Original question:
Hello again!
By simulating the INA818, meeting the required Vcm level, I get nonlinear zones at the ends.
It also occurs before the +-23.5mV differential voltage limit at the INA input.
What could be the reasons for that?
Any way to quantify it? (or to prevent it)
INA818_INA118_ReferenceDesign-TA_SIM.TSC
Thank you!
Hi SrgiO,
Thank you for your post, and for sharing the schematic with me. The INA818 cannot swing all the way to the rail. In the datasheet on page 7, you will see the voltage output swing specified for 0.15V from either rail. That is why your output is limited between approximately 0.15V to 4.85V.
If you need 0 to 5V on the output, you will need to increase your INA818 supplies to handle this.
Hope that helps.
INA818_INA118_ReferenceDesign-TA_SIM EDIT.TSC
Thank you very much for the quick answer.
It is indeed a way to solve it, but it is a design condition to set it to 5 V.
I already had the consideration about the Voltage swing, it doesn't influence me. What I really mean in my question is why non-linear zones occur before this voltage swing level.
Although it cannot be avoided by supplying the circuit with higher voltage, at least know when the linear zone ends (an equation or percentage for example, to apply to other gains).
Thank you very much again, you are helping me a lot.
Hi SrgiO,
the first kink occurs when the input voltage of internal differential amplifier leaves the linear common mode input voltage range, which is about 2V away from the supply rails, in your case at +2V and +3V:
The second kink arises when the output stage of internal differential amplifier goes into saturation.
Kai
Very accurate information, thanks.
So, for linearity, the useful output range goes from (0.15 - 4.85)V to (0.5 - 4.5)V.
Therefore, for a gain of 100, the maximum input range to stay in the linear range is not (±23.5mV), but rather, for power at 5V, a range of ±20mV.
This also, I imagine, will occur at any gain.
Correct me if I'm wrong.
Thank you very much, I love having such instructive talks.
Hi SrgiO,
Thank you for being patient with me while I looked at the internal nodes of this device. It looks like on either end, you are saturating the output of the INA (that is the output of the diff amp stage). Your input range is perfectly fine, the issue is the output swing is slammed. That is why you see the "kink".
You are changing R4 to be 115-ohm to 125-ohm. This correlates to the following input differential voltage range: -31.78mV to 30.74mV. When R4 is, let's say, 116.2-ohm, that correlates to approximately -24mV. Vout = Gain * Vd + Ref = 100 * -24m + 2.5 = 100mV. This is outside of the linear output swing of the INA. The linear output swing of the INA on a 5V single supply is 0.15V to 4.85V
Let us summarize your circuit:
According to the VCM vs. Vout boundary point plot, you are only able to process input voltages of -23.5mV to 23.5mV. That is why you see a kink in the output curve. You are pushing the device beyond it's specified limits so we cannot predict it's behavior. I have attached a photo of the VCM vs Vout graph below.
If you want to process this Vd range and keep a 5V supply, you will need to adjust your gain and reference voltage to make sure you are not violating the input / output range.
Thank you very much for the reply Tamara,
I understand everything you say, it is very well explained.
The problem comes when the simulation of the device shows that that VCM vs Vout graph range is just the "active zone" and does NOT match the "linear zone". a: Vd, b: Vout (when it's become linear zone)
The only way to find this "linear zone" is by considering what Kai said, to see when "internally" it is in the linear zone.
The case is, if the program shows the VCM vs Vout graph for the active zone and not the linear one. Or if the simulation model is otherwise incorrect.
Thank you very much, it is very enriching to have contributions as good as yours.
Cheers!
