Other Parts Discussed in Thread: INA4230, INA4235
Tool/software:
In our implementation, referring to Figure 26 of INA3221.pdf, March, 2016 edition, “INA3221 with Input Filtering”:
- The value of Rfilter is 100 ohms, 0.1% tolerance, for each resistor <- while this value is 10x the recommended value, I suspect it does not have a great impact on our operation, your feedback would be greatly appreciated.
- Cfilter is 0.1uF, 10% tolerance
- Rload = 0.025ohm, 1.0% tolerance. This is the shunt resistor
- We also have a 1uF, 10% tolerance, capacitor from Vin+3 to GND, as well as from Vin-3 to GND <- This is my area of concern
The power supply is 3.3V, with a ramp rate of 5000V/s from 0v to 3.3v during startup.
During power up, the inrush current measured by the INA3221 across Rload exceeds the threshold of 700mA. Different CCAs (circuit card assemblies) perform differently, some fail, others don’t fail.
We theorize that a difference in capacitance between the two 1uF caps from VIN+/- to GND creates a brief voltage differential between Vin+ and Vin- which is detected as exceeding the critical limits set within the INA3221.
This circuit was originally implemented using a component different from the INA3221, that circuit required the 1uF caps to GND in the filter network. The capacitors to GND were not eliminated when the change to the INA3221 was implemented.
What can you tell us about the INA3221 sensitivity to caps on Vin+/- and how that might create a false overcurrent failure. Specifically related to the capacitive balance between the cap on Vin+ as compared to the cap on Vin-?
As you can see, the Rfilter components are the wrong value as well. I suspect this doesn’t influence the operation, but if you can comment on this as well, I would be appreciative.
We are still trying to confirm root cause, your input is appreciated.
