OPA859: TIA with Gain switching

Hi Anil,

as I already said in the "original" thread, mounting too many components in the feedback loop of an 1.8GHz OPAmp is highly critical because their parasitic inductances and capacitances can destabilize the OPAmp. Heavily increased noise can mean oscillation due to just such an instability.

Instability can also occur due to the junction capaciatances of analog switches, which appear to be installed in parallel to the detector capacitance.

You can try to add a small phase lead capacitance directly across the -input and output of OPAmp, not only across the gain setting resistors. If adding such a small cap will resuslt in a decrease of noise, you may have cured this instability.

You could also mount small resistances in series to the analog switches. I mean the connections to the -input of OPAmp. These small resistances can help to isolate the junction capacitances of analog switches from the -input of OPAmp.

But the best idea is to swap the gain setting resistors and the analolg switches. By this the -input of OPAmp would not directly see the junction capacitances of analog switches.

Another hint is to mount the highest gain setting resistor permanently in the feedback loop and to only switch the three remaining gain setting resistors.

Kai

Hi Anil,

  That is a very wide range, and would benefit from multiple gain options. I agree with Kai. A high speed TIA is sensitive to parasitic, and mux/switches will especially introduce parasitic inductances and capacitances to the feedback of the amplifier. The idea could work with careful layout and specific tuning to the input/feedback components, but would be simpler to use an integrated gain switching TIA or cascading a TIA and PGA/VGA.

  Below is a simplified mux model as a reference. Also, here is a very helpful gain-switching TIA explanation with a link to the reference design and simulation examples. 

  We have the LMH32401, OPA3S328, and OPA3S2859-EP. LMH32401 has a 2kOhm and 20kOhm setting which will be too high of a gain selection for your application, but OPA3S328 and OPA3S2859-EP would work. 

  Below is for 1kOhm gain selection at 65 degree phase margin or in other words, a Butterworth response Q=0.7. Calculator can be found at this link. 

  You could tune the feedback capacitance to extend the system bandwidth if your application can handle some peaking in the frequency response (45-50 degrees of phase margin, but not maximally flat). 

Thank you,
Sima

Hi Anil,

  Thank you for accepting the friend request. I sent over the additional data.

Thank you,
Sima