LV Per-circuit GFCI?

[Joekovar]

I'm in the process of ditching all of my power bricks and moving everything to redundant Mean Well NDR series power supplies.

One of the things I haven't liked prior to this is the way GFCI protection has worked. Unless I setup a mess of GFCI outlets, I was stuck with a single point of failure where one bad piece of equipment shuts down at best a significant portion of the system. In 4 years I only had it happen once, but it happened and I was fortunate to be around when it did.

So with this update, I'm looking to setup device level GFCI, that'll cut off only the offending piece of equipment. On the LV side of the power supplies.

Now, before I get too far down this rabbit hole of hall effect current sensors, comparators, 555 timers, latches, and relays, has anyone seen LV GFCI devices that aren't a couple hundred bucks per circuit?

 

[magallonbuckinbulls]

They are pricy. We use DIN style type on our control panels.

The GFCI should prevent the breaker from tripping. Check your wiring. Make sure the loads are rated properly.

 

[Joekovar]

The issue I had was with the GFCI itself cutting off multiple pieces of equipment, breakers were fine. Sure I couldve wire up a strip of GFCI outlets and made sure each tank uses at least 2 separate GFCI outlets for critical equipment, but that seems overly complicated.

The LV din rail stuff is certainly pricey. I started looking around and didn't feel so bad about the time I had to put $30 GFCI outlets in every receptacle of the kitchen to appease the electrical inspector when we sold a house. (Kitchen was wired screwy when we moved in)

 

[Daveobrien]

I like this conversation. I fed my fishroom with four different 20A circuits, to be able to spread risk around. Two of them feed a GFCI protecting large power strips. Another feeds GFCI outlets, plumbed closest to the pumps. The last was for GFCI tripping equipment, like my chiller. I've tried to spread lights, return pumps, recirc pumps, etc across the three circuits, so if one goes - the whole system isn't a goner. In 20yrs, I've replaced the GFCI's three times - when they start acting up.

You're at a different level of thinking than I am... But I LIKE the idea of a bank of GFCI's, instead of one GFCI per power strip.

You can see my three circuits in this photo. Top Left corner, you'll see lighted plugs in each GFCI outlet, as a visual indicator it hasn't tripped. and you can see the two silver power strips across the back wall, each one is a separate circuit.

 

[Joekovar]

Ultimately my goal is to move completely to LV. While a lot of people don't like SLA/AGM batteries, I have for all intents and purposes an unlimited supply of them through work so they're extremely cost effective for me when it comes to backup power.

In order to do this, I have to get away from AC as much as possible. Doing so however brings me back to a single point of failure where my combined power supplies still have to convert AC to DC. Technically, the typical AC GFCI protection is going to be moot when the system is running on batteries now that I think about it, so I'm kinda stuck with DC side GFCI.

I'm leaning towards designing a protection PCB myself at the moment. Currently I'm running my LV from the NDR supplies to a distribution block, then branching individual device circuits out through DIN mounted fuse blocks with 1.5x rated fuses for each individual piece of equipment with some smart relays in the middle. It'd make sense to just fuse the individual inputs of a sort of GFCI distribution board.

 

[Joekovar]

I forgot about the chiller.

That, and a heater is going to be tough to do DC.

I imagine someone's eventually going to make an MP40 style induction heater that runs on DC and keeps all the electrical out of the water (I started, but hit a road block when I looked into using chemical vapor deposition to coat inconel in a thin layer of diamond for a wet side...)

I don't know about the chillers. I've never done much with compressors, horsepower, etc. Though Milwaukee does make an 18VDC air compressor, so maybe a DC aquarium chiller isn't so far fetched.

I read someone's making a DC skimmer that can run on as little as 2W the other day, which kinda blew me away.

 

[Daveobrien]

Ultimately my goal is to move completely to LV. While a lot of people don't like SLA/AGM batteries, I have for all intents and purposes an unlimited supply of them through work so they're extremely cost effective for me when it comes to backup power.

In order to do this, I have to get away from AC as much as possible. Doing so however brings me back to a single point of failure where my combined power supplies still have to convert AC to DC. Technically, the typical AC GFCI protection is going to be moot when the system is running on batteries now that I think about it, so I'm kinda stuck with DC side GFCI.

I'm going to ask an obvious question. Assume you switch your system to be all DC fed. Are you SURE you need a GFCI? 110v AC will kill you, and if not, sure as heck hurts to get zapped by it. (I'm ignoring the chiller&heater challenge)

I thought the risk with home 110v AC circuits is Black wire is hot, and current is supposed to be returned via White/Neutral wire... Which is also connected to earth ground. My understanding of risk is Hot being in contact with water, and you or I reaching into the tank, and having our body act as the return path going back to earth ground - instead of going through the white/neutral wire. I don't think this happens with DC, right? No earth ground.

A 65W DC pump needs, ~3A at 24v, ~2A at 36v. (most DC pumps I'm aware of are fed by 24v or 36v)

So the topic here is really about Amperage being fed out by each DC circuit, in the event of a short in the tank water, right? Isn't this the primary risk to your health?

(Apply all safety caveats here - I'm asking questions, from the perspective of not being an expert.)

 

[magallonbuckinbulls]

"But I LIKE the idea of a bank of GFCI's, instead of one GFCI per power strip"

Me too, a like to isolate my equipment.

Also adding grounding rod to system. Frame will be grounded too.

Remember, Amperage is the ones that kills.

 

[Joekovar]

I'm going to ask an obvious question. Assume you switch your system to be all DC fed. Are you SURE you need a GFCI? 110v AC will kill you, and if not, sure as heck hurts to get zapped by it. (I'm ignoring the chiller&heater challenge)

I thought the risk with home 110v AC circuits is Black wire is hot, and current is supposed to be returned via White/Neutral wire... Which is also connected to earth ground. My understanding of risk is Hot being in contact with water, and you or I reaching into the tank, and having our body act as the return path going back to earth ground - instead of going through the white/neutral wire. I don't think this happens with DC, right? No earth ground.

A 65W DC pump needs, ~3A at 24v, ~2A at 36v. (most DC pumps I'm aware of are fed by 24v or 36v)

So the topic here is really about Amperage being fed out by each DC circuit, in the event of a short in the tank water, right? Isn't this the primary risk to your health?

(Apply all safety caveats here - I'm asking questions, from the perspective of not being an expert.)

I'm by no means an expert, but this is my understanding.


On an AC system a GFCI does two things. One, watches for an imbalance of current between the hot and neutral, two, looks for unexpected current on the neutral.


The first is indeed obvious, 100mA going out on the hot leg should translate to 100mA returned on the neutral. Unless the current has found an alternative path to ground.


The second isn't so obvious, because it's only going to happen if wiring elsewhere is screwy. You can run into situations where two different breakers are sharing one neutral. There may not be any current flowing between the outlet you're working on and neutral, but if that neutral is shared with a second breaker, that neutral is effectively a continuation of the second breakers hot line. If there's an active load on that second breaker, and you touch that shared neutral, you become an alternative path to ground for that second breaker. That's why breakers that share a neutral are supposed to be tied together. Ideally, breakers shouldn't be sharing a neutral.


There are two applications of DC systems. Floating, which is your battery powered systems that are self contained, and there's your non floating systems which are your systems that use rectification to get DC from AC and transformers to step the voltage down. With a non floating system, that DC current is ultimately looking for a path back to ground, since it's being derived from a grounded AC system to begin with. Looking at the block diagram in my particular power supplies datasheet, the positive and negative rails are tied to ground through capacitors.


With a non floating DC system, which is what my system will be during normal operation, if I'm in the water with a faulty piece of equipment, and I bump into for instance a lights grounded metal enclosure, I'm going to complete the circuit.


As I think through this, I'm starting to doubt whether ground fault protection will be of any use when the system fails over to battery and becomes a floating system. I'll be monitoring the AC side of the power supplies, and when they lose power, relays will disconnect the power supplies from the circuit, and switch over to batteries. Effectively removing the ground reference from the system


I believe I would need a faulty positive in one tank, and a faulty negative in a separate tank, on two pieces of equipment on the same power supply to complete a floating circuit in a way individual conductor current monitoring would catch. Though I suspect since I can't reach any two tanks that aren't sharing plumbing at the same time it's unlikely to be an issue.


I'm not too concerned about fire hazards. My NDR power supplies have overload protection and I'm putting fuses rated for 1.5x each individual piece of equipments normal draw on their respective circuits. My supply to bus feeds are 12awg, and individual circuits are 16awg carrying max maybe 2A 20 feet.


Though I might rethink some of this when I'm able to move equipment not in the water over to DC.

 

[magallonbuckinbulls]

I'm by no means an expert, but this is my understanding.


On an AC system a GFCI does two things. One, watches for an imbalance of current between the hot and neutral, two, looks for unexpected current on the neutral.


The first is indeed obvious, 100mA going out on the hot leg should translate to 100mA returned on the neutral. Unless the current has found an alternative path to ground.


The second isn't so obvious, because it's only going to happen if wiring elsewhere is screwy. You can run into situations where two different breakers are sharing one neutral. There may not be any current flowing between the outlet you're working on and neutral, but if that neutral is shared with a second breaker, that neutral is effectively a continuation of the second breakers hot line. If there's an active load on that second breaker, and you touch that shared neutral, you become an alternative path to ground for that second breaker. That's why breakers that share a neutral are supposed to be tied together. Ideally, breakers shouldn't be sharing a neutral.


There are two applications of DC systems. Floating, which is your battery powered systems that are self contained, and there's your non floating systems which are your systems that use rectification to get DC from AC and transformers to step the voltage down. With a non floating system, that DC current is ultimately looking for a path back to ground, since it's being derived from a grounded AC system to begin with. Looking at the block diagram in my particular power supplies datasheet, the positive and negative rails are tied to ground through capacitors.


With a non floating DC system, which is what my system will be during normal operation, if I'm in the water with a faulty piece of equipment, and I bump into for instance a lights grounded metal enclosure, I'm going to complete the circuit.


As I think through this, I'm starting to doubt whether ground fault protection will be of any use when the system fails over to battery and becomes a floating system. I'll be monitoring the AC side of the power supplies, and when they lose power, relays will disconnect the power supplies from the circuit, and switch over to batteries. Effectively removing the ground reference from the system


I believe I would need a faulty positive in one tank, and a faulty negative in a separate tank, on two pieces of equipment on the same power supply to complete a floating circuit in a way individual conductor current monitoring would catch. Though I suspect since I can't reach any two tanks that aren't sharing plumbing at the same time it's unlikely to be an issue.


I'm not too concerned about fire hazards. My NDR power supplies have overload protection and I'm putting fuses rated for 1.5x each individual piece of equipments normal draw on their respective circuits. My supply to bus feeds are 12awg, and individual circuits are 16awg carrying max maybe 2A 20 feet.


Though I might rethink some of this when I'm able to move equipment not in the water over to DC.

That music to my ears. Good job breaking it down.

 

[Joekovar]

Here's the idea at the moment.

I'm inclined to stay away from microcontrollers here.

A single 24V DC input pair of terminals fed from a power supply.

From there to a 5V regulator for detection circuitry, and branched out to individual protected circuit traces. I need to do more research on trace current carrying capability. Wire I know, PCB traces not so much.

The positive and negative of each branch shall route through hall effect current sensors. I have my eye on the 5A version of the ACS724 at the moment because of it's 0.8mV/mA sensitivity.

From there the positive and negative shall route through the N/O contacts of individual 5V mechanical relays and out via terminals. I'm considering using the release time of the relays as the base delay for activation.

On the 5V side, the outputs of the current sensors shall route to linear window comparators for each branch. The branch's negative going to the lower inverting input, and the positive to the upper non-inverting input. The idea being that the comparator output will drive high as long as the positive current sensor voltage is greater than the negative sensor voltage (inadvertent reverse polarity and current leakage via the negative rail protection), yet still below the top end of the comparators difference threshold. That difference being the maximum current difference between the legs before it's considered a fault. (Aiming for 30mA max)

That comparator output may route through voltage follower op-amps to drive the 5V mechanical relays, depending on the comparator used.

The detection circuitry shall route through individual momentary push buttons. The buttons shall activate and latch the detection circuits. When a fault is detected, that latch shall be released, disabling the detection circuits ability to drive the relays. I haven't put much thought into how to accomplish that yet. I may just try and flip the latch over to a fault
indicator LED.