Zero crossing detection with common ground

[denujith]

I am planning to use LT1671 for zero crossing detection application.


But I have a concern that there might be a grounding issue...


LT1671 is powered from stepped-down voltage of V1. I need to measure zero crossing point of V2 signal - another stepped-down signal.


Since I have to connect grounds of V1 and V2, I need to know will there be any issues with that.


Thanks,
Lakmal

 

[D.A.(Tony)Stewart]

There is an uncertainty of signal integrity if there is either a common mode noise with an imbalanced pair of signals or a differential noise between the different grounds which may be coupled to each signal. Although this is a simple analog detector, the probability of error depends directly on the SNR and discriminator method.

Thus you must define all sources of noise by current, voltage and coupling reactance and determine the transfer function of each or use common practices to raise signal integrity margin then verify with tests.

Shielding, balanced differential methods, raise CM impedance with optoisolators or Ferrite CM choke, large flat strapped grounds ( low inductance. ) minimizing pulse output stray coupling to signal input by impedance control, noise rejection filtering

 

[betwixt]

The LT1671 is a comparator that can be used in zero crossing detection if wired appropriately. If V1 and V2 and presumably the LT1671 as well, are sharing the same ground it should work OK. If the current in the ground wiring are significant you may have to be careful exactly where you take the samples from and if using a resistive 'step down' divider, where it's ground connection is so you don't introduce unwanted ground voltage offsets.

Brian.

 

[Warpspeed]

One solution to this might be to offset both inputs to the comparator by a positive volt or so, to place both well within the allowable common mode range.

Then one input can be grounded to your remote (noisy) ground, and the other input to your remote equally (noisy) signal.

 

[denujith]

Hi Tony,

I didn't get what you mean. Can you please explain a little bit more? I am planning to use LT1671.

Regards,
Lakmal

- - - Updated - - -

I am planning to use LT1671.

I can use a better alternative if there is. I am not constrained to use LT1671

 

[Warpspeed]

Like this :



In this example, both inputs to the LT1671 are raised by the exact same +1 volts.
This allows ground 1 to move around a bit above and below ground 2.

 

[denujith]

Tony, I'll assemble this and see how it works. Thank you...

Lakmal

 

[D.A.(Tony)Stewart]

We aren't sure of your layout and noise levels but normally low current drawing shared grounds are no issue.

There are several methods of ZCS pulses depending on pulse width desired and function.

Even if the pulse is being used for edge trigger or as level duration reset and is a very narrow pulse it may not matter if the pulse leading edge is before and after the actual zero crossing or begins at the zero crossing for a defined period as long. I trust you know the difference. Most uses make no distinction.

For a ZCS signal centered on the zero crossing a separate FW bridge and a low comparator threshold will provide the pulse with the threshold top peak signal level which determines the pulse width.
For example on direct line operated circuits, one can get xx us duration signals using a large series R with a FW diode bridge and diode clamp to Vcc with the clipped full wave line signal

-adding hysteresis gives some immunity with a controlled % of positive feedback to threshold. We add an RC filter LPF with known phase shift before the FW bridge to improve transient impulse noise rejection greatly. This could mean a common high V1 may be used for both powering and detection with separate bridges.

Another method uses and RC filter on line signal, then rectifies and HPF differentiation to a comparator with hysteresis may also be considered.

Another method simply ac couples the line voltage in a R divider to make a logic square wave then using an RC LPF filter on one side of an XOR gate controls the pulse width after ZCS detection. This is my preferred method as it offers high SNR, ease of filtering, limiting , precise PW control and uses standard logic gates.

Of course you may not care about false ZCS pulse signals, but I did, so I used impulse LPF rejection, CM ferrite beads and twisted pair wires to offer the highest SNR.

As I recall, I did something like this.
http://goo.gl/DMj2yz

 

[D.A.(Tony)Stewart]

My falstad sim link
Concept with random neutral noise injection option using logic gates only off grid.Also works with secondary LV.

 

[denujith]

Hmm... I've got a lot of inputs... Thank you everyone... Planning to try with the ideas and suggestions... Will post how it went...

 

[D.A.(Tony)Stewart]

Hmm... I've got a lot of inputs... Thank you everyone... Planning to try with the ideas and suggestions... Will post how it went...

For understanding signal integrity, it helps to know impedance of source at signal and noise frequencies. When this ratio SNR is poor, then we use shielding, ferrite CM chokes or differential mode ferrite beads, RF filters, and examine physical paths as waveguides looking at frequencies that resonate like 1/4 wave etc.

Generally signals should be low source Z and high Z load and balanced giving high differential mode, DM Z but low CM Z.

When using high impedance single ended sources, susceptibility to stray noise by E or H field is worse, thus CM chokes lower the CM impedance and thus shunt stray coupling in series to the single ended source ground reference or if floating to the single ended termination ground. E.g. Ethernet, phone to -48V office . When dealing with RF signals however, matched impedances work better for integrity with higher return loss and lower ingress.

Type and Quality of cables matters.

if possible choose, twisted pair (UTP) , shielded twisted pair (STP) or coax with impedance matching. This works in two ways ingress & egress to improve signal integrity.


An example of Worst case is long inductive non-twisted pair, closely coupled to other noise signals and mis-matched grounds with noise current flowing thru it with arc-welding or ESD near cable. E.g. Stepper motor current and high Z optical sensor. This would result in "glitches"

 

[denujith]

Thanks. I'll keep these things in mind.

 

[D.A.(Tony)Stewart]

I hope you understood

 

[denujith]

I hope you understood

hmm... I am not that educated in electronics as you are quite difficult to digest in the first place. anyway, I can understand up to some extent. I have to learn a lot... thank you so much for your helping hand