Leviton GFCI nuisance tripping and circuit analysis

[gar]

190117-1538 EST

tersh:

I tried several different inductive loads this afternoon, 120 V circuit, and none produced any huge voltage transients. I did not have one of my 8' Slimline fixtures without a line filter to test. None of my test loads produced more than about 20 V.

I am beginning to wonder if your load has some defect causing unwanted leakage current only under turn on or off times.

Does your former college have test equipment and willingness of your professor to study your problem? This would be the best place to do this.

.

 

[junkhound]

190117-1538 EST
unwanted leakage current only under turn on or off times.
.

Consider: 220 V winding, when one side switched off there is a 200 pF or so path to the grounded core. Thus one side open, other side high dv/dt. I = C* dv/dt. I shows up as a differential mode current in the CTs.
120 V sp motor one side of winding usually wired to be the ground side against core.

 

[gar]

190117-2405 EST

tersh:

My latest thought on an experiment that might be useful.

This experiment will use the current transformer from one of your somewhat destroyed boards.

Use an adequately good digital scope to view the output of the differential current transformer with different switched loads.

The object is to see how well the differential transformer works in taking the difference of the two currents, forward and return, to the load. And how different loads change the signal.

.

 

[gar]

190117-2437 EST

tersh:

I just ran an experiment with my Fluke clamp-on Hall device current probe. This probe should have moderately good bandwdth

The switched load was a fan in a 1500 W space heater. No power was applied to the heater, just to the fan.

The cancelation was not good at all. Saw some peaks around 10 A. Turn on duration of oscillation was much shorter than turn off.

A much better differential current transformer is needed than what I saw with the Fluke probe.

.

 

[tersh]

190117-2437 EST

tersh:

I just ran an experiment with my Fluke clamp-on Hall device current probe. This probe should have moderately good bandwdth

The switched load was a fan in a 1500 W space heater. No power was applied to the heater, just to the fan.

The cancelation was not good at all. Saw some peaks around 10 A. Turn on duration of oscillation was much shorter than turn off.

A much better differential current transformer is needed than what I saw with the Fluke probe.

.

But note it never trip on the Siemens GFCI breakers which I have fully tested.

Anyway. There seems to be a poor man self-test in the china unit. The yellow marks and labels are the normal circuit. The red marks and labels are the electronic reset button electronics. The blue marks labels are the poor man self test or defect indicator. When the diode below the SCR was removed (2 days ago). There was red light in the receptable. The red light also appeared when the SCR was fried. What kind of circuit is the one shown in blue where it can detect the missing diode? Note the poor man self test circuit is connected to Pin 5. Recognize what kind of circuit is it? What is that thing in the green question mark (below C3)? and what do those 2 diodes do?

The following was old file I shared before. The ??? is thermal fuse. I didn't add the reset electronics and poor man self test electronics:

 

[tersh]

But note it never trip on the Siemens GFCI breakers which I have fully tested.

Anyway. There seems to be a poor man self-test in the china unit. The yellow marks and labels are the normal circuit. The red marks and labels are the electronic reset button electronics. The blue marks labels are the poor man self test or defect indicator. When the diode below the SCR was removed (2 days ago). There was red light in the receptable. The red light also appeared when the SCR was fried. What kind of circuit is the one shown in blue where it can detect the missing diode? Note the poor man self test circuit is connected to Pin 5. Recognize what kind of circuit is it? What is that thing in the green question mark (below C3)? and what do those 2 diodes do?

The following was old file I shared before. The ??? is thermal fuse. I didn't add the reset electronics and poor man self test electronics:

​

I bought another 2 pieces of the china oem GFCI and shared the youtube video of the tripping when used with the Waterpik shaded pole motor (turn on your volume so you can hear the contacts opening and closing). The latest GFCI has 5.5mA tripping tested compared to the 4mA tripping before. It seems to be improving with 1 trip out of 4 to 5 instead of 4 trips out of 5 (which occurred when I got samples with 4mA tripping). Don't know if it's coincidence. I turned to the right after the 4th switch because the cellphone was blocking my view and I was looking at the units. Later I switched to the hair dyer to show other appliance doesn't trip. Only the water shaded pole motor and maybe some other I haven't encountered yet as of now.

What do you make of it? leaked current? But where does the waterpik shaded pole motor leak the current when it's put on plastic and and whole unit was put on glass surface??

I should have shared the actual weeks ago when it tripped 4 out of 5 times. They were already installed on the wall and I don't want to experiment on those already on the wall. But I'll see if I can get another 4mA tripping to see if there is correlations between tripping times and tripping thresholds.

If it's leaking current. How come the Siemens GFCI 2-pole Breaker never trip? The whole room is current protected by 30A version of the Siemens GFCI breaker and I have 5mA tester for it which works (i'll share how I tested it sometimes). Also my fridge never trip for 2 weeks already (before when I put it on the china GFCI, it tripped twice a day).

Leaked current or inductive kick inducting capacitive coupling within the circuit?

 

[gar]

190118-1129 EST

tersh:

This is not a comment on your last post, but a continuation from my last post.

The Fluke Hall proble has problems doing the differential measurement at high frequencies because the waveforms and mangnitudes make no sense.

This morning I quickly wound a current transformer. Thus a passive only device. It needs refinements, which may inclucde an electrostatic shield. There are high frequencie resonances, damped oscillations around 100 kHz. Contact arcing up to 20 mS on contact closure. Not huge currents. No more time now.

.

 

[tersh]

190118-1129 EST

tersh:

This is not a comment on your last post, but a continuation from my last post.

The Fluke Hall proble has problems doing the differential measurement at high frequencies because the waveforms and mangnitudes make no sense.

This morning I quickly wound a current transformer. Thus a passive only device. It needs refinements, which may inclucde an electrostatic shield. There are high frequencie resonances, damped oscillations around 100 kHz. Contact arcing up to 20 mS on contact closure. Not huge currents. No more time now.

.

I have fully traced the circuit:

I'm still analyzing how the fault circuit works that can light up the Red Led when say the main diode was removed (as you instructed a few days ago). If you have any clue, please let me know. Also I'm identifying what exactly are these parts (SCR or transistor, etc. in the black one. Diodes or Zener, etc. in the colorful tubes).


I need to understand how they work so I can know how good the china gfci can detect and display fault (and what fault). I'm spending time with this circuit because it's one of those few that can actually save lives besides those electronics in airplanes (in contrast to say electronics in the apple Ipad which is for entertainment only).

Btw.. I never talked with my engineering professors for over 20 years already and don't know if they were still teaching. Also they may not consider this important for any analysis.

 

[tersh]

I think I have isolated the problem. I disconnected/cut the main current inbalance sense coil in the 4th test pcb (see diagram with "cut" label above because I don't want to spend another week analyzing it as it's taking my time off theoretical physics which is my interest now). And the gfci no longer trip even after 50 switches! Neither could the Test button work. Nothing else was disconnected. Pin 5 (line) is still intact as there is no red led fault indicator warning. So it's not transient getting into pin 5 or pin 7 or C3 directly affecting the SCR.

Now here's the things to figure out. Is it really a subtle current inbalance due to the motor (some said capacitors inside the motor)? But why can't it affect the Siemens 2 pole GFCI breaker that is guarantee to trip at 5mA (I have tested it manually with an amazon 120v GFCI tester with one leg connected to either line and the ground connected to the soil (it trips)). Do the Siemens have better electronics that can filter those motor line inbalance? How does it differ to normal line inbalance, how can it distinguish?

Or did somehow the transients to the china gfci can make the sense coil wires leading to pin 1,2,3 pick up signal enough to cause false positive? That is, not from transients directly triggering the SCR but from the three wires actually inducting current inbalance enough for to chip to make it's normal function.

What do you think? And can you share exactly how pin 1, 2, 3 work in great details?

 

[gar]

190118-2454 EST

tersh:

I believe your blue circuit is incorrectly drawn.

Assume the device is a transistor. Then the upper terminal connected to a heavy trace is the emitter. At this time I don't know if it is NPN or PNP. Your circuit is not correct enough to determine that.


See https://www.alliedelec.com/m/d/1e9376b215eb11e4d8f38e0bb6b703c9.pdf for a MOSFET.
See https://www.westfloridacomponents.com/mm5/graphics/ds10/SST2222A.pdf for an NPN. You have to enlarge the datasheet many times to see the pin numbers.

It appears there is a 10k from base to emitter. There is a 4.7k from collector to diode. You can use a Simpson 260 to determine polarity of the diode. If the other is a Zener then with a DC source, a current limiting resistor, possibly 10k, and a meter you should be able to determine polarity and breakdown voltage of the Zener.

If it is a bipolar transistor, then a Simpson may allow you to determine NPN vs PNP. A more accurate circuit diagram may also identify whether NPN or PNP.

I don't know what L1 and L2 are, top or bottom of the two power wires.

What do you mean by reset coil comes from trip coil?

If two crossing lines connect put a dot on top of the intersection.

You have dangling leads that do not have a good definition of to where they connect.

.

 

[tersh]

190118-2454 EST

tersh:

I believe your blue circuit is incorrectly drawn.

Assume the device is a transistor. Then the upper terminal connected to a heavy trace is the emitter. At this time I don't know if it is NPN or PNP. Your circuit is not correct enough to determine that.


See https://www.alliedelec.com/m/d/1e9376b215eb11e4d8f38e0bb6b703c9.pdf for a MOSFET.
See https://www.westfloridacomponents.com/mm5/graphics/ds10/SST2222A.pdf for an NPN. You have to enlarge the datasheet many times to see the pin numbers.

It appears there is a 10k from base to emitter. There is a 4.7k from collector to diode. You can use a Simpson 260 to determine polarity of the diode. If the other is a Zener then with a DC source, a current limiting resistor, possibly 10k, and a meter you should be able to determine polarity and breakdown voltage of the Zener.

If it is a bipolar transistor, then a Simpson may allow you to determine NPN vs PNP. A more accurate circuit diagram may also identify whether NPN or PNP.

I don't know what L1 and L2 are, top or bottom of the two power wires.

What do you mean by reset coil comes from trip coil?

If two crossing lines connect put a dot on top of the intersection.

You have dangling leads that do not have a good definition of to where they connect.

.

Ah. diagram corrected:

the Red line in the diagram is from the Red line in the following labels:

According to electronic people I asked. The unknown transistors is likely PNP transistor. See: https://www.mouser.com/ds/2/302/BC856_BC857_BC858-1147070.pdf

For PNP transistor. Current is from emitter to collector with Base as the one that controls the current flow.

And the diodes might be a zener in series with a diode in some sort of transformerless power supply from the mains. There’s a chance it’s a 6.2V Zener and the diode is a LL4148. The cathodes have the band, so they are in inverse series.​

 

[gar]

190118-2545 EST

Verify that it is PNP, and correctly draw the circuit. And fill in some of the other details.

.

 

[tersh]

190118-2545 EST

Verify that it is PNP, and correctly draw the circuit. And fill in some of the other details.

.

Corrected including the black power supply to the reset switch.

 

[tersh]

Correction: The 10k is connected between Emiter to Collector. You can trace using the actual circuit front and back.

Full PCB trace front and back

 

[tersh]

redrawn.. this is the clearest ever. so any idea what makes the red led light up when there is fault like the main diode removed?

 

[tersh]

gar, here is something so puzzling.

First, my multimeter can measure polarity in diodes so I have checked the polarity.

Second. If only a removed diode (like you asked me too) can activate the red fault led and it did, maybe not so hard.

But when I removed pin 1 in the 2nd pcb a few days ago and it immediately destroyed the SCR. However, the diode and resistor is still contacted to the thermal fuse and the AC (red), and DC voltage of about 12v still getting to pin 5. And the red fault led turns on! How can it know the SCR was disconnected??

To make it simpler, what kind of signal to input into the violet dot at middle to turn on the red fault led?

Btw.. about removing the sense coil and the tripping gone. Yes. Can do that consistently. You can
get same effect by shorting pin 1,2,3 in a fully working pcb. This will disable the current different sense coil and the Test button won't work. And you won't get tripping this way either. It's reversible because once you no longer short pin 1,2,3. It will work again with the Test button working everytime. So it is the sense coil or wires leading to pin 1,2,3 that can cause tripping. It's so consistent one can repeat it and get same results.

 

[gar]

190119-0727 EST

tersh:

The 10k is between pins 1 and 2. The 4.7k starts at pin 3, collector, and goes to diode. 1 megohm goes from collector to AC line. red. Your blue line seems to imply that the LED is between junction of 4.7k and 1N4148, and red AC line.

Note: A 1N4148 is a relatively low PIV diode, 100 V. A 4.7k current limiting resistance for an indicator LED implies low voltage , like possibly 50 V, 2500/4700 is 1/2 W, sort of high, thus not likely as much as 50 V.

See if you can correct the circuit more.

.

 

[gar]

190119-0820 EST

tersh:

Looks like reset button, starts somewhere not clear where, goes to cathode of diode, diode anode to small 47k resistor, to base of transistor, 10k from transistor base to transistor emitter.

On collector side looks like collector to both 4.7k and 1M. 4.7k to LED other end of LED to other end of 1M. Why shunt the LED and 4.7k with 1M? Does not seem too logical.

Also from collector is the 1N4148 and Zener to somewhere. Appears that diode and Zener are in series opposition. This means the Zener forward direction is turned into a high impedance, and the combination only clamps at the Zener breakdown in one direction. Was that necessary? Don't know because I don't yet know what the whole circuit looks like.

Without rereading I think this says what I wanted to say.

.

 

[tersh]

190119-0727 EST

tersh:

The 10k is between pins 1 and 2. The 4.7k starts at pin 3, collector, and goes to diode. 1 megohm goes from collector to AC line. red. Your blue line seems to imply that the LED is between junction of 4.7k and 1N4148, and red AC line.

Note: A 1N4148 is a relatively low PIV diode, 100 V. A 4.7k current limiting resistance for an indicator LED implies low voltage , like possibly 50 V, 2500/4700 is 1/2 W, sort of high, thus not likely as much as 50 V.

See if you can correct the circuit more.

.

Thanks for the corrections. I was using symbol and actual components designation. The following is the corrected one. But note that I double checked,

LED is between junction of 4.7k and 1N4148, and red AC line. Won't this make sense?​

Also note the one in the reset section is not transistor but same model as the main SCR that pin 7 triggers.

 

[gar]

190119-0850 EST

tersh:

The PNP is only rated 65 V. Can't put PNP, 4.7k, and LED in series across an AC 240 V line. Things will blow up. Not one of the components could tolerate this. Both the forward and reverse directions are a disaster.

.