Mains sine wave through opto isolator ?

[userx2]

Hello

I have been struggling here for some time but not had a decent result yet.

I would like to isolate a mains sinewave for further processing and zero x detection on the low voltage side.

The constraints are that there can be no op amps or devices that require power rails on the mains side.
I am not ruling out very low power processor yet.

Whilst this seems simple with 2 optos at first, their current transfer ratios vary so much (>100%) that I cannot rely on a simmetrical sinewave on the other side.
I have not been able to source any isolator that has 2 matched leds and outputs or a device with a guaranteed transfer ratio.

Those HCNR300 types are not of any use here since as they are just as bad without the compensation of the second photo diode.

Unless someone knows of a more suitable opto, I think the only solution is to get the whole signal through a single opto. I can fullwave rectify it and send it through one opto but then I have no solution on re-assembling it on the other side.
I could use the rectified signal but I have found that low pass filtering that as is does not produce satisfactory results.
The signal distorts and then there is the phase shift as well...

I cannot use a transformer for this application either.

Could I somehow sample the mains with a small processor and transmit the data through the opto?
I think then the problem is sampling the negative half.

Any ideas are greatly appreciated.

Regards
X

 

[FvM]

I presume that zero crossing detection won't be a problem, but transferring the analog voltage isn't that easy. Different solutions are possible, you'll need to specify intended accuracy, bandwidth, acceptable input current, board space, costs for a qualified trade-off.

Some points in your considerations are self-contradictory, however. If you think about digitizing the input voltage, you'll have some kind of primary power supply and other solutions like analog optocouplers with linearization would apply as well. All in all it's a matter of input current and board space.

Some points are missing an explanation, e.g. "can't use transformer". Is the problem about transformer size (it's variable, in fact) , or something else?

 

[userx2]

I presume that zero crossing detection won't be a problem, but transferring the analog voltage isn't that easy. Different solutions are possible, you'll need to specify intended accuracy, bandwidth, acceptable input current, board space, costs for a qualified trade-off.

Some points in your considerations are self-contradictory, however. If you think about digitizing the input voltage, you'll have some kind of primary power supply and other solutions like analog optocouplers with linearization would apply as well. All in all it's a matter of input current and board space.

Some points are missing an explanation, e.g. "can't use transformer". Is the problem about transformer size (it's variable, in fact) , or something else?

Hi FvM, group.

Because I am in the dumps at the moment with my failure to solve this, I sort of typed in all the different avenues I have come up with. They may contradict each other but I don't want to rule out any solution completely yet.

I need a highly reliable solution with zero x jitter / accuracy of 50us or better and bandwith between 40 - 70 Hz or <100Hz.

Transformers are out for this as they are too big and heavy. the really nice pcb mount small ones I have evaluated do not output any usable sinewaves.

My original plan was digitising on the mains side and digitally processign it into pulses but this seems to need as much power as having analogue circuitrly doing the whole job.

I would really like to see if I can get away without a power supply on the mains side and do all processign on the lower voltage side. I am hoping to have <1/2W power use for whatever circuit I come up with per phase.
If power is needed, it should be low enough so I can obtain it through resistive dividers.

Solving this whole problem if a power supply is added, is not an issue and I can handle that easily.


Regards
X

 

[FvM]

To start with a simple option, a 3-windings transformer can be used in a feedback circuit with zero core flux.

 

[userx2]

To start with a simple option, a 3-windings transformer can be used in a feedback circuit with zero core flux.

I would like to know more about that.
However, my experience with getting transformers safety and UL approved (which it will need to be) has not been straighforward either.

Regards
X