PGA281 oscillates with 100nf input filter

Hello Randy,

The PGA281 data sheet specifies a typical capacitive load drive of 500 pF. If you connect a larger capacitor than that to the output, stabilization with isolation resistors will be required. 

I recommend that you watch our new TI Precision Labs video series on op amp stability to better understand this issue. Specifically, Stability 5 provides a method for stabilizing op amps with an isolation resistor in series with the output.

A SPICE model is available for the PGA281 as well, which allows you to easily test for stability and functionality in TINA-TI.

Best regards,

Ian Williams
Linear Applications Engineer
Precision Analog - Op Amps

Thank you Ian, your reference to Stability 5 was good and I have broken the loop many times on both oscillators and amplifiers. The problem I am having I believe I can isolate on the input side. The evaluation board ; www.ti.com/.../sbou130.pdf shows a filter on the high impedance input. When I increased the filter C it oscillated. I assume that I can isolate the C with 50 ohms in series. Sorry if I made you think I was talking about the output. The output is fine, just the input. Thanks again, I am looking forward to your post. -Randy

I agree that it should be modeled. I have used TINA in the past and will try it again. Thank you, you answered my question.

Hi Ian,

Yes I may try Tina soon. I am using Ltspice and made up an instrumentation amplifier for my simulations. I may need Tina for the correct model to break the loop and check stability.

It is true that the capacitive loads are to 500pf max. I am driving a differential pair that is just resistive so I am thinking the output side is OK. 

The input impedance is ~1G and the input capacitance spec is 12pf. I am thinking my problem of oscillation is on the input side. I am using the eval-board(s). One is 0.100 Rsense and the other is 0.010 for Rsense. The evaluation board has both an input filter and output filter designed in. 

When I tried to set the input filter to 20kHz, it oscillated with both increased C and increased R. I tried them individually, one at a time thinking the input may not like to have more C hanging on the differential input(s). I may try to isolate the C by adding 50 ohms to both input.

I will be looking at the front end tomorrow and do need help. Is there any TI person that I can speak with by phone? Is there a TI FAE for Honeywell in Redmond WA? I will check My-TI tomorrow.  Thank you,

Randy

Notes on PGA.doc

I am not sure if this attached correctly. It is a file that shows input filters with 75ohms and not 0 ohms. 

Perhaps this is one of the problems. Strange that the documentation goes into great lengths to describe the filter cutoff freq. I am a little confused and assume that I need to let R=0  ??

Randy

Another recent piece of data I am looking at, shows a sine sweep of current to be flat to 100kHz and then peaks 5X(of 1mA) at 5MHz. This is way beyond the bandwidth of the PGA. I have 2 other amplifiers downstream that can be swept without any issues. This is on the Eval-board. I will be checking this today since a junior engineer did the sweep.
The peaking tells me that again I am slipping into instability, however this is beyond the bandwidth, perhaps it is EMI, or other 5kHz sampling folding back into my bandwidth. ....just a few thoughts before I jump into the lab.....
Randy