How to stabilize the millivolt signal of mains

[mheruian]

Good day folks,

I am creating an AC signal small enough to act as a signal (like sensor value) on a PMIC. I want to use it like a testing rig for PMICs before actual setup to check if PMIC really works. Also to test it when an inverted phase is fed on it. I research a lot of ways to do this but I chose the easiest one since I'm limited to components and I am just getting started to know the basics stuffs of electronics (my knowledge are just around voltage dividers and stuffs ).

I choose using the mains itself since I'll have the free sine wave phasing etc. for me to use since i needed a sine wave analog signal to be fed on the PMIC analog voltage input. But since mains are high voltage and analog voltage input of PMIC accepts around 250mV only, as to be used also on actual, using voltage divider was my way to do it where I make the R2 to be variable to meet analog voltage values from 10mV to 250mV. I also included inverting op amp since as I said I wanted to test it also in inverted phase.

My newbie basic circuit works like this:



now, when values are around above 190mV, signal is quite stable (random 190mV - 192mV), but when it reaches like 12mV, the signal is unstable like it'll go random 12mV - 12.9mV - 13.3mV - 11.7mV - 13.8mV, etc etc. which is not tolerable for me. I am fine if atleast it'll have a ±0.3mV deviation from the desired value like 12.0mV - 11.7mV - 12.3mV but more than that, i wouldn't get my target values of: 12.4mV, 13.0mV, 13.6mV, etc. that are lowest.

Now, i tried doing some filtering like putting mylars after voltage divider or using voltage follower or even AC coupled voltage follower to make it stable etc. but failed.

Can anyone teach me how to regulate/stabilize AC signal specially on its millivolt value?? I only know basic stuffs and not a math wizard yet

[ArthurDent]

That is not a safe way to do what you want to do. The theory may be sound but if the 5M resistor were to fail you would have a real problem. I've seen circuits where there were resistive sampling circuits for high voltage D.C. and for a 5M value they would use five 1M or ten 500K resistors in series to make sure the resistors were operating well within their voltage ratings. They would also use protective diodes from the opamp input to each supply rail to try to protect the rest of the circuit.  The fact that you have no isolation from the line is also scary. I would be using a 240 to low voltage transformer which besides making the circuit safe would also attenuate higher frequency crud. 

[DannyTheGhost]

In case of millivolts, you will almost always deal with noise. if you look at absolute voltage error in both cases, its around 2mV there and there. Source of noise could be anything, from AC network to inner parts of circuits, i.e. your Op-Amp.

[mheruian]

That is not a safe way to do what you want to do. The theory may be sound but if the 5M resistor were to fail you would have a real problem. I've seen circuits where there were resistive sampling circuits for high voltage D.C. and for a 5M value they would use five 1M or ten 500K resistors in series to make sure the resistors were operating well within their voltage ratings. They would also use protective diodes from the opamp input to each supply rail to try to protect the rest of the circuit.  The fact that you have no isolation from the line is also scary. I would be using a 240 to low voltage transformer which besides making the circuit safe would also attenuate higher frequency crud.

Hi, Ok thanks a lot.
so using transformer like 240 to 9v would reduce noise??

In case of millivolts, you will almost always deal with noise. if you look at absolute voltage error in both cases, its around 2mV there and there. Source of noise could be anything, from AC network to inner parts of circuits, i.e. your Op-Amp.

then there is no way around regarding this?

[DannyTheGhost]

then there is no way around regarding this?

For external sources of noise you use, for example metal shielding, but its a bit overkill for circuits like this, so it should placed far away from grounded metal parts of your desk and high power devices, which create a lot of EMI for sensitive circuits
For internal sources, well, there is too much to tell in one msg. In your case, you should design your circuit with less traces between parts, especially for Op-Amp and its feedback part (resistors and other parts, which included in feedback circuit, must be placed as close as possible to Op-Amp). Furthermore, cheap general-use Op-Amps has generally high internal offsets, like input voltage offset etc. If you care about accuracy, use precision-type Op-Amp, which are designed with low internal offsets
i.e. your TL072 has its input offset voltage around 2mV, so, I suppose, that's what you see in your results, while precision Op-Amp has ~1uV input offset voltage, or even less.
btw, its better to use inverting voltage follower when you need unity gain Op-Amp