INA190: First stage stability issue

Hey Pengfei,

Would you please explain why the load currents (CH4, blue waveform) look different between INA190 and INA199 tests?

Does the INA190 ever track the load current aside from the output disturbances?

Would you please share what the INA190 output pin is driving in the schematic?

---

Overall you are correct to assume that this behavior could be due to the capacitive 1st stage of INA190. In the INA190 datasheet (section 8.1.3, page 23), it explains that harmonic current transients can cause periodic output disturbances in the INA190 due to the first stage's topology. The best and really only way to fix this is to attenuate the differential transients the INA190 is sensing. This is done by placing a capacitor (>5nF) across the inputs. Right now the customer only has a 100pF. If this doesn't attenuate the signal enough, then populate R425 and R428 with resistances > 10-Ohms. This should further drop the cutoff frequency of the input differential filter and thus further attenuate the current transients.

Additionally, I should note that an output RC filter on the INA190 OUT pin can make the disturbance worse if the cut-off frequency is especially low. This is so because if there is an output disturbance, a large output cap would spread this energy out over time, thus lengthening the output disturbance.

Hope this helps,

Peter Iliya

Hey Pengfei,

Increase C422 to greater than or equal to 10nF as good starting point. Repeat the test with R425 and R428 as 0-Ohm and as 100-Ohm. You can increase these resistors up to 1kOhm with minimum effect on the DC accuracy to maximize the filtering of high frequency current noise. The only trade-off to decreasing fc of input filters is that it will slow down system current step responses, which might or might not be needed for this design. 

We do have SPICE models for INA190 using the TINA-TI software. This model is a great tool to determine DC, transient, and output driving operation, as well as, determining system output stability (e.g., INA190 driving a capacitive load and how to compensate for this). I have included a simulation file that should represent customer's circuit to the best of my knowledge.INA190Schematic.tsc

http://www.ti.com/product/INA190/toolssoftware

http://www.ti.com/tool/TINA-TI

Please note that the output disturbance behavior seen in this post is not modeled and cannot be debugged or analyzed using the SPICE model.

Sincerely,

Peter Iliya

Hey Pengfei,

When actually testing hardware you will need the 10nF input cap. This cap will be more important than the input resistors since there seems to be high-frequency current transients in the real system. The TINA simulation is not simulating this and the INA190 model does not model its chopping behavior. Your conclusion that the input resistors are playing the main role here is potentially incorrect. Of course increasing the input resistors will lower the fc of the input filter and this helps attenuate noise better and smooths steps responses, but decreasing fc too much will slow down system response and may not be necessary as an input capacitor by itself may practically accomplish what is needed to fix the problem.

This is why I said try testing hardware with just the input cap (10nF) first and if this does not work, then start increasing the input resistors.

I cannot disclose the specifics about the first stage other than that it is extremely complex and there are switching capacitors and chopping circuitry to overall eliminate input voltage offset and input bias currents. The first stage's capacitors increases the differential impedance (~4.6MOhm) which lowers in the input bias current and thus allows to measure small currents in the micro-amps. This gives the INA190 very good DC accuracy and low power consumption.

Best,

Peter