DIY Thoughts on Building a Grid Backfeed Prevention Device

[ricardocello]

I’ve been considering the issue where even short surge amounts of grid backfeed are causing people to get a knock on the door from the PoCo. We all know “zero-export” inverters do not limit backfeed power to zero, but they try. For those of you who don’t believe this, please search the forum.

I’ve looked for devices that will actively monitor the current direction and take an action when reverse current flow is indicated. The only thing I’m finding are “Reverse Power Relays” meant to prevent generators from motoring. I’ve even seen schematics for these, and they are all cheap low-performance analog circuits. I believe a modern digital approach could yield a high-performance solution.

It should be easily possible to sample voltage and current simultaneously using a CT, and calculate both power factor and current flow direction on a per-cycle basis. Sampling rate should be like 240 Hz (or 200 Hz for 50Hz countries) to capture the zero crossings cleanly. I’m aware that CTs have a time constant, and that averaging over multiple cycles will be necessary to minimize noise. Some calibration will likely be necessary. For the sake of argument, let’s say this is all feasible, and I’ve constructed such a detector in a microprocessor with thresholds for power and duration of reverse transient.

What action should be taken when the reverse current flow detector activates?

• Disconnect the mains via a contactor or relay?
Where will the excess power go, back into the battery (hopefully)?

• Switch in a resistive load (could be as high as 5 kW) for the second or two to absorb the power before it goes to the grid?
This would need thermal and timeout monitoring for safety, and would be as big as those huge resistive loads Will was using.

• Any other ideas?

 

[SparkyJJO]

Isn't that the point of SBU mode? SBU has the grid input relay open so there's no way it can backfeed, unless I'm misunderstanding somehow.

 

[ricardocello]

Isn't that the point of SBU mode? SBU has the grid input relay open so there's no way it can backfeed, unless I'm misunderstanding somehow.

That's right, but I want to have my cake and eat it too!

In other words, I want the flexibility to not rewire my entire house and get a true zero-export capability.
It also helps to avoid buying 2x the batteries that will have to be heavily cycled daily.

 

[BentleyJ]

Isn't that the point of SBU mode? SBU has the grid input relay open so there's no way it can backfeed, unless I'm misunderstanding somehow.

Yes, but that is limited to Off-Grid type inverters. With a UL1741, grid interactive inverter, there is blending of utility with inverter power. It takes a moment for the inverter to respond for example when a large load is turned off there could be some momentary back-feed.

 

[Rednecktek]

2 big diodes and a heat sink? Current can flow in from the grid but the diodes block any current heading out?

 

[agtcovert]

I should've taken some EE classes. So when this sounds utterly moronic, sorry.

<pay no attention to the computer science guy>
Would it be possible to route inverter or grid power through a capacitor instead of the direct path? Essentially, whatever mode you're in feeds the capacitor feeds the load, based on input from the CTs, etc. Even if there's a ramp down, your load is pulling from the cap. The inverter can finish dumping power (giving it a place to go) and then adjust again based on input from CTs, etc.

The other option you mentioned is interesting, essentially opening a relay when load changes. Cycle time would be critical, and I wonder if the wire length of the CTs would inhibit the desired response time.
</<pay no attention to the computer science guy>

 

[ricardocello]

2 big diodes and a heat sink? Current can flow in from the grid but the diodes block any current heading out?

Diodes are great for DC, but AC goes both ways.

Actual power direction is determined by the phase offset of AC current and voltage, so it is a lot more challenging.

 

[SparkyJJO]

That's right, but I want to have my cake and eat it too!

In other words, I want the flexibility to not rewire my entire house and get a true zero-export capability.
It also helps to avoid buying 2x the batteries that will have to be heavily cycled daily.

You don't have to heavily cycle the batteries daily. Set your upper and lower charge limits so they don't cycle very far at all if you want under normal circumstances (i.e. switch to grid when battery gets to 60%, only charge up to 90%, or whatever). Besides, even if you did heavily cycle them, it seems with LFP you're going to calendar age them by the time you cycle age them.

Yes, but that is limited to Off-Grid type inverters. With a UL1741, grid interactive inverter, there is blending of utility with inverter power. It takes a moment for the inverter to respond for example when a large load is turned off there could be some momentary back-feed.

Don't have to run a grid interactive inverter, though if you don't want to install a sub panel I guess you kinda have to. I get the niceness of not having to do it, and having it blend power seamlessly (no "blip" switching from solar/battery to grid and back since it is all in sync), but if you want absolutely zero chance of backfeed then I don't know a better way.

The other thing though - go look at your meter. Does it even have the capability? My old digital display meter on my house - which really wasn't very old, it was installed in 2018 - only did one direction. I could (and did by accident) - run grid interactive, and did backfeed. I got charged for it lol (my first poco bill with solar running was higher than the month before... hahh..). I misunderstood and thought I had permission to operate and I didn't yet. They swapped my meter to a bidirectional meter, then my bill dropped. Anyway, my point is, if your meter isn't bidirectional, then a tiny leak on occasion won't make any difference.

 

[ricardocello]

I should've taken some EE classes. So when this sounds utterly moronic, sorry.

All ideas are good in this thread.

Would it be possible to route inverter or grid power through a capacitor instead of the direct path? Essentially, whatever mode you're in feeds the capacitor feeds the load, based on input from the CTs, etc. Even if there's a ramp down, your load is pulling from the cap. The inverter can finish dumping power (giving it a place to go) and then adjust again based on input from CTs, etc.

I haven't considered capacitors yet, but switching one into the circuit when backfeed is detected would probably work too. I like this idea better than huge resistive loads.

The other option you mentioned is interesting, essentially opening a relay when load changes. Cycle time would be critical, and I wonder if the wire length of the CTs would inhibit the desired response time.

Let's say the relay could drop in 10 msec (which is pretty common I believe).
I think the limiting factor on time will be the response of the CTs and the averaging necessary to get a reasonable measurement.
I'm not sure if "one-cycle" (16.66 msec or 20 msec) detection is a reasonable goal or not.

One effect of opening the relay is that the inverter would immediately go into inverter mode, as grid power is dropped.

 

[Mattb4]

It would be much easier to have a policy based solution versus a technical one. If small amounts of spurious export were acceptable than a zero export AIO would meet the needs. The electric companies are being too restrictive.

 

[agtcovert]

I'm working on a different system for inside my house, but I was testing some different modes on the Victron gear in my RV. (My wife loves that I've turned it into my personal testbed when we're not using it).

Anyway, I kinda accidentally backed around 0.9kwh according to VRM. I was expecting to hear about it. The meter that feeds our shore power is new (last 4 years). But, I haven't heard anything. Can't tell from the billing portal.

That said, I know the issue exists, and I know people have gotten knocks. What is odd to me is that spurious feedback shouldn't be a big deal. Unless the power is out, linemen would know for certain the lines are energized. The critical piece is anti-islanding. As soon as loss of power is detected, the inverter needs to open a relay. And as far as I know, most do.

So at that level, I'd be interested to know the ramifications of some spurious feedback and why some pocos care so much.

 

[Rednecktek]

Diodes are great for DC, but AC goes both ways.

Actual power direction is determined by the phase offset of AC current and voltage, so it is a lot more challenging.

Right, but it goes back along the Neutral and the meter reads only the incoming hot lines, so any backfeed between the 2 hot legs would be blocked, any backfeed on 120v would feed on the unmeasured neutral line. The goal of anything going towards the PoCo meter being stopped would still be stopped, anything going from the meter to the loads would pass right through the diodes. Right?

Yes, AC won't flow back and forth through a diode, but isn't that the whole idea?

 

[BentleyJ]

Anyway, my point is, if your meter isn't bidirectional, then a tiny leak on occasion won't make any difference.

That is true with many of the so called smart meters. Our single channel meter got changed out to a 2 channel meter when the solar system was approved. Same as your experience.
That said, it depends on the meter programming. There is nothing preventing the utility from setting up the meter to detect back-feed even though they are not going to give credit and may even charge for it.

 

[ricardocello]

You don't have to heavily cycle the batteries daily. Set your upper and lower charge limits so they don't cycle very far at all if you want under normal circumstances (i.e. switch to grid when battery gets to 60%, only charge up to 90%, or whatever). Besides, even if you did heavily cycle them, it seems with LFP you're going to calendar age them by the time you cycle age them.

You are certainly right abut that.
My objection to the heavy cycling is that at some point in the day, the batteries will be at 20% SoC.
That will be the exact time the grid will go down, and my wife will wonder why we paid all of this money for a 20 minute backup system.

It would be much easier to have a policy based solution versus a technical one. If small amounts of spurious export were acceptable than a zero export AIO would meet the needs. The electric companies are being too restrictive.

No argument there at all!
But this is a DIY site, so I'm trying to DIY it.

 

[ConnerLabs]

The only thing I’m finding are “Reverse Power Relays” meant to prevent generators from motoring. I’ve even seen schematics for these, and they are all cheap low-performance analog circuits. I believe a modern digital approach could yield a high-performance solution.

What exactly is wrong with cheap low-performance analog? I see analog, I think simple, reliable, no buggy firmware.

 

[1201]

@ricardocello good post. if this issue could be solved, then guerilla solar can be fixed and adding a small inverter would become easier than ever.

the concern about relays is they could have a limited lifetime based on activations.
I like the idea of a sink. It would have to be a very large one.

this is one I would like to see solved.

 

[ricardocello]

Does it even have the capability? My old digital display meter on my house - which really wasn't very old, it was installed in 2018 - only did one direction. I could (and did by accident) - run grid interactive, and did backfeed. I got charged for it lol (my first poco bill with solar running was higher than the month before... hahh..). I misunderstood and thought I had permission to operate and I didn't yet. They swapped my meter to a bidirectional meter, then my bill dropped. Anyway, my point is, if your meter isn't bidirectional, then a tiny leak on occasion won't make any difference.

I'm in a temporary state of grace as my 1993 meter is a spinning wheel with ratchet and limited electronic reporting.
I watch it go backwards briefly on a regular basis (never more than 1 second fortunately).

When the PoCo comes around to eventually replace it, I need a plan.
One approach is to attempt net metering, but I already know they won't approve Victron.

 

[SparkyJJO]

That is true with many of the so called smart meters. Our single channel meter got changed out to a 2 channel meter when the solar system was approved. Same as your experience.
That said, it depends on the meter programming. There is nothing preventing the utility from setting up the meter to detect back-feed even though they are not going to give credit and may even charge for it.

Well depends on the meter capability. They switched my meter out because the original 2018 one COULDN'T be programmed to go both ways. If you look up your meter and it may be able to, then be wary. If you look it up and find (like my old one and your old one) that it can't detect anything, then there is no risk.

 

[ricardocello]

What exactly is wrong with cheap low-performance analog? I see analog, I think simple, reliable, no buggy firmware.

Nothing! I love simple analog solutions, they are more robust.

But those circuits allow between 2-20% reverse current to get through.
They aren't sensitive enough to get down to single cycle measurements.
The filtering is crude. Maybe a great analog designer could improve it.
I'm a digital guy.

 

[agtcovert]

@ricardocello good post. if this issue could be solved, then guerilla solar can be fixed and adding a small inverter would become easier than ever.

the concern about relays is they could have a limited lifetime based on activations.
I like the idea of a sink. It would have to be a very large one.

this is one I would like to see solved.

This is really an interesting idea.

Inverter output -> Sink -> out to load center
Then the sink is making decisions and is essentially and overflow pool. As long as it was reliable.

I was wondering how you'd fit something into an ecosystem. Everyone's setup is unique in the DIY community. Would be nice to have something, if possible, at a reasonable price point that can be integrated anywhere. Kind of like an AC soft start. Don't need to buy a whole unit with it. (Though, I don't understand why they just don't come with that from the factory).

 

[ricardocello]

I was wondering how you'd fit something into an ecosystem. Everyone's setup is unique in the DIY community. Would be nice to have something, if possible, at a reasonable price point that can be integrated anywhere. Kind of like an AC soft start. Don't need to buy a whole unit with it. (Though, I don't understand why they just don't come with that from the factory).

It's probably a unique product concept, but I can't imagine the market is very large, no offense to the forum members intended!
Maybe if the draconian PoCos start cracking down such a device becomes more valuable.
I may do a quick prototype if I can find the time.

 

[1201]

This is really an interesting idea.

Inverter output -> Sink -> out to load center
Then the sink is making decisions and is essentially and overflow pool. As long as it was reliable.

I was wondering how you'd fit something into an ecosystem. Everyone's setup is unique in the DIY community. Would be nice to have something, if possible, at a reasonable price point that can be integrated anywhere. Kind of like an AC soft start. Don't need to buy a whole unit with it. (Though, I don't understand why they just don't come with that from the factory).

in my perfect world it would be a device on the main load center that instantly diverts power to a battery when it sees one cycle of export or senses a large load shutting down. i dont know if thats possible, just dreaming

 

[ricardocello]

in my perfect world it would be a device on the main load center that instantly diverts power to a battery when it sees one cycle of export or senses a large load shutting down. i dont know if thats possible, just dreaming

That's why I posted in the "General Discussion and Brainstorming" section. Dreaming is encouraged.
I agree with your thinking, it should be inline at the main load center.

 

[agtcovert]

in my perfect world it would be a device on the main load center that instantly diverts power to a battery when it sees one cycle of export or senses a large load shutting down. i dont know if thats possible, just dreaming

Hm, I wonder if that could be something similar to a whole house surge suppressor. Those are hooked into a double pole 50A breaker (mine is anyway). The energy would still need a place to go that's not 'out'. Be interesting to take that concept and have a double pole breaker on the 2 busses, and then a set of wires on the device that's outside the panel that you hook to your main system bus bar. (I say that thinking about the Victron Lynx ecosystem because, while I am doing my home system with other inverters for various reasons, most of the DC stuff I have will be blue). At least within that ecosystem, you could route it to the distributor. It'd be fused, and I think the shunt would be able to see and account for the power.

Alas, I am not versed enough to know if and how such a thing would work.

 

[ricardocello]

I should point out that whatever detection circuit is designed, it needs to be faster than the circuit used by the PoCo smart meters to detect grid backfeed. That will be hard to know in advance, obviously.

Adding the time to switch away or add a load on top of the detection time may make this device infeasible.
But that’s why we prototype things.