INA821: Large DC offset when REF < 1.53 V

Hi Helge,

Helge Langen said:

The input signal is at Vcm = 2.5 V, Vdiff up to 12 mV, frequency ranging from 1 kHz to 1 MHz. We want the output signal to swing around 1.0 V.

OPA821 is unable to support 1MHz differential input signal at Gains = 8.35 V/V. 

If the differential input is required to be 1MHz, you may have to consider INA849. I would recommend to combine the gains of 8.36 and 2 in one as 16.7V/V in OPA849 IA.  

Helge Langen said:

Raising the reference voltage (for both the INA821 and the OPA835) to 1.53 V or higher makes the circuit operate as expected.

The minimum output swing is shown in the image below. INA821 is a 36V IA and you are operating at the minimum required single voltage rail. The OPA821's output has (V-) + 0.15V requirement in addition to Vref voltage. So you are not able to configure Vout at 1V without dual supply voltage rails. In the simulation below, I set Vee = -0.5Vdc (or lower) and it will work as you desired (except it can only work at approx. 400-500kHz range without output attenuation). 

INA821 Offset Issue 04242024.TSC

If you have additional questions, please let me know. 

Best,

Raymond

Thank you for your reply. Yes, I am aware that the output cannot swing below 0.15V above the negative supply, but 1.0V +/-100 mV should be well above 0.15V, so that still does not explain why the reference needs to be as high as 1.53V to work properly - where does this additional 1.38V margin requirement come from?

Hi Helge,

Helge Langen said:

where does this additional 1.38V margin requirement come from?

The output voltage swing rating (V-) +  0.15 is characterized with ±15Vdc dual supply rails at gain of 1. The output swing rating may be affected by different supply rails, especially that is close to the minimum single supply rail at 5V. 

This IA is tailored for 36V or  ±18Vdc nominal operation. The minimum single supply rail can power INA821 is 4.5V and the output voltage swing specification is affected by the single supply voltage under the configuration including 8.35V/V gains effect. In addition, this is 3X op amp IA and I do not have individual op amp characterization, and unable to calculate the typical figure.

The simulated Vout is approx. 1.5V range, and Vref = 1.0V, so the actual difference is only approx. 0.5V from GND, not 1.38V when powered by 5Vdc.  

If you have other questions, please let us know. 

Best,

Raymond

ok, it's strange that this parameter changes so dramatically (a factor of 10x) across the supply range, at least this should have been indicated in the datasheet somehow.

Regardless, I tried running a spice simulation where I set the supply to 30V single-sided, and it behaves exactly the same, in that the reference voltage needs to be above 1.518V (to be precise) in order to operate correctly, so I don't think running it at 5V is the problem.

Additionally, I tried setting the reference voltage close to the positive rail, and experienced the same behavior, in that the reference needs to be less than 1.5 V below the positive rail to function properly.

Does this mean that the allowed range of the REF pin is (V-)+1.52V to (V+)-1.5V? If so the datasheet should be updated to indicate this.

Hi Helge, 

Helge Langen said:

Regardless, I tried running a spice simulation where I set the supply to 30V single-sided, and it behaves exactly the same, in that the reference voltage needs to be above 1.518V (to be precise) in order to operate correctly, so I don't think running it at 5V is the problem.

I simulated what you suggested, and it looks fine with the circuit. 

It is probably the Vcm issues when the Vs is powered at 5Vdc, which is 0.5Vdc above the minimum specification. 

Best,

Raymond

ok, I was able to reproduce that, but still, the datasheet does not in any way indicate that these strange behaviors are expected when operating well within the required supply voltage range. Is this an error with the amplifier, or with the datasheet?

Hi Helge, 

Helge Langen said:

when operating well within the required supply voltage range. Is this an error with the amplifier, or with the datasheet?

I would not characterize as "operating well within the required supply voltage range".  The datasheet is specified at ±15Vdc and Vout swings are ranged within the figures shown below. 

Indeed, your INA821 configuration is operating near the low end of the 36V instrumentation amplifier; not all the performance is captured in the datasheet. I would not use OPA821, if your application has only 5Vdc available as the supply rail. I would suggest to select low voltage instrumentation amplifiers, so that  you are able to maximize the IA's performance. We have tons of different low voltage IAs to select from,  see the image and link below,

https://www.ti.com/amplifier-circuit/instrumentation/products.html?keyMatch=instrumentation%20amplifier%20selection#358max=5.5%3B5.5&

I hope that I answered your questions. If you have additional questions, please let me know. 

Best,

Raymond