# Modelling differential amplifier input circuit

I'm trying to understand the behavior of a filter connected between a mixer and fully-differential amplifier. Here are the relevant portions of the schematic. The first part is the mixer output

5VF is 5V. And here's the input to the FDA.

The amplifier is the ADA4940. I was able to track down a spice model for the ADA4940, but unfortunately I'm having trouble getting Ngspice to accept it. So, instead, I'm trying to design an input circuit that looks roughly equivalent from the perspective of the filter. Here's what I've come up with:

The voltage source is the mixer output, then everything up through the 10n caps is part of the filter or mixer output bias and then everything to the right of that is meant to simulate the amp input. I've used $$\50M\Omega\$$ resistors for the dividers because this is the reported common mode input resistance from the data sheet (I guess theoretically since the resistors are in parallel, the 50Mohm should be the combination of both in parallel, not each individually, but this value doesn't seem to affect the simulation much). The dividers also set the common mode input voltage to half the supply voltage. The data sheet also says to use twice a single gain resistance for the differential impedance, hence the $$\2\times 549\$$ value.

When simulated I get the gain and phase response shown here, which is more or less what I'd expect given the desired pass frequency range (about 30kHz to 1MHz).

and zoomed into the passband

Have I correctly described the diff amp input circuitry for this simulation?

Separately, the ADL5802 mixer datasheet says to use pull-up chokes, but this design uses pull-up resistors. I tried to simulate this with inductors instead of resistors but couldn't get the simulation to work. Why were resistors chosen here in place of inductors? Is there any downside to using the resistors? Why would the datasheet insist on inductors when resistors seem to work fine?

## 1 Answer

If you look at the schematic, the input is differential, this means there are actually two inputs w.r.t. ground, in antiphase (as is the case here). The same for the output, because if you look at its datasheet, the FB pins are the same as OUT.

Soo you have a simple LC lowpass, followed by a differential active filter, which is a simple 2nd order bandpass. The LC filter is not loaded with any resistor, so you would expect the whole result to be an undamped (to a degree) lowpass combined with a 20dB/dec bandpass.

It looks like what you plotted is similar, but the way you have drawn the equivalent schematic does not reflect the whole truth. Here is an attempt with LTspice. I didn't use the model for 4940, but I used another differential opamp, only for the show:

To the right is the same version, but single ended, and the result is the samem save the amplification due to the bandpass being an active filter, as opposed to a passive one.