Sigma-delta modulator (ADC), can it be used without sinc3 filter?

Question

Purpose: isolated measurement of DC voltage (about 600V). For these purposes, I found two options:

1. ADS1115 (I2C ADC) + I2C digital isolator
2. CA-IS1204 Isolated Sigma-Delta modulator

I am considering the second option. He needs to apply an external clock frequency (From 5 MHz) to the input and at its output there will be a stream of units, the density of which depends linearly on the input voltage.

At the output, they suggest using a sinc3 low-pass filter.

At the moment, I have no way to find an external sinc3 filter, but I'm not sure about using a software filter (since there is no such experience yet).

Questions:

1. Is it possible to work without a sinc3 output filter if I measure direct current where there is no sinusoid? I will change the voltage value using a DC/DC converter (600-200V range). Or without a filter, do not know the actual value of the voltage? What if we sample values ​​and get the average.
2. CLKIN will be at 5MHz, what frequency will the output data be? I don't plan to check the voltage value so often. The datasheet has the 'fdata' value, but it is associated with the sinc3 filter.

P.s. As a microcontroller I use STM32F3

Thank you!

Answer 1

Is it possible to work without a sinc3 output filter

Not really in the way you are thinking. You really need to sample every bit coming out and low-pass filter that data stream. The low-pass filter doesn't need to be a sync filter, but a sync filter is the easiest to implement in an FPGA.

if I measure direct current where there is no sinusoid? I will change the voltage value using a DC/DC converter (600-200V range).

That doesn't save you. The key word in sigma-delta modulator is "modulator". It takes the continuous input and it modulates it into a digital data stream. That modulated stream then needs to be demodulated, and the proper demodulator is a low-pass filter.

Or without a filter, do not know the actual value of the voltage? What if we sample values ​​and get the average.

In theory you might be able to decimate the output and take the average of that. In practice it's just not going to work, or at least there will be serious consequences. I certainly wouldn't try it without a thorough understanding of how that particular chip implements its sigma-delta.

CLKIN will be at 5MHz, what frequency will the output data be? I don't plan to check the voltage value so often.

5MHz, and if you don't process every bit coming off of the chip then you run into that "in practice it just won't work" problem.

Depending on your required precision, there's a chance that you could take your 5MHz bit stream and apply it to an analog active filter on your isolated side, then read that with whatever ADC you wanted -- but if you want more than 10 or 12 bits of precision you'll need very solid analog skills, and you're probably looking at more circuit board area (not to mention production-line grief) than using a no-isolated converter with an SPI interface and using plain old digital isolators for the SPI.