How can I eliminate this spike like noise?

[doncarlosalbatros]

As a part of a big system Arduino is being effected from a switch outside(so the spike is not always there only when an outside switch or surge happens ect.) and I guess through its USB producing spikes on its output. Arduino is set normally to output around 40Hz pulse train and when the spike hits it upsets the Device it couples to below:




I dont have enough time to figure out all the system and block whats happening. I rather want to eliminate or filter the spike which looks like below:



I have a scope screenshot but it is not very good one that's why I draw it above. But anyway below is the screenshot the pink one is the pulse signal:




You can see the spike above during the pulse ON time.


How can I eliminate this spike? Would a low pass filter, zener didoe or an opto-isolator help? Is there a topology at the output of the pulse Hsignal from Arduino I can try out?

 

[KlausST]

Hi,

The spike is about 4ms after the rising edge of the pink signal?
Is this true?
And is it always that "4ms"

Then I assume the pink signal controls something like a relay or so...?

****
But the reason and the filter depends on a lot of things.
I asume it is an EMI pulse.
Then the pulse may be induced by wire or by air.... We can´t see/know this, because we dot see your application.

There´s a chance that the signal is (just) coupled by the scope´s wires.

Klaus

 

[doncarlosalbatros]

Hi,

The spike is about 4ms after the rising edge of the pink signal?
Is this true?
And is it always that "4ms"

Then I assume the pink signal controls something like a relay or so...?

****
But the reason and the filter depends on a lot of things.
I asume it is an EMI pulse.
Then the pulse may be induced by wire or by air.... We can´t see/know this, because we dot see your application.

There´s a chance that the signal is (just) coupled by the scope´s wires.

Klaus

All the issue comes from an external switch through mains.

And here are how the 12VDC 5VDC power supplies and a 10VAC step down transformer is affected by the switch in the same system:



The isolation transformer didnt help as well. It comes in anyway and upsets PSUs and the Arduino board which gets power from them and the control from USB hub inside the system.

So I just want to eliminate that spike we are not able to block that AC mains interference to come in even though we have EMI filter and tried iso transformer.

Spike is not systematic, it only occurs when an outside syetem a switch or ect is turned on ect.

 

[KlausST]

Hi,

show us a photo of your application. On this photo we want to see:
* all your GND wiring
* all your scope connections
* the "switch" (we still don´t know what this is)

Give explanations to this photo.

Klaus

 

[doncarlosalbatros]

Hi,

show us a photo of your application. On this photo we want to see:
* all your GND wiring
* all your scope connections
* the "switch" (we still don´t know what this is)

Give explanations to this photo.

Klaus

It is remote far away and very complicated we tried everything already. I just need to filter this spike. I dont have any other photo or diagram at the moment and in a hurry. Would an RC 330 ohms and 1uF ceramic help?

 

[KlausST]

Hi,

Would an RC 330 ohms and 1uF ceramic help?

We even don´t know where you want to install these parts.

Without more information it´s impossible to answer. --> try it and see.

Klaus

 

[dick_freebird]

Would an RC 330 ohms and 1uF ceramic help?

Depends where you put it, and seems you do not know where.

You could try some diagnostic "cuts" at the problem-space
like:

Trace the source of the "stimulus" by continuing with the
trigger scheme that catches the glitch, trace back through
all attached conductors (incl ground network), to the
prime source event

Common-mode AC line choke for conducted EMI susceptibility

Grounded copper cage for radiated EMI susceptibility

If you can find the precursor event's point of entry and a
local timing trigger that is determinstically related, maybe
you can add a "blanking" circuit or code that suppresses
the misbehavior.

 

[crutschow]

Spikes like that can be difficult to suppress.
A common-mode choke on the input AC, as Dick suggested, and/or better EMI filter may help.

What EMI filter are you presently using?
Is it specifically designed for the AC input?

 

[CataM]

Would an RC 330 ohms and 1uF ceramic help?

Depends how it is connected. RC filter is lossy.

Try with bypass caps with close SRF as the ringing freq.

 

[doncarlosalbatros]

Here is the basic diagram of what is inside the metallic enclosure:


Here are some photos:







The mains power entry is a module which is followed by a switch. The mains power entry module has also a filter. Here are the photos and details:


This link is the line filter used the exact model is called "5707.0801.312 - IEC Filter"

This link is one of the six daq boards installed.


The Arduino's 33Hz pulse train output goes into one of the digital inputs of the DAQ. And the spike is causing the DAQ board upset.
It is because if the Arduino output is not connected the DAQ does not get upset.

The spike is generated when a device is switched in the same mains or a surge ect. So we are not sure how it is happening.
The spike is observed by an oscilloscope bot at power supplies and Arduino output.

So my aim is to eliminate this spike instead of finding the root cause since we couldn't. It is enough to eliminate the spike only between the Arduino output and the DAQ digital input.

So how about filtering the Arduino output and using a Schmitt trigger?, I tried to simulate the spike and use an RC and a schmitt trigger with hysteresis as:



Signals at input, output and comparator's inverting and non-inverting terminals:



I put R6 C5 to suppress the spike in the system to not to upset comparator, is that a fine way to filter the spike from Vcc? What do you think about the circuit?
I can provide more information if needed.

 

[dick_freebird]

What "the DAQ board gets upset" means is unclear to me.

If the spike is presented to the DAQ input, and captured as
proper per the normal operation, that isn't "upset" in my book.

On the other hand a data level of any sort should not "upset"
the DAQ board in any sense of functional interrupt (provided
it is within signal range specs). But this bears checking, does
the spike actually exceed input range spec to the point that
it might induce component latchup etc.?

You should consider the likelihood that "the" spike is not one,
but a pervasive injection of some current pulse on multiple or
even all long wires in the harness / backplane. If so then it is
not true, that the first thing you found is the only thing. Maybe
the data is not the problem, but some similarly-provoked but
not-looked-at control line.

RC networks are commonly used for glitch filtering. However
since the situation seems still to be largely mysterious, how
will you know that the time constant that works now, will
work broadly (for example, say this really has to do with the
mains and its attribtutes-as-network - would a larger branch
present a fatter pulse from time-of-flight, or some such?).

You declare that you couldn't find the root cause, but that
is simply giving up. I suggest more discipline and determination,
because

"When you believe in things
that you don't undertand
then you suffer...."

- Stevie Wonder, "Superstition"

 

[FvM]

Unfortunately the photos don't show the mounting position and wiring of the Arduino board. If the Arduino has no other connections than USB and the two output wires, it's most likely that the interferences are entering the system through external USB cabling. I guess you can't perform any tests without connecting a computer. You might try common mode choke cores for the USB.

Another possible entry path is the external signal wiring, not clear if it was already connected during the tests.

Like dick_freebird I'm awaiting clarification about "DAQ board upset".

 

[doncarlosalbatros]

Unfortunately the photos don't show the mounting position and wiring of the Arduino board. If the Arduino has no other connections than USB and the two output wires, it's most likely that the interferences are entering the system through external USB cabling. I guess you can't perform any tests without connecting a computer. You might try common mode choke cores for the USB.

Another possible entry path is the external signal wiring, not clear if it was already connected during the tests.

Like dick_freebird I'm awaiting clarification about "DAQ board upset".

Yes the Arduino has no other connections than USB and the two output wires. This happens without any external signal wiring as well.

I will consider your argument "it's most likely that the interferences are entering the system through external USB cabling.". But could you elaborate how could that be the path for interference? (Because the issue occurs when a nearby external device's mains switch is switched on.)

For the choke do you mean a ferrite core or a choke coil?
As an alternative to a choke, would using a USB isolator between the USB Hub and the PC's USB port help?

 

[doncarlosalbatros]

I redraw the system:

 

[KlausST]

Hi,

In this picture there is a GND_loop.
Via USB_GND, Power-, GND.

Try to wire it as a star connection instead.

Klaus

 

[doncarlosalbatros]

Hi,

In this picture there is a GND_loop.
Via USB_GND, Power-, GND.

Try to wire it as a star connection instead.

Klaus

Klaust could you show/draw the loop on my diagram? It would be really helpful to minimize the conversations. Thank you.

 

[KlausST]

Hi,

There is the red loop.

And most probably there is a second one across the power supplies and the earth GND.
This usually is capacitvely coupled...no loop for DC.



Klaus

 

[FvM]

The critical ground loop is through USB ground, PC and PE. The loop allows lower frequency components and higher interference energy to strike the victim.

An USB isolator can help, particularly if it exposes low coupling capacitance. I would first try with high permeability toroid core.

 

[doncarlosalbatros]

Hi,

There is the red loop.

And most probably there is a second one across the power supplies and the earth GND.
This usually is capacitvely coupled...no loop for DC.

View attachment 150135

Klaus

Thanks, the SMPSs are isolated so the pink loop can only be capacitively coupled. But the issue occurs when the red loop is there by connecting the Arduino output to the digital input. Do you think that red loop might be picking up the EMI? And how can I starground these since they are all far ports and pins? Thanks all your support so far.

 

[KlausST]

Hi,

As already mentioned: your photos don't show the wiring.
Thus: For me it's impossible to give a recommendation.

Maybe modify the placement if the components to cut the loop.

Btw: your single ended signals are not useful for lengthy wiring in industrial environment .... as you see

Klaus