I have to use a pulse waveform (0-2.5V) as the input signal (signal band: 1KHz) of sigma delta modulator. An analog LPF following input before modulator is used for anti-aliasing. The modulator is just a normal 2-order one.

The frequency of the input pulse is 1MHz and I can only change the duty-cycle of the pulse (10%-90%). The purpose for the pulse input is to check how the modulator response to the DC input. (Unfortunately I can not use DC directly in this case )
The LPF is two-pole and w3dB=10KHz. The sampling frequency of modulator is 1MHz.

Now I am running some tran simulation (say use 50% for duty cycle -> ideal DC=1.25V) and find the output signal of the LPF is not ideal DC and has some very small ripples. This is understandable since the LPF is not idea.

1) The LPF out signal with ripples is applied to the modulator, so the real INPUT signal is DC(wanted) + ripples. If the DC level is very small, will the ripple shade the DC and modulator can not detect the real input so that cause some deadband on output if I do DC sweep???

2) Since the ripple is generated by input (the sharp edge includes high-frequency components), does the ripple can be regarded as (white?...sth else) "noise" and folded into signal band (1KHz) and reduce SNR?

3) Based on my understanding, noise should be something not related to any input signal. If it's not noise, it also can not be "signal" since only DC is studied here. Will this ripples can be shaped by modulator and not a bad issue?

3) Does anyone can provide some document or papers about deadband of modulator?

Thank you very much!


xue