Oscillation with AD8137


I have an AD8085 driving a AD8137 with gain 10 with a roll-off at about 1 MHz:

The board is FR4, two layer with 1.6 mm thickness. Bottom layer is mostly ground with lots of ground vias. Running the circuit for some time, an oscillation with about 20 mV and about 100 MHz appears, observed with a probe on DO+. Touching e.g. DO- with an isolated  screwdriver or switching off and on again the supply removes the oscillation , to appear again at random. Copper from the ground plane below the summing nodes is not removed, but stray capacitance should be small, as overall area of summing nodes is small, FR4 thickness high and C41/C42 form are the "small capacitors across the feedback resistors" as in the datasheet "Layout considerations".

Any hints for me?

  • Hello,

    when the oscillation happens, it is clearly visible on the AD8137 side of the 1k at the input and not visible on the AD8065 output side. The 1k/10K combination is the one recommended in the AD8137 datasheet. Do you really recommend to go to lower impedance? AD8137 is fast, but not really  fast as the AD8139.

  • > It  sounds like he has the feedback caps for compensation rather than  filtering,            

    > so I would recommend that he remove the capacitors and see  what happens.

    I need component defined low pass properity. So Cf is for filtering and compensation is only secondary effect. With a possible "low-level oscillations with large Cf" as tertiary effect. Argh. Thanks for the openness to tell.

    Any other part recommendation for single ended to differential conversion with G=~10, component determined fg = ~ 1Mhz, Vsupply = +/- 5V and not to low effective input resistance to not load the AD8065 too much?

  • Thanks Harry,

    I forgot the obvious.

    Should I keep Cf with about 1pF to (over)compensate C(stray) at the summing nodes? I guess this is not a "feedback capacitor value is too large"?