[SOLVED] switch off relay inductive load arcing waveform on scope, I don't understand.

[pgib8]

Hi, I'm looking at PCB for switching AC mains voltage, and I'm only looking at the contacts side and not worry about the solenoid side, which includes a free-wheeling diode for the solenoid. I believe it is a poor design because there is no protection across the contacts and there is also a power supply on the same board that is producing voltage spikes when the relay switches.
I only notice this when I switch an inductive load, for which I'm using a 75W blower fan.
My theory was that when the contacts open, the inductive load produces a high voltage that arcs across the contacts. I'm pretty sure I have confirmed this by looking at the voltage across the contacts on the scope. See below:
First what it looks like when switching off an incandescent light bulb (yellow trace is the contact voltage):

Now the same thing with the fan instead:

So when I zoom in on that spike I see this:

And when I zoom in further, at the beginning of it:

So here is my question, if this is really because of arcing, I would like to understand this waveform better.
I imagine at the beginning the voltage increases because the contacts have already separated at this point and the inductive load is building up the voltage as the parasitic capacitance gets charged up by it. Then when the trace comes vertically straight up is where the arc occurs. The ringing that follows is either an artifact from the measurement or some kind of natural oscillation in the circuit like a spark-gap transmitter but I'm not too worried about this. I assume the arc extinguishes right when the voltage starts to build up again.
So the thing I don't understand what are the flat spots. Later on in the waveform they become really pronounced, take a look:

It appears to me that during the flat spots, the contacts have about -160V across and are just sitting there and then out of the blue it decides to arc again? Why wouldn't the voltage keep increasing and then once it reaches a maximum then right away the vertical line happens? Or maybe I'm completely misunderstanding what's going on and any help would be greatly appreciated!

 

[KlausST]

Hi,

when the contact opens
* the voltage across the contact begins to rise.
* air becomes inoized
* air becomes conductive --> spark
* current flow
* the gap (spark) becomes low impedance
* thus the voltage decreases
* the spark turns OFF
* the gap becomes high impedance
* voltage again begins to rise.

***
Indeed I also wonder why ther is a flat top. Maybe inside the fan there is overvoltage protection.
But it seems the voltage is high enough for the air to become inonized... until the spark appears.

Klaus

 

[pgib8]

At least I'm not the only one that wonders that. I tried the same thing with a smaller 28W blower fan and I see the flat spots too. Maybe the flat spots are the arc, no clue.

--- Updated Nov 9, 2023 ---

I tried something else. Instead of a fan I connected a 120-to-240 transformer by itself and I got this (when turning the relay off):

 

[dick_freebird]

An arc has a negative differential resistance and with a LC tank
this can make an oscillator. The Tesla coils used spark gaps (this
being the only "active element" available) to make the primary
oscillate.

 

[KlausST]

Hi,

so the contact side wiring is:
Mains -> contact -> inductance -> Neutral
Nothing else?

Can you really see the spark?

What exact relay (link to datsheet) is it?

What are your scope probes?

Klaus

 

[pgib8]

@KlausST that is correct, although there is also a power supply in parallel to this but I don't foresee that causing such spikes.
I can't see any spark, that is just a theory, the relay is enclosed.
For good measure I posted the schematic below. It is for a light that can turn on and off with ambient light but also measures power consumption. If you're looking at it anyway, one comment I had about that is that I questioned D5 that comes after the transformer because I wasn't sure if that was meant to go to GND and/or if there should have been a second zener from 3.6V to GND. If it was me designing it I would have gone with an LDO from 24V to 3.6V like he does with 5V, but that's just me. Keep in mind line-in is the same as GND, I guess for power measurement reasons. Btw. the cyan trace is the 3.6V rail.
On the scope I'm using the probe it came with for the 3.6V and a differential probe (DP10013) across the relay contacts mainly to protect my scope.
Datasheet of the relay: http://www.hongfaamerica.com/hq/PDF/HF21FF_en.pdf
Edit: I forgot to mention that this PCB is hooked up to an isolation transformer so that I can use the ground from the scope where needed.

 

[D.A.(Tony)Stewart]

Each quirky looking "flyback" arcs and resonant signal can be modelled with the above explanations including the 25 MHz ringing from your long 10:1 probe ground resonating with it's coax capacitance.

Reducing that can be done by removing the ground lead, and probe top replaced with a spring probe around the ring , a common solution for a test engineer.

Dry contacts switching inductive loads will always spike with the same power circulating before it opens. RC Snubbers across are used to reduce this EMI or simply a 1n to 4n7 film capacitor across the contacts.

 

[FvM]

I agree that the observed ringing is likely caused by probe and other circuit resonances. The characteristic flat top waveform with sudden breakdown can be however assigned to arc behaviour. We see that arc formation is relative slow, taking 100 or more microseconds. Due to exponentional increase of ionization there's a sudden breakdown.

 

[danadakk]

Proper probing considerations :

https://download.tek.com/document/ABCs-of-Probes_Primer_60W-6053-16.pdf


Regards, Dana.

 

[pgib8]

Thank you guys so much for helping me to understand this, especially @FvM, you showed me the door and it helped me look up the right stuff.
Probably one of the best pictures I found was from https://en.wikipedia.org/wiki/Electric_discharge

This looks exactly like from one of my scope shots (when viewed upside-down):

Here are some more results:

relay arcing dark glow discharge - Google Search

This was a complete mystery to me and now I understand, so thank you again to all of you.

PS: I can't find my little spring probes right now, I've never used them but I know what they're for and I will still try and read up on the probe usage tips that were shared and appreciate those too.

 

[KlausST]

Hi,

I just want to mind:
* on the one diagram the X-axis is "current"
* while on the other diagram the X-axis is "time".

So if you found some similar waveform in both diagrams ... it is wrong. One can't compare current with time.

Klaus

 

[pgib8]

Hi,

I just want to mind:
* on the one diagram the X-axis is "current"
* while on the other diagram the X-axis is "time".

So if you found some similar waveform in both diagrams ... it is wrong. One can't compare current with time.

Klaus

I noticed that too but I said that's close enough. My intuition told me that time is still part of the X-axis because of the letters where we go from A to B to C and so forth. I figured we are following this curve and that it can have a different shape depending on the current but that overall it will look similar to this and it does. My main concern was the flat portion and this diagram shows that it is part of the discharge. Thanks again.