diy solar

diy solar

Incrementally Adding AC Batteries

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
7,199
Location
Key Largo
TL;DR: Now that we've solved (?) Incrementally adding DC batteries, how can you add AC batteries from different manufactures or a DIY AC battery?

Since an AC battery has its own inverter/charger you don't have any of the internal resistance problems you do with DC batteries.

Instead, the problem is trickier. Basically, it's the same problems you see with AC Coupling.

In order for an AC battery to work, it must be AC coupled to the roof array's inverter. This is a problem for grid-tied inverters as they act as charge sources rather than a normal inverter which is a voltage source. A voltage source is what most people use every day and are familiar with (e.g., batteries, a wall socket, an off-grid inverter). A device connected to it draws amps and the source tries to hold the voltage steady. A current source is the opposite of that, it varies the voltage in order to push the available power as current. So if your house's grid power is 122V, a grid-tied inverter might up the voltage to 125v or so in order to push its current through the house and out onto the grid. The grid acts like a huge sponge so can absorb what the panels are cranking out. If the panels aren't able to supply enough current the voltage drops back allowing the grid to make up the difference. When the grid goes out grid-tied inverters shut down, otherwise, they'd supply power to the grid (a legal issue).

An AC Coupling solution allows storage to work with a grid-tied system. Basically, it's a smart inverter (aka hybrid inverter) that can control the power from grid-tied inverter either by shifting the frequency to request the inverter to throttle back output, shut them off altogether, or some other means. The inverter also needs to be able to rapidly handle surges from the solar panels, e.g., the solar is pumping 5 kW into the house to power the AC and the AC shuts off. Generally, this is done by keeping a large battery bank that the surge can flow into.

So, what's the actual problem?
The problem might be that there can only be one "brain", and that whatever you pick in the beginning you're stuck with. Let's look at an example...

Imagine you install a Tesla or Outback hybrid inverter and setup the AC coupling solution. Great!

But next year you want to come back and add another 10 kWh. If you had an Outback you could just add more batteries to the DC side and all done.

But what do you do if you had a Tesla or IQ8? Are you always locked into Tesla or Enphase which are controlled by the Gateway or Envoy controllers? That is, the brain tells the AC battery's inverter what to do; but that brain can only work with batteries it knows how to talk to.

Or, is there a way to add your own DIY AC Battery? It seems like you'd need a synchronous inverter that also operated as a matching voltage source rather than a current source (which removes all existing grid-tied inverters) or one that was smart enough that it could play nicely with other charge sources.

Anyway, that's the boat I'm in. I'll have Enphase IQ8s and Envoy, but would like to be able to add a DIY AC battery in the future.

WARNING/UPDATE
Enphase does use frequency shifting from the grid to throttle its microinverters. But as that's relatively slow the IQ8s use high-speed IP over powerline for throttling when offgrid; so much of the "hypothetical" discussion in this thread is incorrect.
 
Last edited:
Both Tesla and Enphase's brain communicate to the battery... why is that... let's use a visual aid... so thinking out loud....

Capture.PNG

Let's say the AC batteries switch to charge mode over 120V and discharge below that. Okay, the grid goes out and the sun is shining.

The "Brain" will throttle the inverters to supply the needed load, if there isn't enough power then the voltage will start to sag and the batteries will cut in providing supplemental power. When the cloud goes away, the voltage picks up and the batteries go back into recharge. Now, if the inverters are powering the air conditioning and it cuts off, it takes a bit for the brain to throttle back the inverters - so that surge would need to be funneled back to the battery. AH! The brain would have to know the Batteries state of charge. Basically when the power first went out it would have to drain the batteries a bit to make sure there was head-room for surges.

Assuming that's the case (no guarantees) then since the inverter is grid-tied, it'll sync to the Original AC batteries frequency so that inverter just needs to be like a regular off-grid inverter - it can operate as a voltage source (not a normal off-grid inverter since it has the whole charge/recharge voltage thing going on).

Does that sound reasonable or is there a huge faux pas?

To add another AC battery from the same manufacturer probably just requires a sync cable between them. But to add a DIY "New AC battery", the inverter in it must synchronize to the original AC batteries frequency; but other than that it could operate as a voltage source too.

So, for the DIY AC battery bank we'd need a synchronous inverter that operated as a voltage source below 120 and automatically went into recharge above 120. Anyone know of anything like that on the market?
 
Last edited:
The "easy" solution is to add a "normal" voltage source inverter with an ATS and critical circuits subpanel as this requires no synchronization of load:

Capture.PNG
But this solution means the amps from the new inverter don't combine with those from the Original AC battery; It also means the second bank won't kick on until the first bank is exhausted. An ideal solution would draw from both to minimize DoD and C-Rate.
 
Could CA Rule 21 Resolve this?

CA Rule 21 allows the grid to control the output of the inverter.

So, if the brain is measuring power from the solar and batteries, it might work. Don't really know enough about how the "brain" is working for any given technology, but it might be possible to do something like this:

Capture.PNG
Might be some problems with this. For the original AC battery to work correctly it would need to handle the scale of the power introduced by the new battery (ideally to it, the new battery "looks" like more solar panels with the same response times (or faster) than panels.).

Update: IEEE 1547 requires DERs (e.g., a PV or wind system) to trip at 60.5 Hz for 300s. The EU has a similar regulation ENTSO-E.
Update: I suspect the Envoy is not using frequency shifting to throttle back the microinverters, it would be to slow...It's probably using it's Ethernet over Powerline communication. From their AC coupling guide, there's a minimum of 40 ms ramp down time.
 
Last edited:
Bit of a breakthrough! While researching this stuff I ran across this bit:

Because renewable energy sources generate DC or variable AC voltages, they require an inverter to interface with and help regulate or “form” the microgrid. Accordingly, these inverters are called grid forming inverters (GFI). Unlike grid-tie inverters (GTI) which typically interface to a utility grid with well-regulated voltage and frequency, GFIs are tied to microgrids where they are required to assist in the regulation of both voltage and frequency.

Since several different renewable energy sources typically feed a single microgrid, GFIs must be capable of operating in parallel. In essence, these grid forming inverters must be designed as parallelable voltage sources with very good load sharing capability, while maintaining a stable AC output voltage and frequency with varying loads.

Frequency regulation is typically achieved by controlling real power (kW) while voltage regulation results from reactive power (kVAR) control. Furthermore, droop control methodologies can be employed to allow GFIs to operate in parallel without the need for communication or synchronization.
So whoo hoo! Now we have a name for these types of inverters.

Update: So, probably not. I have found these in the MW range, but nothing in the 500-2500 W range; which would be about right-sized for incremental adds. They seem built for military bases or islands. I did see "grid forming" on SMA's SunnyBoy; but they achieve that as you can parallel their inverters. On the upside, the CA Rule 21 thread seems to indicate the throttling based on voltage and the Enphase IQs can do so in <= 4s (ref); so whatever surge solution is required only needs to handle the array size for less than 5s and be fast switching.
 
Last edited:
As I read through this, one thought sticks out in my mind.

If the original system is properly designed to meet the needs of the loads being serviced, then the inverters in the original AC battery would be sufficient. This just leaves us with adding more DC batteries behind the inverter to prepare for longer periods of power outage.

Yes, this means a longer charge cycle along with the longer discharge.

I would definitely rather build my own AC battery to run with an Enphase system. I haven't seen the prices in the Encharge, but I'm certain it isn't anywhere near the cost of building it from cells.
You can get 10kW of cells for about $2k. Find a way to communicate with the "brain" that isn't more than a couple hundred, and I'll do it myself.
Then its back to the Incrementally Adding DC Batteries.
 
Less, Ampster just did it for $116 kWh here.
True, the raw cells come out about there.
My cost for 4 280Ah cells was $117.75/kWh. Add the BMS and you're at about $500 total. Rough calculation to use the same cells means that X4 plus bus bars, cables, etc. and you're in the neighborhood of $2k.
 
If the original system is properly designed to meet the needs of the loads being serviced, then the inverters in the original AC battery would be sufficient. This just leaves us with adding more DC batteries behind the inverter to prepare for longer periods of power outage.
I originally had a deposit on a Tesla Powerwall which was a good deal for its capacity, However the only way to add battery capacity was to add another Powerwall and that started to look expensive when all I would need was more batteries. I cancelled the Powerwall order and purchased an Outback Skybox which can AC Couple to my existing Grid Tie inverters. I already had some old Nissan Leaf modules that would give me some time to get a better handle on my battery needs. The power outages that we have typically had are during sunny weather so I will probably never need the full capacity of the 28kW that I just purchased.
 
Find a way to communicate with the "brain" that isn't more than a couple hundred, and I'll do it myself.
I have not found a less expensive way to do that than AC Coupling. I good 5kW hybrid inverter with UL certified AC coupling is going to cost around $3000 and up.
 
Since I'm Enphase I'm fairly sure I could extend my soon-to-be-installed Encharge with a DIY IQ8 + LiFePO4s (My installer has said that a powerwall is compatible with Encharge, but I'm dubious, just can't see them playing nice with frequency control). But I'm pretty sure that now that energy storage is under $150 kWh that there will be a lot of of new AC battery technologies cropping up (e.g., LG, Samsung, GE, Whirlpool), and if they want to compete they're going to have to be compatible with Tesla & Enphase. My thinking is some sort of CA rule 21 compatible grid following inverter will appear on the market first, at that point we'll have something that can sync to phase/frequency and we can just plug in as many as we like.
 
My thinking is some sort of CA rule 21 compatible grid following inverter will appear on the market, at that point we'll have something that can sync to phase/frequency and we can just plug in as many as we like.
My Outback Skybox cost $6000 in late 2018. It has since dropped to $3000 for competitive reasons. I hope you are correct. It is not just CA rule 21 but Hawaii has a similar rule and those markets may continually evolve benefiting users throughout the country. I know very little about the market outside the USA.
I will be interested in your EnCharge experience. As far as I can tell Enphase is restricting them to dealer installs.
 
The Encharge uses IQ8s to make the AC/DC conversion, right?
Are the inverters doing all the communication to the IQ Combiner too? If so, there's nothing stopping us from adding IQ8s to our battery bank and attaching it to the system.
I suspect there is more to it than that.
 
I suspect there is more to it than that.
Yes, I would guess there is some proprietary protocol that the battery will use. Enphase is not going to leave an opportunity for some hacker to use an IQ8 and a commodity LFP battery to cannibalize a product they have been working on for three years.

In some markets it is not just making a GT inverter work off grid but in markets like California and Hawaii with high rates it is the operating modes such as grid zero and self consumption which require programming and a user interface.
 
Last edited:
... It is not just CA rule 21 but Hawaii has a similar rule ...
The applicable statutes I know of are:
  • IEEE 2030.5 for utility and virtual power plant interoperability,
  • IEEE 1547, UL 1741 SA, Rule 21, and HECO Rule 14H for utility interconnections in United States,
  • PREPA for Puerto Rico, and
  • CSA 107.1 for Canada
Probably every country will adopt some form of that.

...Enphase is not going to leave an opportunity for some hacker to use an IQ8 and a commodity LFP battery to cannibalize a product they have been working on for three years.
It's what every other vendor has done. I really don't know enough about the communication; but I suspect someone will eventually reverse engineer it.
Not an expert of course, but it might not need to be anyway, see speculative post #6 above about how Rule 21 can be used by vendors to sneak their AC batteries in with existing systems.
 
Last edited:
...... Not an expert of course, but it might not need to anyway, see speculative post #6 above about how Rule 21 can be used by vendors to sneak their AC batteries in with existing systems.
I took a look at that post and one thing that is missing is communication from the battery about its state of charge. As I understand Tesla's implementation, that as the bettery approaches 95% full the brain (in the Powerall) begins to raise the frequency to power down the GT solar. Older GT inverters all turn off at once, but newer ones can modulate. This avoids the recycling that would happen with older inverters.

I assume what you are describing as a Grid Forming Inverter or an AC battery is what my Outback Skybox with batteries will do? I most often hear those kind of systems referred to as bimodal hybrid inverters.
 
Last edited:
... missing is communication from the battery about its state of charge...
You're right about that...but a competitor AC battery wouldn't have to worry about it by making use of frequency shifting as per the rules listed above.
Here's why. First off, it's a given the "brain" of the base system knows when the house needs power and when it doesn't. From there it'll set the frequency to control the inverters on the solar panel to provide the house the needed power from the power. There can only be one "brain".

So how does the competitor AC battery work in that scenario? If the frequency is shifted the competitor AC battery outputs no power - in fact it shifts into charging mode pulling some number of amps based on the frequency. Why? Because as long as the frequency is shifted the solar panels are being throttled back because they have excess power. There's no reason for the competitor to provide any power in that situation and since there's excess power it might as well charge up.

So, what about when the panels don't have excess power? In this case the brain doesn't shift the frequency, so the panels will output all that they can and normally the battery makes up the difference. This is also where the competitor AC battery wants to supply power, that is when the frequency is not shifted.

When the frequency is not shifted, the competitor inverter can hold the voltage at 240V. Microinverters from the solar panels operate as current sources, so they'll exceed that voltage and therefore their power will be consumed first while available. But either battery/inverter can make up any lag. When the panels are no longer producing, both AC batteries will supply more of the load up to their maximum or fault from to much of a power drain. Ideally the competitor AC battery matches the voltage of base system's battery to load share - but it might have user-configurable algorithms to stretch the power out (after all, the longer it lasts the more inverter power there is for lags from power surges).

The last piece of the puzzle is that the competitor AC battery needs to know when the system is off-grid so it knows whether to be in charging mode or active inverter mode. A CT sensor would work, but I'm hoping they come up with something simpler so installation is as easy as plugging it into the wall (fingerprint of the inverter's power?). Ideally these would be small modular units in which over time you could add more DC watt-hours or watts inverter power as you needed them.
 
Last edited:
You're right about that...but a competitor AC battery wouldn't have to worry about it by making use of frequency shifting as per the rules listed above.
My hybrid inverter controls both GT inverters. Most of the time I am on the grid but in grid zero mode with sell back of excess power to the grid. Most of the time I am self consuming. It is a much simpler system. I guess it is a little of topic from an AC battery thread.
 
Last edited:
Back
Top