Attila Kinali

On Sat, 15 Dec 2018 19:02:30 +0200 Anders Wallin <anders.e.e.wallin@gmail.com> wrote: > That got rid of the self-oscillation - but now I am left with severe (16 dB > @ 230 MHz) gain-peaking! (see attached figure) Could it be, that the stray capacitance to ground of the non-inverted input is now too big? You increased the feedback resistors by factor 5, so effect of the stray capacitance also increased roughly by a factor of 5. Attila Kinali -- <JaberWorky> The bad part of Zurich is where the degenerates throw DARK chocolate at you.

Am 18.12.18 um 20:46 schrieb Attila Kinali: > On Sat, 15 Dec 2018 19:02:30 +0200 > Anders Wallin <anders.e.e.wallin@gmail.com> wrote: > >> That got rid of the self-oscillation - but now I am left with severe (16 dB >> @ 230 MHz) gain-peaking! (see attached figure) > Could it be, that the stray capacitance to ground of the non-inverted > input is now too big? You increased the feedback resistors by factor 5, > so effect of the stray capacitance also increased roughly by a factor of 5. That amount of peaking is a sure sign that the amplifier is still very, very close to oscillating. John's dead bug construction with the inputs in mid-air is probably the only way to make the SO-8 chip work at all at this low gain. And still it's not comfortable. cheers, Gerhard

I'd suggest that if you wish to use a DIP or SO-8 version of a single CFA, and attain maximum BW and flatness, on a regular board, it should be mounted on the ground plane side, and the feedback resistor should be underneath on the opposite surface, directly between the output and inverting input pins, (through small vias, in the case of an SO-8 package). Also, the bypass capacitors should be on the ground plane side, adjacent the power pins. Dual CFAs are a different story, due to different pinout - the same principles apply, but crosstalk becomes an issue. If you desire to keep the feedback network R as low as possible, the inductance of the runs going around the IC on the same plane will become a bigger factor, possibly causing the observed peaking. So, it's best to keep it a short trip, either underneath, or even tacked on top of the IC. Likewise, the shunt resistor to ground from the inverting input should go straight to ground plane, right next to the non-inverting input termination resistor, so they both see nearly the same ground point. Ed

Gerhard Hoffmann writes: > Have you seen the new THS3491 ? LMH6702 on steroids. > > This one begs to be designed into new distribution amplifiers: > > Everything from DC / 1pps to a few 100 MHz into 50 Ohms, without any > changes. While everybody is looking for opamps with even higher GBW figures… I'd think that for driving significant power into 50Ohms VDSL line drivers might actually be a better fit for output frequencies below 100MHz. Regards, Achim. -- +<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+ SD adaptations for KORG EX-800 and Poly-800MkII V0.9: http://Synth.Stromeko.net/Downloads.html#KorgSDada