INA326EA instability

Hello Vlad,

Thank you for considering the INA326.

Let's start with a few questions concerning the circuit and troubleshooting:

a)  What is the relative location between the gain-setting resistor and the INA326?  Could there be parasitic capacitance at pins 1 and 8?

b)  Are you able to readily re-create this issue?

c)  On boards that exhibit this behavior, have you replaced the INA326 with one from a working board?  If so, what was the result?  Another way to phrase this question is:  Does the instability follow the devices or the boards?

d)  If you can re-create the issue, have you been able to capture the input waveforms?

Thanks!

Hello Pete,

Thank you for the swift response.

Here are the answers:

a)  The distance between INA326 and first stage gain setting resistor is ~3mm

b) Yes. The problem is consistently reproducible on selected boards. I have 3 boards rejected from production with that problem

c) It is very hard to move the chip from one board to another since the boards are conformally coated. The analysis of failures in production shows that some chips are more prone to oscillation than others

d)  I have attached the input waveforms during power-up. Keep in mind that if a scope probe is attached to pin 8 the oscillation never occurs.

I've got an additional information about this issue. It seems INA326 become unstable during power-up if some voltage is present on the IN+ (pin 3). In my case there is ~0.2V residual voltage on the input capacitor during power-up. Since I have introduced the bleeding resistor 1M across 1uF capacitor on the input the problem vanished. Simple removal of the cap does not work because of racing condition between INA326 power and input voltage.

Any comments?

Thanks,

Vlad

Hi Vlad,

I am entering this discussion as Pete must take leave to tend to a family matter. I have followed this thread with interest and have conferred with Pete.

We have experienced sensitivity and tendency to oscillate due to excessive capacitance on the gain set resistor pins 1 and 8. This was the origin of Pete's questions about the connections to these pins. We are suspicious that this is the fundamental issue but we will not rule out other possible causes. Typically, somewhere around 10pF on these nodes can cause issues.

The additional information you have provided on the power-up and signal conditions is interesting but not necessarily inconsistent with the suspected cause. This is a marginal situation--only some boards do it. In a marginal cases there will always be signal and supply conditions that can trigger the oscillation.

You have relatively short connections to the gain set resistors, that's good. Do you have a ground plane under these connections or under the resistor? This could increase the capacitance on the gain-set nodes, especially if a ground plane is on an internal layer of a multilayer board. The conformal coating will also increase capacitance somewhat. Though I would not expect this to be a large effect in itself, there may be several small factors that combine to create the problem. I suggest that, if possible, you temporarily break the connections to the existing gain-set resistor by lifting these pins from the circuit board. Then tack in an external resistor that is not subject to any capacitive effects on the board and see how the circuit behaves. Also consider any other possible layout factors that could contribute to stray capacitance. Are there any other signal-carrying nodes nearby that could be capacitively coupled to these nodes?

Regards, Bruce.

Hi Bruce,

Thank you for the reply.

I did what you suggested to do. I lifted pins 1,8 and soldered the gain-set resistor on top of the chip. It did not help, the oscillation persisted with the same occurrence rate.

What actually helped dramatically is the change of the power rising time. The original rising time was ~ 500us as shown in the previous post. After change to 6ms the oscillation did not occurred on the boards where oscillation was consistently reproducible.

I am going to propagate the change into production.

Regards,

Vlad

Vlad,

Thanks for the feedback on this issue. I'm pleased that you found a solution. It makes me wonder if the particular circuit conditions of your application make the INA326 sensitive to power supply rise time.

Anyway, I will tag this thread appropriately so that we can be alert to other possible instances of this behavior.

Thanks and regards, Bruce.