Accurate sensing of rectified mains voltage through an ADC

I'm working on sensing the mains input for a power factor-corrected boost converter so I can detect the zero-crossings and also detect the peak voltage. I have placed a voltage divider circuit before the voltage bridge so that the voltage can be stepped down to a range that falls within my ADC input range. An example of what I'm using can be found in the image. The capacitor on the output is 1uF and used for anti-aliasing purposes.

I'm having a problem where the rectified voltage isn't behaving as I would expect. Ideally, I want the zero-crossing of the rectified mains to correspond to the GND potential after the bridge. Instead, I'm seeing the voltage get close to 0 V, but never actually on it. In fact, as the voltage increases (e.g. from 120VAC to 240VAC), the gap from the minimum point on the rectified voltage to the GND of my circuit actually increases.

Now this wouldn't be a problem if the gap was consistent since I could add an offset or compensate it during software, but it's a big problem since it's varying with respect to peak input voltage. Is there any explanation for why the gap is increasing? I also measured at the input to the circuit right where it's plugged into the mains and that gap is still present, so I don't think it would be an EMI filter problem (there's one right after the diode bridge).

If I add a resistive divider (without diodes) for the live and neutral AC mains, the voltage is clamped to GND without any issue. I don't want to do this since it means I'll have to use two ADCs and add the signals together in software. I don't want to isolate this section either since there's isolation farther down the conversion chain (AD/DC to DC/DC).

Any comments or suggestions? Thanks

The capacitor at the ADC input is configured to 'fill in the trench' and cause the problem you report (if that's where you're taking the scope readings). However the scope diagram is not exactly what capacitive filtering normally looks like.

Or did you remove the capacitor and still get the problem?

Possibility:

With a coil (or 2) in the circuit you may be getting counter EMF. Probably from the coil at left. It's hard to be sure.

It may be a net effect of several diodes switching current throught two coils. Again, can't be sure.

What did you plan to be current flow in the bottom left circuit? The two diodes are oriented so neither can conduct during either cycle, except if they conduct down to the ground symbol. It's a match to the ground at right. This ground is created through the bridge rectifier. However during one half of the AC cycle its current goes through the coil at upper left. A prime candidate to create the voltage shift you observe.

Operation might become balanced if you were to add a second identical coil in the other leg coming from your AC source.