modified GFCI

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
Prototype2
Posts: 16
Joined: Fri Oct 30, 2015 9:06 am
Location: United Kingdom

Re: modified GFCI

Post by Prototype2 » Sun Nov 01, 2015 5:39 pm

I should add that your device sounds like a great idea, and I'm sure you'll be able to find buyers should you decide to have them manufactured. What's your target unit cost?

User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

Re: modified GFCI

Post by JR. » Sun Nov 01, 2015 7:20 pm

I really have no idea yet... the parts cost is not very significant One mosfet and a handful of diodes, LEDs and resistors. I have decided not to pursue a patent but UL approval could be a couple $10K.. I was hoping to not start another business but if nobody is willing to partner I will do this... to maybe save a few lives.

JR

PS: Regarding ground, with GFCI outlets ground it is not completely necessary but it is worth indicating when present or not.

User avatar
mediatechnology
Posts: 3276
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: modified GFCI

Post by mediatechnology » Sat Nov 07, 2015 6:26 pm

Well see, I do not expect UL approval to be cheap or easy. .
Can you get CSA for less cost and then effectively have UL?

User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

Re: modified GFCI

Post by JR. » Tue Nov 10, 2015 11:03 am

mediatechnology wrote:
Well see, I do not expect UL approval to be cheap or easy. .
Can you get CSA for less cost and then effectively have UL?
The combined approval is called CUL don't know that it's any cheaper.

I purchased the UL spec that covers the existing (cheap) outlet tester. The cheap tester may be $5 but the UL spec was $400 :shock: The spec was clearly written after that tester was designed. The spec stipulates fine print to add to the cheap tester instructions listing things it doesn't do (but mine does). :oops:

My original tester design used a 1M touch probe input resistance, which IMO is adequately safe << 1mA. The UL spec calls for 100M insulation resistance, and I figured it would be easier to pass that than negotiate higher leakage current. Yesterday I swapped out the 1M input resistors with 100M. I had to make other circuit tweaks and add another part*** but I got it working with 100M input resistance, sweet. :D :D :D

Another UL spec I need to work around is a 2 mA max current to the ground lead. Again that supports the typical neon lamp current used in the cheap tester. I can get by just barely with 2mA using a LED. They make green neon lamps but I suspect they are more expensive and hard to source.

A few of the traces in my current PCB don't meet the UL spacing spec, so I need to do another cut (and to add a part). Another thing I need to evaluate, they test for the 100M insulation resistance with a 500V magneto, I'm using 300V mosfet and diodes, so while I should still pass the insulation test the circuit may be damaged by the 500V test. I need to find out from UL if that matters (and exactly how they test). If it has to work after the 500V test I may need higher voltage circuitry.

Next step is to see if the $400 spec buys me something. I expect to pay for them actually testing my product but answers related to that testing should be free or reasonable.

It may be awkward that my tester doesn't need all the fine print UL caveats... still needs one or two (can't parse ground-neutral swap or ground quality).

JR

PS: Still no answer from the company I tried to partner with so I am resigned to probably doing this myself. I plan to use an off the shelf line cord plug as a housing and base, so that should already be UL listed and simplify that part of the design approval. I will probably approach the plug maker to see if they want to work together. Should be a no brainer for them, especially if I have already done the heavy lifting, but I'm not counting on that either.

*** Sorry I decided a couple months ago to stop sharing specific circuit details, since I am not going to patent this (too much cost for too little benefit).

User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

Re: modified GFCI

Post by JR. » Tue Nov 10, 2015 12:59 pm

Thanks, I can probably rig up a 500V supply (carefully) if needed. A 20VAC transformer backwards gets me there and more (assuming the transformer insulation can handle the voltage).

In fact 500V/100M is only 5uA so even if the semiconductors breakdown at 500V the 5 uA leakage (2.5mW) may not permanently harm the semiconductors.

I just dug into the UL spec and as usual their descriptions of the test procedure is vague but they suggest 3 separate tests using 3 product samples each. The functionality test is separate from the insulation test, and a third test for accelerated aging.

This suggests to me that the tests are performed concurrently on different sample sets in a specific sequence.

Based on this it looks like the 2 mA to ground test is performed "after" the 500V insulation test, so semiconductor failure could have an impact on that. maybe.

I may need to test this... (time to research semi failure modes).

Thanks, I'll let you know if I need to borrow a tester... I am trying to spend as little as I have to as this will not be cheap to pay them to test. Of course it will be cheaper if I know that I am going to pass first time, and can avoid the expense of repeat testing.

JR

ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: modified GFCI

Post by ricardo » Wed Nov 11, 2015 1:48 am

JR. wrote:Thanks, I can probably rig up a 500V supply (carefully) if needed. A 20VAC transformer backwards gets me there and more (assuming the transformer insulation can handle the voltage).
Be VERY careful about doing that.

Iron transformers have core & windings sized to start saturating near the 'working' voltage. To support a higher voltage needs a bigger core and/or more windings.

Putting 60Hz 110V onto a 20V winding will saturate big time, 'magnetising current' (current with no load on the transformer) would go through the roof, probably blow fuses and other nasty stuff. :o ... that's in addition to the insulation and creepage & clearance (sparking) issues.

You could do it with a 110V/20V x 60Hz = 330Hz 110V oscillator as the Flux (and core+winding) requirements are inversely proportional to frequency :mrgreen:

User avatar
mediatechnology
Posts: 3276
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: modified GFCI

Post by mediatechnology » Wed Nov 11, 2015 9:29 am

Richard - You seem to be speaking from experience. Any stories to share? :lol:

Reminds me of a story of an SSL organ repair...

It had huge bulk electrolytics and a tech came in to replace a bad one.
The organ was powered-up.
After it was replaced and quite proud of his repair - with his back to the instrument - the organist politely asked the tech: "Excuse me sir is the organ supposed to be on fire?"

The transformer visuals in ricardo's post made me think of that story.

Sorry for the veer John.

User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

Re: modified GFCI

Post by JR. » Wed Nov 11, 2015 10:51 am

ricardo wrote:
JR. wrote:Thanks, I can probably rig up a 500V supply (carefully) if needed. A 20VAC transformer backwards gets me there and more (assuming the transformer insulation can handle the voltage).
Be VERY careful about doing that.
:lol: :lol: :lol: :oops:
Iron transformers have core & windings sized to start saturating near the 'working' voltage. To support a higher voltage needs a bigger core and/or more windings.

Putting 60Hz 110V onto a 20V winding will saturate big time, 'magnetising current' (current with no load on the transformer) would go through the roof, probably blow fuses and other nasty stuff. :o ... that's in addition to the insulation and creepage & clearance (sparking) issues.

You could do it with a 110V/20V x 60Hz = 330Hz 110V oscillator as the Flux (and core+winding) requirements are inversely proportional to frequency :mrgreen:
OK I need to KISS so how about plan B? I have several transformers sitting around, so if I wire 4x transformers with the secondaries in parallel but with the primaries in series.

Apply 120V to bottom primary, if I get the winding polarity correct the the other 3 primaries should combine(like an auto-former) to make roughly 480VAC. All the windings should be in their happy place regarding saturation or insulation breakdown (well top transformers have higher than design voltage between windings, but not spark plug high voltage.) The 20V windings can be floating so not sure where they will end up.

==========
I asked UL for clarification about how they perform the test... one end of the megger 500V probes the housing, and other end one or more of the plug contacts, Individually or all at the same time could make a difference.

I have looked at upgrading to 500V diodes and MOSFET but they are physically bigger and more expensive.

I need to school myself more on semiconductor failure modes. Just because a diode or mosfet avalanches due to over-voltage, if that breakdown is not accompanied by high current and localized heating, their may not be any permanent damage. Since my 500V stress is effectively in series with 100M the current and heat will be minimal.

My gut tells me the 300V parts will be OK, but my gut also said to reverse the 20V transformer. :lol:

JR

User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

Re: modified GFCI

Post by JR. » Thu Nov 12, 2015 5:00 pm

back.jpg
back.jpg (30.26 KiB) Viewed 11210 times
side.jpg
side.jpg (60.74 KiB) Viewed 11210 times
Found what looks like a good package. The plug is already UL approved so I don't have to go through that again...

The attachment screw in the middle of the back can be the touch probe contact.

Time to start on rev 3 PCB layout....

JR

User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

Re: modified GFCI

Post by JR. » Mon Nov 16, 2015 11:45 am

I think I have reached a reasonable solution for surviving the 500V megger test. While my suspicion is it would probably already survive because the current even after device breakdown is so small (uA), but by adding just one more 300V diode in series with a common node for three other diode strings that 500V is now across 2x300V diodes so they should all survive. The other polarity has a shunt resistance across the MOSFET that will effectively divide that 500V in half exposing the 300V MOSFET to only 250V. :D

Any or all mains power leads should survive 500V (to tester case and my input probe). The circuit would of course have some difficulty with 500V line to neutral or line to ground, but they do not get tested that way AFAIK.

JR

PS: I finished my 3rd cut PCB layout without a single via,,,, While I may add one or two to increase spacing.

Post Reply