diy solar

diy solar

Incrementally Adding AC Batteries

Yesterday I postulated a competitor could build an AC battery that could be mixed into a Solar /Encharge home without having to do a lot of reverse engineering using UL-1741 (e.g., compatible with CA Rule 21).

What about DIY in that situation? Could we build our own using a UL 1741 compliant inverter?

Short answer: Not as far as I know currently unless you go the full AC coupling route.

Long Answer
The problem is a simple UL 1741 compliant inverter is a grid tied inverter - a current source. So once turned on it would immediate drain the battery as fast as it could invert pumping all the battery power into the grid.

You could get around this by controlling the inverter, say with an Arduino and some logic:
  • When the grid is up or the frequency is shifted (e.g., excess solar power), the inverter stays in charge only mode
  • Anti-bounce logic for when the grid is down but the solar is providing some fraction of the power load (that is the inverter coming on could provide too much power for the house's load, which would cause the brain to frequency shift).
By itself, that's still not good enough.

First is the insidious problem of surges since frequency shifting takes time. Just having a brain to frequency shift isn't enough, you need the hardware to handle it. For example, from the chart below you can see an Encharge 10 is only rated to handle up to 18 IQ7Xs, if you have more than that on the roof, they shut down when the grid is off because the Encharge 10 doesn't have the capability to handle more surge energy than they could provide.

encharge_pairingchart_image2.png

Obviously as the number of AC coupled solutions show, the problems are not insurmountable. But it's probably beyond the skills of an average DIYer to build/size something and not worry about it catching fire.

The IQ8 inverters are designed for this, but as Ampster pointed out, may have a proprietary interface making it difficult for a DIYer. So, might not be able to just slap a battery on them as I'd like to do.

In the future I expect to see inexpensive small wattage (500 to 1,000W) third party inverters that are designed for this. Vendors that already have AC Coupled solutions have the tech and worked out the bugs... it's just that their existing line up is to large to mix and match to make scaling up easy. So when Outback or Fortress Power come out with a 500W unit where I can add my own LiFePO4, I'd definitely buy one, maybe two!
 
Last edited:
That is a thorough analysis of the issue.
I wonder what miniimum size issue will be economic. I assume the cost will be influence by the cost of UL certification. One of the things I have discovered with AC coupling is all of the power of the GT inverter must flow through the hybrid inverter. So a 500 Watt hybrid would only handle a couple of solar panels. A typical household circuit is 20 Amps or 2400 Watts. The relays & power handling devices start to escalate in price. Since most GT inverters are 240 volt devices using 120 volt inverters may be more complicated than it is worth.
 
... So a 500 Watt hybrid would only handle a couple of solar panels...
Suppose we let the original system worry about frequency shifting and controlling panels and the add-on AC battery didn't worry about the panels?

Could it be a lot lot simpler to create small inverters for this? I think so... here's why...

First, let's look at a simplistic example of a pure sine wave inverter, in this example we'll say incoming is 12Vnom DC. It's switched back and forth across a torrid and magnified to 240 VRMS:

concept-of-power-inverter.gif

When we say our AC is 240V, we really mean that it's 240VRMS. That's convenient as the RMS value gives the same amount of energy as DC for calculation purposes. But what you'd really see on your oscilloscope is:

articles-rms1.gif


Vpeak > 240V. If it were just a simple magnification across the transformer for 12V to 339V (Vpeak for 240VRMS), it would be 28.28x

Similarly 12Vnom doesn't mean 12V. As the battery discharges, it's voltage changes slightly:

Lithium-12V-Discharge-Rates.jpg
When fully charged it's ~13V and with a low voltage cutoff around ~12V.

So, with a simple 28x magnification the actual AC peak voltage would be 13x28=367V to 12x28=339V. Similarly, if they don't adjust the power under the curve, then VRMS is 247V to 228V.

For this new proposed inverter to work it would need to sense the AC to match the frequency and voltage (either VRMS or Vpeak would work, but peak gives wider separation). Given that, from the voltage the new AC battery can roughly know the SoC of the other battery.

In fact, if the chemistry of the batteries were the same the two batteries would naturally stay at the same SoC because if either had a higher SoC the AC voltage would be higher and therefore it would supply more current and drain faster (until it matched the other battery).

Given the grid voltage is more variable, this hypothetical new battery would still need some way to know if the grid is up. If the grid's up or frequency is shifted it can charge. Possibly it could just be programmed with a starting voltage like some UPS models.

There's one other critical need for this hypothetical inverter in the AC battery... consider a case where the grid is out and both batteries and solar are powering the air conditioner which suddenly shuts off. The IQ8s can handle the surge from the solar inverters and itself...but they won't be able to handle the new battery too. Somehow the hypothetical AC battery must handle it's own surge.
 
Last edited:
Suppose we let the original system worry about frequency shifting and controlling panels and the add-on AC battery didn't worry about the panels?
I am going to need some time to wrap my head around that assumption. I have spent years trying to understand the complexities of AC coupling. In the meantime can you clarify in the above statement which is the GT and which is the hybrid inverter? You can use "battery inverter" if that is easier than "hybrid inverter". I just find the term AC battery confusing. Can we agree that whatever we call this is, it is a system that consists of an inverter, a DC battery, some CT clamps, some programmable logic most likely a transfer switch and other bits and pieces.
 
... In the meantime can you clarify in the above statement which is the GT and which is the hybrid inverter? You can use "battery inverter" if that is easier than "hybrid inverter". I just find the term AC battery confusing.
I used AC battery specifically because it is neither a hybrid or GT inverter...but has some properties of both (e.g., phase matching of a GT...but NOT a current source).

I came up with the idea after reading your post and a couple of drinks last night...it might make more sense if you've had a couple of drinks too.

My viewpoint was thinking of it as an Enphase system.... so IQ7s on the roof (they're UL 1741 GT inverters obviously, synonymous with any existing GT inverter), the the IQ8s in the Encharge (they have "modes" set by the Envoy, so they act in different ways: Grid forming, Grid following, charging, possibly others), and should be synonymous with a Powerwall... but possibly not a AC coupled system. This "small" hypothetical AC battery is something you'd connect in to augment the IQ8 watts and watt-hours; preferably without burning the house down.
 
Last edited:
... Can we agree that whatever we call this is, it is a system that consists of an inverter, a DC battery, some CT clamps, some programmable logic most likely a transfer switch and other bits and pieces....

We can definitely agree on the DC battery, and other bits & pieces... ;)

Helps to always put an adjective in front of "inverter" for this discussion. AFAIK there's:
  1. grid-forming inverters (aka Off grid inverters that are voltage sources: set voltage and frequency),
  2. Grid Tied inverters (frequency matching current sources: ramp up voltage to push available watts), and
  3. grid-following inverters that try to stabilize voltage and frequency.
Like the IQ8s, the inverter in this AC battery isn't any of them, for example:
  • Shouldn't be a current source but should match frequency
  • Should stabilize voltage, but not frequency
  • Act like a voltage source, but not form a grid.
Not sure about the logic and would prefer to not have the CT if possible. Ideally I put it in my laundry room and plug the drier into it, and it into the wall. Guess that means we do need some logic, that is if there's no active grid it goes dormant.

No transfer switch like an off-grid inverter that has one built-in. But... they might need one internally to switch between being a charger and an inverter.

Wonder if I could come up with a picture of this thing... if you're trying to wrap your head around it I've probably lost a lot of others. I definitely don't want to lose the high-spirited entrepreneur who's actually going to build/sell these things.
 
Last edited:
I'm tracking.
It's a pretty cool concept to jack up the AC voltage generating a voltage that is relative to the SOC.
The theory seems sound. Detecting the frequency and synchronizing shouldn't be too hard.

I just want some of what you're drinking! ?
 
So here's a stab at it...

You'll see I broke down and used a CT on the grid feed to a transmitter. When the grid's down it transmits a low-range tone over RF. You can have multiple Magic batteries plugged in any 240V outlet (yeah, these could be 120V in any socket, but we're keeping it simple for now). They'd all be using the same transmitter. Would still like to get rid of that if someone thinks of something clever.

1593207511538.png
Logic
  • No Voltage/frequency on L1/2, no output power, also satisfies anti-islanding requirements
  • No CT Tone received (Grid Active)​
    • Charge Mode​
  • CT Tone (Grid offline)​
    • Frequency Shifted - Charge Mode: max possible charge rate set by shifted amount (but of course might be less based on SoC)​
    • Frequency not shifted - Inverter timing source synchronizes (probably use the existing AC signal as transistor's base with cap offset for perfect phase alignment)​
    • Voltage > VMax (~247VRMS) - OverVoltage protection kicks in: charges battery if they can absorb the energy fast enough or if battery full dumped to waste heat. This needs to be fairly fast, it can occur when things like inductive motors stops suddenly (e.g., air conditioning turns off, see inductive kickback spikes). It only needs to sink as much as the inverter's maximum output and probably less than a second. Heat sink is sort of a worse case scenario, big caps and clamping voltage? Also, Jack Rickard said if the voltage on a standard inverter exceeded the output voltage, the current would back-flow into battery (e.g., collector/emitter reversed)...could probably make use of that with the right transistors (might be how existing charger/inverters work). This is a swag, don't know enough about this other than it's possible as existing inverters do have this built in (e.g., Outback, Enphase, Tesla).​
    • if Surge TSink > Thot and battery > 80% disconnect until Tsink < TLow
    • If Voltage <= VMax, Inverter mode, V set to match existing V or VMin, whichever is higher. If they have the same chemistry V should naturally stay in sink equalizing SoC for all the batteries in the system. But with VMin, the AC battery is guaranteed to exhaust before the Energy Storage controller - which is important since its the one providing the frequency reference.​
    • DC Voltage <= VLowDC disconnect​
The AC Batteries "power cord" is not a "suicide plug" because it won't output any current if it doesn't detect voltage, similar to any GT inverter.
 
Last edited:
Okay, I will have some wine tonight and respond in less detail. I tried to find out more about IQ8. I did find out Enphase does their R & D in India. Good place to test how equipment behaves when the grid goes down.
 
So here's a stab at it...

You'll see I broke down and used a CT on the grid feed to a transmitter. When the grid's down it transmits a low-range tone over RF. You can have multiple Magic batteries plugged in any 240V outlet (yeah, these could be 120V in any socket, but we're keeping it simple for now). They'd all be using the same transmitter. Would still like to get rid of that if someone thinks of something clever.

Logic
  • No Voltage/frequency on L1/2, no output power, also satisfies anti-islanding requirements
  • No CT Tone received (Grid Active)​
    • Charge Mode​
  • CT Tone (Grid offline)​
    • Frequency Shifted - Charge Mode: max possible charge rate set by shifted amount (but of course might be less based on SoC)​
    • Frequency not shifted - Inverter timing source synchronizes (probably use the existing AC signal as transistor's base with cap offset for perfect phase alignment)​
    • Voltage > VMax (~247VRMS) - OverVoltage protection kicks in: charges battery if they can absorb the energy fast enough or if battery full dumped to waste heat. This needs to be fairly fast, it can occur when things like inductive motors stops suddenly (e.g., air conditioning turns off, see inductive kickback spikes). It only needs to sink as much as the inverter's maximum output and probably less than a second. Heat sink is sort of a worse case scenario, big caps and clamping voltage? Also, Jack Rickard said if the voltage on a standard inverter exceeded the output voltage, the current would back-flow into battery (e.g., collector/emitter reversed)...could probably make use of that with the right transistors (might be how existing charger/inverters work). This is a swag, don't know enough about this other than it's possible as existing inverters do have this built in (e.g., Outback, Enphase, Tesla).​
    • if Surge TSink > Thot and battery > 80% disconnect until Tsink < TLow
    • If Voltage <= VMax, Inverter mode, V set to match existing V or VMin, whichever is higher. If they have the same chemistry V should naturally stay in sink equalizing SoC for all the batteries in the system. But with VMin, the AC battery is guaranteed to exhaust before the Energy Storage controller - which is important since its the one providing the frequency reference.​
    • DC Voltage <= VLowDC disconnect​
The AC Batteries "power cord" is not a "suicide plug" because it won't output any current if it doesn't detect voltage, similar to any GT inverter.
There will likely be a little delay in the circuitry for synchronizing the phase. Matching it exactly will probably take something a little clever.

I think a small bank of super caps would do a good job of handling surges to feed it back to the system when things return to normal.
 
After a glass of wine I couldn't figure out which iteration of the design I was supposed to respond to. There was a 120v plug in version which I have seen in the market place for several years. The 240 volt version with or without using CTs.

Nevermind the latest just popped up on this thread. I will have a look at it after I have a cup or two of coffee.
 
Also, Jack Rickard said if the voltage on a standard inverter exceeded the output voltage, the current would back-flow into battery (e.g., collector/emitter reversed)...could probably make use of that with the right transistors (might be how existing charger/inverters work).
Yes, i saw that video, and that is the crux of an AC coupled hybrid inverter and how they work AFAIK. The brain even calculates the SOC of the battery and if there is not enough reserve in the battery it cranks down the GT generation to so that hybrid inverter batterys get to a SOC that gives it a reserve. That way it can buffer any swings in load or solar production that it cannot control. Increase in load and reduction in solar can be handled by battery drain. The opposite is treated with the excess energy charging the battery.That is why those systems always need the battery at less than 90% SOC. Most systems like this have a minimum battery size to GT inverter relationship. I only mention AC coupling because whether one uses frequency shift as used by AC coupling devices or some other method of modulating the GT inverter, that is the process that is fundamental to managing the inputs and outputs of such a system.
Speaking of Jack Rickard, i saw his last video and he did not sound very healthy.
 
Last edited:
You'll see I broke down and used a CT on the grid feed to a transmitter.
I think some grid sensing is built into most ATSs. The system will need more CTs than that. One would be needed on the load and one on the GT inverter AC output. Until recently the Tesla Powerwall used a Nuerio which transmitted over 485 protocol to the brain module contained in the transfer switch or the Powerwall.

Nuerio is now owned by Generac which has made some strategic moves including the purchase of a battery system manufacturer. That battery is expensive right now so they dont have a competing product to the Powerwall. However they do have a user base of devoted customers some of who may want to leverage GT solar which the cannot do with a Generac right now. If Generac were to develope an inverter based generator then they would have all the things necessary to AC couple that to a GT inverter. That would give them a system that would start within milliseconds of the grid going down, not have to use fuel until the sun went down, thereby doubling the time that a fixed volume of fuel would last.
 
There will likely be a little delay in the circuitry for synchronizing the phase. Matching it exactly will probably take something a little clever.
I don't know why that system just doesn't provide an AC source to the GT inverter and be done with it. GT inverters are designed to follow the grid. The most effective way, so far, is simple AC coupling. That is what the Tesla Powerwall uses. I have been following Elon Musk and am acquainted with several of his Engineers from SpaceX and Tesla. If there were a cost effective way to control a GT inverter without using AC coupling I think he would have done it. He does not believe in reinventing the wheel, unless he can make it better.
 
There will likely be a little delay in the circuitry for synchronizing the phase. Matching it exactly will probably take something a little clever.
GT inverters do this today and the 120V ones on Amazon don't cost much so at least we know it can be done. I suspect what they do is to use the existing signal on the transistor's base, so easy to match frequency. Obviously there's a bit of a time lag in the response, but to match phase a small capacitor in front of the base to account for the lag (which should be a fixed delay for the circuitry) should be able to offset whatever the time delay is. I bet you could have phase matching on the first cycle.

...I think a small bank of super caps would do a good job of handling surges to feed it back to the system when things return to normal.
Truthfully, I've no idea how you handle inductor kickback spikes, surges, and from over-production. GT's cap out at a max voltage, so I'm not sure how big of a deal over production is (wind turbines use load-diversion technologies, but that's a different can of worms). Straight-up inductors store energy as a magnetic field which gets dumped when the current stops (Elctroboom has an amusing video on this) and they can have big spikes. Similarly, spinning motors don't stop immediately because of their momentum. I know you could have a surge protection via clamping (video) ... but ultimately I don't know what would work best in this situation. Real EEs (e.g., @PHoganDive or @ghostwriter66) could probably tell you. I know it's done in the IQ8 and other inverters so assume it could be done here.

From @Ampster 's post above it sounds like AC Coupling solutions disconnect the renewable energy until it's battery has discharge enough that can absorb that energy fast enough. That technique wouldn't work with the magic AC battery as it doesn't have any control over the renewable energy source. It's not as bad as it sounds, by definition the renewable energy source has to be able to handle both the surges from the source and from it's own battery. The magic battery only needs to be able to absorb spikes equal to the maximum wattage it can put out, and these will probably be pretty small.

...I think some grid sensing is built into most ATSs.
All the big ones have a relay that flips so it can be used to activate auxiliary systems (e.g., a generator), even if it didn't probably easy to add a relay.
So, might not need the CT...but that's still a transmitter. Ideally it would be nice if you could just plug it in and you're done...having a transmitter/receiver just adds potential problem points. Say for example it could recognize the grid's fingerprint from the energy storage's fingerprint and just magically know or learn which was which. UPSes don't have external sensors, they just work off low voltage, so that technique might be a possibility.

...The system will need more CTs than that. One would be needed on the load and one on the GT inverter AC output. Until recently the Tesla Powerwall used a Nuerio which transmitted over 485 protocol to the brain module contained in the transfer switch or the Powerwall.
I agree the main brain needs them.

But I don't believe the magic AC battery needs more. The main energy controller in the diagram is handling the renewable energy production, the magic AC battery just matches voltage when the conditions are met -- keep in mind this isn't a current source like a GT inverter, it's a voltage source... so no current would flow unless the voltage lags.[/QUOTE]
 
Last edited:
That technique wouldn't work with the magic AC battery as it doesn't have any control over the renewable energy source. It's not as bad as it sounds, by definition the renewable energy source has to be able to handle both the surges from the source and from it's own battery.
What renewable energy source are you talking about in your diagram, that has a battery? What would a surge look like and where would it come from?
 
What renewable energy source are you talking about in your diagram, that has a battery? What would a surge look like and where would it come from?
It's any renewable energy source that has an energy storage system that can control said renewable via meets UL 1741. For example the Enphase Ensemble solution qualifies. What I don't know is if it would work in an AC Coupled solution as those systems seem to route all the power through their brain/inverter system. Need to cogitate on that.
 
Attached is a simplied schematic of a Grid Tie Battery based Inverter/Charger. Pretty much same whether Outback, Xantrex, etc. manufacturers. Rest of it is digital/firmware features by a given manufacturer to control the hardware.

If a bi-directional inverter is strange to you, it is accomplished by just slight adjustments to the MOSFET's PWM sinewave sequence duty cycling which can change power flow direction and rate of flow, on the fly, from DC to AC or AC to DC ports. The inverter can suppliment any AC port or draw from AC to charge batteries.
 

Attachments

  • Simplified GT battery inverter_charger schmatic diagram.pdf
    197.7 KB · Views: 17
This is an SMA AC coupled with a Radian using a DPST style ATS to have whole house backup rather than a critical circuit panel. See post above for AC battery logic. Hopefully obvious you could strip out the ATS and use a critical circuits panel with the same electrical impacts.

As you'll see when reading through there is a problem as Ampster previously brought up.

1593290617112.png


  • When the grid is up the dpdt ATS configuration lets the Radian back-feed the grid. (illustration above indicates grid active)
  • When the grid goes down both switches flip, the grid is disconnected.
  • The Radian's AC in/GT-out line is now dead, the AC out from the Radian's Grid forming inverter goes live. It's a voltage source, so only supplies current as needed, sets VRMS-nom = 240V and frequency to 60 Hz (U.S.).
  • The radian takes power from the solar panels and batteries; using frequency shifting and relays to control their power output.
  • The batteries are sized appropriately for the arrays so the Radian can handle surges from the renewable energy source only.
  • The load center supplies power to devices
  • The owner wanted to augment either the watts available, the watt hours available, or both so has installed 1 or more magic AC batteries.
  • In this case, the Radian only frequency shifts the renewable energy source (in the previously example it shifted all AC power)
  • If no tone from CT Transmitter (grid active), and both voltage and frequency are within range then AC Battery is in charge mode.
Grid Down Cases:
  • AC battery's CT receiver picks up the CT Transmitters tone and knows the grid is down.
  • The AC from the Radian's Grid Forming battery is always 60 Hz. This is a problem as the AC battery cannot know how much, if any, excess renewable power is available, and therefore not know if it is acceptable to charge or not.
    • As Ampster indicated earlier, it could be accomplished with a CT on the Radian's DC line (that is, no battery draw, then there's excess power). The AC battery could ramp up the charge rate util the Radian's DC CT hit some preset value (e.g., you'd still want the Radian's battery to charge too).
    • Given this discussion, it might be possible to know if there's some excess power as Vpeak would be higher while the Radian's battery is being charged.
    • Still not ideal as without knowing how much frequency shifting the AC battery can't know the maximum charge rate. Plus I hate CTs... need to find a better solution.
  • The AC battery uses the Radian's sine-wave to synchronize phase.
  • If the AC Battery detects excess power (e.g., CT, Vpeak, other magic) it enters charge mode
  • Otherwise both inverters are now acting a voltage sources, so each supplies current to prevent voltage sags from appliances.
  • If they have the same chemistry, each should maintain a similar state of charge (see this discussion)
  • The AC battery has a minimum voltage setting, so it should exhaust first near the end, important since the radian provides the frequency.
Ideally an AC battery, especially a magic one, should work in either scenario.... so this idea still needs work.
 
Last edited:
What I don't know is if it would work in an AC Coupled solution as those systems seem to route all the power through their brain/inverter system. Need to cogitate on that.
I believe the Encharge can AC couple to almost any GT system. That is an assumption that i will try to verify. Certainly, as we have discussed, it may use a propriety communication to communicate with other Enphase products which would allow it to control the other micros much faster. That would mean the GT kW to battery kWh ratio could be less.

I have four IQ7s that are AC coupled to my Skybox but have not witnessed the modulation personally. I recently installed an Envoy and added CT clamps so with the IQ 7s Enphase has a modular system that can be configured for zero export as may be required in some jurisdictions. I continue to be amazed at their adaptability and can't wait to hear first reports
 
Back
Top