modified GFCI

[JR.]

I continue to work on my "hard to fool" outlet checker. The problem is simple outlet checkers can be fooled since the voltage is AC there is no easy way to determine wiring polarity without an external reference.

As mentioned a neon lamp or high impedance VOM will work using the human body for a ground reference while it can not discriminate between ghost (high impedance) leakage voltages and real faults.

My first approach was to make mosfet source follower using a pair of Nch and Pch 400V power mosfets. This did not work reliably.

My next approach was to use separate circuits for each outlet lead. This proved too sensitive at first with both the line and neutral leads indicating hot when I touch the gate with my finger. If I shunt the touch node with 100k neither one indicated. A 1M shunt was just right for the hot line to indicate hot, while the quiet neutral line indicated cold.

This suggests to me that the electrical model of the human body includes an antenna (voltage source) adequate to fire the unterminated mosfet (like when you touch the tip of an audio jack. !00k termination forms a RC with the body capacitance such that the body charges up quickly enough to not turn on the mosfet. With 1M termination the residual noise of the human body (antenna) is damped enough to not fire the mosfet, while high enough impedance that the body capacitance holds the gate relatively steady while the swinging source, turns the device on.

This is really interesting and kind of stretching my gray matter (my analysis is mostly speculation). My next step is rebuild my circuit with lower voltage and lower current mosfet that will exhibit lower input capacitance. The current mosfets are around 300pF, so I can get that down to 100pF which will make them more sensitive, while that sensitivity is a bit of a trapeze act, requiring a goldilocks just right termination.

JR

[emrr]

I've probably already said it: the simple testers are easily fooled for even ground connection. Plug a 3 prong cable into an ungrounded outlet. Plug the tester into the other socket of same outlet; it will tell you ground is present because of the other ground pin presence, NOT because ground is actually present.

[JR.]

I've probably already said it: the simple testers are easily fooled for even ground connection. Plug a 3 prong cable into an ungrounded outlet. Plug the tester into the other socket of same outlet; it will tell you ground is present because of the other ground pin presence, NOT because ground is actually present.

Yes, if that gear is connected to other gear that is plugged into a grounded outlet elsewhere, it may find a ground via that path. Also if outlets have a floating ground (like all the outlets in my house) leakage current can accumulate say between products plugged into the same outlet strip. This ground leakage can cause phantom or ghost voltages that sensitive testers can see.

Further proper testing of grounds involves injecting a pulsed current into the outlet ground to confirm it follows a separate path back to the panel and is not just bootlegged to neutral.

JR

[emrr]

The tester I have is fooled by a 3 prong AC cable that goes nowhere; I don't know why it thinks ground is present.

[JR.]

The tester I have is fooled by a 3 prong AC cable that goes nowhere; I don't know why it thinks ground is present.

Perhaps the tester is faulty... ? A cheap tester will connect a lamp between line and ground so that ground needs to sink a few mA to light the lamp...If an open floating ground still lights the lamp something is wrong, like ground shorted to neutral or something like that.

Even neon lamps require a half mA or so light brightly. I doubt a line cord has enough capacitance to the environment to sink that much current.

JR

[JR.]

Good news, I heard back from my guitar amp expert tester. He was busy because his guitar pedal company (Amptweaker) has been slammed with sales since NAMM show. As a favor to me, he checked this out yesterday.

The GFCI strip I sent him was modified with a 3 position switch so he could compare a hard ground connection, a floating ground, and a cap in series for the ground path. I used a 0.15uF cap so a mains voltage fault through that cap coupled ground would trip the GFCI 6 mA threshold.

For noise floor the floating ground position was noisy, and both the hard grounded and cap grounded positions were quiet. An additional benefit arose from preventing ground loops. If the amp was plugged into one outlet, and say a preamp or powered pedal was plugged into an outlet on a different branch circuit, it hummed in the hard ground position, but was silent in the open or cap grounded position.

So this is a win-win.... 2 wins and no losses.

JR

[mediatechnology]

Great news!

Sounds like things behaved the way they should.

Link to Amptweaker: http://www.amptweaker.com

[emrr]

Cool

[JR.]

I have already shared my KISS GFCI design. Simple just add a cap in series with a GFCI protected outlet ground lead. A 0.15uF cap will sink enough current to trip the GFCI but still be too low for humans to get stuck to.

My premium GFCI uses a lot more parts and provided extra protections.

#1 a 3 pole relay in front of the GFCI outlet, but powered from the output side of the GFCI latches on when the GFCI is powered up, and disconnects the ground too, when GFCI trips.

#2 a momentary touch on circuit, provides short term power to latch the GFCI on. This touch circuit will not work if the outlet is reverse polarity so automatic protection agains RPBG (reverse polarity bootleg ground) miswired outlets.

#3 A current sense in the ground lead will force the GFCI to trip and open the relay if several mA is detected in the ground lead.

All these extra parts are not cheap. The 3 pole relay at 16A is not a common relay so not cheap (especially in small quantity). The momentary start circuit needs to use triacs and opto-isolators for both the high side and low side of power supply, so 2x 16A triacs and 2x opto triac drivers. A low capacitance mosfet is used in the touch circuit so input touch switch has 1M resistor in series. Finally a third opto-triac is used to sense the ground current and imbalance the GFCI to force a trip.

I will draw up a schematic and post it (later), since I do not plan to sell these.

JR

[JR.]

gfci_1.jpg (43.57 KiB) Viewed 11533 times

This is my second pass schematic... I will add values and correct part numbers later.

The GFCI box is a commercial GFCI outlet.

The opto-isolators I used for the schematic show bipolr transistor outputs. The actual parts I used are opto-triacs.

There's several things going on that aren't obvious... I'll write up a description tomorrow.. right now I need to mow my lawn.

JR