya'acobi

Part Number: INA219
Other Parts Discussed in Thread: , INA226, INA230, INA233, INA3221

Hi,

We are using the INA219EVM Evaluation module.

When connecting the INA219 Vbus to 1.5V alkaline battery (with no load) we are seeing 18mV difference between the Vbus read of the INA219 and 2 other DMMs.

We have connect the battery between Vin- and GND of the INA219.

We are also seeing this difference when connecting the INA219 to our platform rail.

Please see below the graph and the SW setting.

Please advice: what may be the reason for this difference?

Thanks,

Oren Y.

over 6 years ago

0 kai klaas69 over 6 years ago

Guru 140200 points

Hi User,

input impedance of INA219 at IN- input is 320k. This is much lower than the input impedance of a DMM. So, the INA219 is loading the battery more than a DMM which results in a lower battery voltage.

Kai

0 Oren Ya'acobi over 6 years ago in reply to kai klaas69

Prodigy 105 points

Thanks Kai,

We would appreciate your advice on the following regarding the battery measuremnts:

1. Is there a way to use the INA219 and to get the same voltage read as the DMMs?

2. Is there any calculation which we can use in order to calculate the voltage drop due the INA219 low input impedance?

so we can use this calculation and add it to the actual read of INA219.

Thanks,

Oren Y.

0 kai klaas69 over 6 years ago in reply to Oren Ya'acobi

Guru 140200 points

Hi User,

take also into consideration the DC accuracy of ADC and the input bias current of INA219 of 20µA. How precise must the battery voltage be measured? Is the input leakage current allowed with the battery?

Kai

0 Oren Ya'acobi over 6 years ago in reply to kai klaas69

Prodigy 105 points

Thanks Kai,

The precision of the battery voltage measurement can be +/-1LSB (+/-4mV).

Yes, the input leakage of 20uA current is allowed.

We add our INA219 test board setup to the following diagram:

1. Is there a calculation which we can use in order to calculate the 18mV drop?

2. Please advice how does the INA219 sees this 18mV drop due the 320Kohm input impedance?

Thanks,

Oren Y.

0 kai klaas69 over 6 years ago in reply to Oren Ya'acobi

Guru 140200 points

Hi User,

the 18mV result from battery chemics, which is very complex.

During the test are the IN+ and IN- inputs connected to each other? The In+ input should not float.

Kai

0 Oren Ya'acobi over 6 years ago in reply to kai klaas69

Prodigy 105 points

Hi,

Please note we are seeing the 18mV difference also when we measure our platform rails which are not powered from a battery (see graph below).

Our platform rail connect to a buck converter.

When we measured our platform rail the IN+ was connected in order to measure the current (IN+ was not floating)

We have short circuit the IN+ to the IN- at the battery setup and still we get the same results.

We are evaluating the INA219 because currently we are using DMM in order to measure our platform rails.

So we need to understand this 18mV difference.

1. Is there a calculation which we can use in order to calculate the 18mV drop?

2. Please advice how does the INA219 sees this 18mV drop due the 320Kohm input impedance?

0 Oren Ya'acobi over 6 years ago in reply to Oren Ya'acobi

Prodigy 105 points

Hi,

Due to testing 0-4V voltage range, we used the 16V FSR bus voltage range.

We have found that when we select 32V FSR bus voltage range - we don't see the 18mV difference, and the results are as expected - see below graph.

Please advice why selecting the 32V range improve the results?

We expected that selecting the 16V range will give better accuracy than the 32V range.

Thanks,

Oren Y.

0 Peter Iliya over 6 years ago in reply to Oren Ya'acobi

TI__Mastermind 29825 points

Hello Oren,

Could you try increasing the conversion time of your bus voltage measurements? Maybe there is a settling issue when using the fastest conversion time. Also it is good practice to keep the input pins shorted very close to each other, either with your shunt resistor or with wire. Keep in mind that if you have a shunt resistance in between IN+ and IN-, the platform rail connected to IN+, and your load connected to IN-, then VBUS will read {rail voltage - shunt voltage}.

Also check to see if there is not a layout issue here. Could you probe the VBUS pin of INA219 with your DMM and see if this also is 18mV off? Maybe there is some parasitic resistance in between your rail/battery and the VBUS pin?

Finally, ensure the grounds of your platform rail, INA219, and DMMs are all common.

Sincerely,
Peter Iliya

0 Oren Ya'acobi over 6 years ago in reply to Peter Iliya

Prodigy 105 points

Hi Peter,

Thanks for your help.

I have increased the conversion time - but I still see a diffrence when INA219 reads the Vbus voltage when using the 32V FSR and 16FSR of the INA219..

When I measure the battery using the INA219:

When using the 16V FSR - INA219 reads Vbus value: 1.468V

When using the 32V FSR - INA219 reads Vbus value: 1.484V

Using a DMM I have measured the voltage on the IN- pin of the INA219 and the DMM reads: 1.486V

I see that the DMM read on the IN- pin is as the read of the INA219 for the 32V FSR setting, but the 16V FSR is 18mV below the voltage that is on IN- pin.

I am using the INA219EVM board , I have connected the battery between IN- and GND, and I have short circuit IN- and IN+.

Does the 32V FSR setting supposed to be more accurate than the 16V FSR setting?

Please see below the INA219 EVM resistor setup:

Thanks,

Oren Y.

0 Peter Iliya over 6 years ago in reply to Oren Ya'acobi

TI__Mastermind 29825 points

Hey Oren,

I am testing this now and actually see worse accuracy for the 16V FSR compared to the 32V FSR similar to you. Changing the averaging did not affect this. I am not sure why this is but I will try to get you some information as soon as possible.

Sincerely,
Peter Iliya
Current Sensing Applications

0 Peter Iliya over 6 years ago in reply to Peter Iliya

TI__Mastermind 29825 points

Hey Oren,

I have conferred with my team and the behavior you are seeing does make sense.

Firstly, the 18mV error is within spec. In the datasheet we specify a +-0.5% max error at room temperature for the VBUS measurement when VBUS = 12V. This case implies a 60mV offset maximum possible assuming offset error is the dominant source of error. If you reduce the bus voltage to 1.5V, the 60mV offset may not change, and thus the error of a 60mV offset when measuring 1.5V could be 4%. Although this is very unlikely. With Vin+=Vin- = 1.5V, I measured 4mV and 20mV offsets with 32V FSR and 16V FSR settings respectively and this was with only one average. So this explains your 1.2% error seen.

Secondly, the 32V FSR setting switches the ADC to a trimmed resistor, while the 16V FSR setting does not. So if you see less error with the 32V scaling then you should definitely stick with this setting.

Hope this helps and please post back if you have more questions.

Sincerely,
Peter Iliya
Current Sensing Applications

0 Oren Ya'acobi over 6 years ago in reply to Peter Iliya

Prodigy 105 points

Hi Peter,

Thanks for your detailed answer.

Reading the spec. I understand that the spec. values are for Vbus=12V.

From your answer I understand that the INA219 has 60mV offset error for 12V and also for 1.5V.

I thought that the max offset of 0.5% is relative to the input voltage, and not 60mV constant.

For example: the max error for 1.5V input will be +/-0.5% from 1.5V which yield +/-7.5mV.

Please advice.

Thanks,
Oren Y.

0 kai klaas69 over 6 years ago in reply to Oren Ya'acobi

Guru 140200 points

Hi Oren,

you are not the only one who is thinking that 0.5% is meant relative to the actual bus voltage. Datasheet tells that the ADC has a basic resolution of 12bit and a LSB of 4mV. Differential nonlinearity is even specified to +/-0.1LSB! This makes you believe that the ADC is very precise. A constant offset error of 60mV is really disappointing then! The datasheet should make much more clear what sort of accuracy the customer has to expect from this chip!

Kai

0 Peter Iliya over 6 years ago in reply to Oren Ya'acobi

TI__Mastermind 29825 points

Hello Oren,

The INA219 accuracy spec for VBUS is only for when VBUS=12V and this spec incorporates the total error from offset and gain error. You can look to the Figures 5 and 6 to get an idea of how large the offset and gain errors can be. Of course we know now that the 32V FSR will give you the lower offset. For the INA219, you should just always use this option because the LSB is not reduced if you change the FSR to 16V. We realize this is confusing and not explained well in datasheet and we will try to revise it.

Please know that the VBUS offset can be calibrated out since it will be constant over the VBUS range. It also is relatively stable, give or take a couple millivolts, over temperature, which is shown in Figure 5. This is so because the bus voltage is divided down and read directly by the ADC and this ADC will not experience the CMRR effects of the PGA where a changing VCM will change the shunt voltage offset. Thus a changing VBUS will not change the VBUS offset, making it easy to calibrate out. I also saw this in my testing where the VBUS offset was the same at 1.5V and 12V.

If you would rather have a device with more comprehensive specs you can look at the INA226, INA233, INA230, or INA3221. These devices do not have variable-gain front end, but do provide 6-sigma specs for maximum possible VBUS offset and gain error.

I hope this helps clarify your efforts.

Sincerely,

Peter Iliya

Current Sensing Applications, TI

0 Oren Ya'acobi over 6 years ago in reply to Peter Iliya

Prodigy 105 points

Thanks a lot Peter for your detailed help.

Amplifiers

Amplifiers forum

INA219: Bus Voltage Accuracy