OPA 381 stability problem

Hi Sameer,

This likely has to do with the stray capacitance of these digital potentiometers.  Please send a full schematic showing all input and output connections and loads collin@ti.com and we'll get it worked out.

Thanks,
Collin Wells
Precsion Linear Application Engineer

Hello Sameer,

I wasn't sure if I'd be able to reply to the "info@.... " e-mail address you sent me a message from, so I replied here instead.

First let me summarize some of the information you've provided.

1.)  You're first revision of this circuit was on a prototype single-sided PCB.  The photodiode was connected to the PCB through a shielded cable.  In this original configuration you experienced oscillations.  The oscillations were fixed by placing a 0.1uF capacitor across the feedback network but the signal settling time increased to an unacceptable level.

2.)  To make it easier to separate issues from each other you decided to spin another board with only the photodiode circuit (not ADC or uC) and at the same time included the MCP potentiometer in the feedback network.  Oscillations still remain in this design which is why you contacted us.

Please look over the above information and make sure it's correct.  Then please help answer a few of my questions and read through the comments I provided.

A.)  If you look at the "phase-margin vs. load capacitance" graph in the OPA381 datasheet you will see that with a 50 Ohm series resistor, a 330pF capacitor will degrade the phase-margin to less than 30 degrees.  With the 10 Ohm resistor you have in your circuit I expect your phase-margin to be somewhere between the 50Ohm and 0 Ohm case which means your circuit will have at most 25 degrees of phase-margin.  This means that due to the output network alone of 10 Ohms and 330pF your circuit is basically unstable already and will have large overshoot and ringing when a transient event is applied to the inputs.  This is before we even begin to consider the effects of the input and feedback network parasitics that I'll also point out.  Either reduce the capacitor or increase the resistor value of your ADC anti-aliasing circuit to help fix this issue. 

B.)  How long is the shielded cable that connects the photodiode to the circuit?  Shielded cables can have a large parasitic capacitance to GND.  In your circuit this shield capacitance will be in parallel with the photodiode capacitance and the total capacitance will appear directly at the summing node ( IN- ) of the amplifier circuit which will cause stability issues.  Can the shielded cable be removed by placing the PCB at the same location as the photodiode?  The OPA output is less sensitive than the input so if you have to choose to run one cable longer than the other, choose the output.  This is not to say that long cables can't cause issues on the output as well but it's less sensitive than the inverting input.

C.)  I looked at the datasheet for the MCP potentiometer and found that the PA and PB terminals each have 75pF of stray capacitance and the PW wiper terminal has another 125pF of stray capacitance.  These capacitors in combination with the potentiometer resistance will form multiple delay elements in your feedback network which causes a reduction in the phase-margin of your circuit.  Placing the capacitor across the entire feedback network helps with this effect but only contributes a single pole to the feedback network which may not be enough to cancel the effects of the multiple delay elements from the potentiometer and the cable shield on the input network.  It may not be practical to accomplish this type of circuit with an external potentiometer due to the parasitic elements that come along with their use. 

D.)  To answer you question about the feedback capacitor, unless you're working with very small currents  (<1pA) where ceramic capacitor leakage properties may cause issues then a standard multi-layer ceramic capacitor will work fine for the feedback capacitor.  These capacitors can be found in very small values (>1pF).

In summary, this type of circuit may not be possible due to the parasitic elements involved with these digital potentiometers.  We need a little more information about the shielded cable you use but this can also be adding to the stability issues.  Finally the ADC anti-aliasing filter is not sized properly which is also causing stability issues.  Try to follow the advice provided in this post and see if you can come up with a working solution.