diy solar

diy solar

SMA SunnyIsland + Enphase microinverters + A/C compressor load = enough electrical noise to interfere with microinverters?

Get a look at your waveforms. GXMnow believes rectifier/capacitor front end of some power supplies is what caused his problems.


 
Update: SMA service has responded to my service ticket and they seem to be thinking about it. No good advice yet, but better than more than a month of silence. I made some better video clips of the issue occuring, if anyone's interested. https://photos.app.goo.gl/dYe4DZsSzcGr4B1VA The info tab on each picture or video has some description.

This patent, https://patents.google.com/patent/US7899632 might help explain the issue. I wonder if the SMA inverters are not always robust to this anti-islanding behavior from the micros. Just under certain conditions of loads, higher system load from the A/C, car charger, etc, but with the SI inverters not really doing much at all, since all the microinverters are supplying the loads.
It looks like the SI instability is caused by A/C & charger loads so that should be corrected first with SMA support hopefully. Did the same loads used to run without causing instability?
I set 27104 and 27105 to values at or near zero instead of the default values. Have to _not_ try to connect to the grid (for charging or grid-feed), with these settings. It will initially synchronize with the grid, but within a couple seconds it opens the relay and gives a fault for AC2 shorted.
Have you tried disabling droop control (27129 Inv Drp Ctrl Normal)? Droop control is typically used for power sharing/synchronization between parallel inverters and/or the grid. For testing the off-grid scenario it would be good to see the test result without droop control enabled.
 
It looks like the SI instability is caused by A/C & charger loads so that should be corrected first with SMA support hopefully. Did the same loads used to run without causing instability?
Any load or combination of loads is fine, except with A/C and the microinverters approximately match the total load, with FLA battery current close to zero, then it goes unstable, then SI's start taking a whole bunch out of the FLA battery. Last summer I had 9 panels DC coupled to the FLA SI battery. I didn't notice an issue last summer running the A/C, and it could have been because the SI were inverting significantly from the FLA battery bus (power dumped on there by the DC coupled solar), while the FLA current was near zero. Now I put those 9 panels on a separate lithium bank, with micros that discharge the lithium bank. So now there is only AC coupled solar and AC coupled battery microinverters.
Have you tried disabling droop control (27129 Inv Drp Ctrl Normal)?
Hah! No I will try that next, that could be it, turn off the control completely. I initially thought that might mean the AC coupled load shedding control (which I need for sure). So far I haven't found 27xxx settings documentation, only guessing based on the names, units, etc.

And, I realized I have something to measure current on the oscilloscope, it seems to work. I added more videos here. I haven't spent any time pondering it, I'm not sure what it shows, please feel to chime in. I measured L1 current & voltage, at the feed to the basement panel which has the A/C on it, from the Sunny Island AC1 https://photos.app.goo.gl/dYe4DZsSzcGr4B1VA
 
Nope, 27129 InvDrpCtrl turned off ('FrqCtrlOff', instead of 'Normal') does not help. It seems about the same as having 27104 & 27105 already close to zero (27104=0, 27105=-.030)

Interestingly, adding the load of one or two ferro-resonant battery chargers seems to stabilize the system, at least under these conditions. Adding a 120v ~1.2kw EV charger load destabilized it. Today at least....
1659564039177.png
 
Last edited:
So that gives me an idea.... I'm an engineer, but only an electrical engineer by imagination, so this may be all wishful thinking.

I have a boost transformer laying around, and extra capacitors. What if I hooked it up to the A/C unit, after the contactor, analogous to the charger circuit without the charger's power take off rectifier? Would that be helping or hurting the power factor? Would it potentially be counteracting the microinverters anti-islanding detection feature behavior, by storing power from one cycle and then sending it back on the next cycle?

I could do the same experiment by pulling the battery disconnect on the tractor, plugging in the charger, but not connecting the battery.
 
So that gives me an idea.... I'm an engineer, but only an electrical engineer by imagination, so this may be all wishful thinking.

I have a boost transformer laying around, and extra capacitors. What if I hooked it up to the A/C unit, after the contactor, analogous to the charger circuit without the charger's power take off rectifier? Would that be helping or hurting the power factor? Would it potentially be counteracting the microinverters anti-islanding detection feature behavior, by storing power from one cycle and then sending it back on the next cycle?

I could do the same experiment by pulling the battery disconnect on the tractor, plugging in the charger, but not connecting the battery.

That will likely have limited effect. The scope traces show significant and dynamic harmonic distortions that likely won't be fixed by a simple capacitor bank. It's pretty clear that your A/C is a main cause of the harmonic distortions. It looks like the third harmonic distortion is the main culprit but it would be good if you could measure the harmonic distortions to verify. Generally, you need to maintain THD < 5% and any individual harmonic distortion < 3% to achieve the grid-like power quality that GTI's expect. The conventional fix would be to put in a power harmonic filter or power conditioner.
 
Just an update. Still no particular useful feedback from service ticket case with SMA. They keep talking about FSPC settings, but that is not what's happening, it's not a situation where it is trying to curtail.

They so far claim no documentation or knowledge about the droop control settings (direct access 600 menu, 27104, 27105, 27129) I am very sure this control is what is becoming unstable for whatever reason. Reasons could be THD from system loads, anti-islanding perturbations from microinverters, and interaction with the SI droop control.

I agree there could be 3rd order disturbance, some of the scope traces looked that way, especially when plotting L1 - L2 voltage. Various loads in the house result in different looking signatures when looking at L1-L2. Some look totally worse to me, especially 120v microwave, but maybe that's just with the load on one leg. In any case, the SIs should be totally capable of handling it, the load is nowhere near the continuous capability, and the load is near zero with AC coupled solar.

Does anyone have recommendations on a low budget device to measure THD or harmonics? If I had a way to measure that, If it was something that the raspberry pi could know about, I could add a feature maybe, that would control the electrak chargers which seemed to have a positive effect. That's if I had a way to detect that the SI freq was oscillating in the first place.
 
Oscilloscope or other sampling device, sufficiently above highest harmonic of interest.
Even if it doesn't do FFT, Excel or other math program can.

How high a sample rate can the PI do?
 
Does anyone have recommendations on a low budget device to measure THD or harmonics? If I had a way to measure that, If it was something that the raspberry pi could know about, I could add a feature maybe, that would control the electrak chargers which seemed to have a positive effect. That's if I had a way to detect that the SI freq was oscillating in the first place.

Depending on what "low budget" means to you, you may find the Eastron meters that support THD measurements acceptable. They are much cheaper than other THD meters specifically designed for AC power. For even lower cost DIY, you can hook up a laptop with a sound card to capture transformer isolated voltage and current signals and use free audio analyzer software to do FFT/THD analysis.
 
Another update: SMA just wants to close the ticket. Not offering any helpful advice beyond "don't AC couple those microinverter on the AC1 off-grid side of the Sunny Islands" They will make no comments about the droop control parameters that have a significant impact on it.

Oh, and they point me to this Enphase discussion which ends in the customer tossing their Sunny Islands for a Sol-Ark due to the AC coupling issues.... https://support.enphase.com/s/quest...ual-sma-sunny-island-inverter?t=1668108411622

I guess I'll see next summer, if we still have the old PSC motor air conditioner. Maybe an inverter driven compressor unit will be better.
 
Last edited:
Maybe an inverter driven compressor unit will be better.

Only if it has good power factor (at least if you have Sunny Boys)

Not offering any helpful advice beyond "don't AC couple those microinverter on the AC1 off-grid side of the Sunny Islands" They will make no comments about the droop control parameters that have a significant impact on it.

If the microinverters are well behaved, ought to AC couple OK.
Wonder if thing would work better with Sunny Boy as wells as microinverters?
 
Wonder if thing would work better with Sunny Boy as wells as microinverters?
Yes of course they allege that as a solution also. My microinverter arrays are not really conducive to being set up as high voltage strings for various reasons. They think I own twitter and have funding to try every other hardware combination that _might_ fix the issue, while they offer zero advice on what is going on and ignore the undocumented parameters in the SI inverters that have a significant effect on it.

I think their droop control is just making some basic assumptions and can't be bothered to be stable in certain conditions. Changing those parameters or changing relatively small loads on the systems turns the issue on & off. I'm sure it could be stable, the SI inverters have way higher capability than the collection of dinky 200w micros, even with 35 of them, still only 5-7kw. But SMA only works with SMA, once there's an issue, and they probably only really test with their own stuff. Forbid any testing using the most popular microinverter brand out there, or curiosity to help out with troubleshooting.

In this condition, the SI inverters are not pushing or pulling much from main battery, and the micros are mostly running the loads directly. The SI inverters are probably not programmed or tested to handle much disturbance if they're sitting doing not much at light load. Any GT inverter or microinverter has to do some sort of disturbance for anti-islanding, it seems like they'd do some testing and development for it. Or at least they'd be interested when issues come to look into it or offer advice or solutions. And someone else out there must expect to run a 4kw AC unit. I thought the SIs were honey badger inverters. I still like them, but they're not as fierce as I expected.

Maybe there are parameters or settings in the SI that have to do with how much power factor correction or THD correction they would do? And same for microinverters grid profile (maybe in the newer IQ ones, but I don't see anything like that in the old M series grid profiles).

I'll see what happens with my next load I'm working on: I want to use off-peak off-grid excess for hydronic boiler water heating, but need a variable dimmer sort of control on the heater element to match the available solar. The WH element is 3800w, 16a, and the SCR chopper is going to make very poor sine wave loads. See how that works during AC coupling when it's the micros supplying all the load and the SI are just the lazy hosts in charge of 60Hz and a smooth sine wave.
 
I think the SI will keep operating into poor PF loads, "honey badger" as you say. Don't know if they have adjustable correction levels.

I may be seeing issues where they disconnect from grid, having them in parallel and PV or loads in excess of what one alone wants? Just a guess, I've seen power go off and found breakers tripped. Current shouldn't have been enough to trip two breakers paralleled, but if one SI opened the other might have been overloaded.

Why can't you put some PV panels in series? You could pick up a used Sunny Boy, maybe 5kW or 3.5kW (check tables for which support "backup"), and system should be good with some or all PV on them.
 
I think their droop control is just making some basic assumptions and can't be bothered to be stable in certain conditions. Changing those parameters or changing relatively small loads on the systems turns the issue on & off. I'm sure it could be stable, the SI inverters have way higher capability than the collection of dinky 200w micros, even with 35 of them, still only 5-7kw. But SMA only works with SMA, once there's an issue, and they probably only really test with their own stuff. Forbid any testing using the most popular microinverter brand out there, or curiosity to help out with troubleshooting.

In this condition, the SI inverters are not pushing or pulling much from main battery, and the micros are mostly running the loads directly. The SI inverters are probably not programmed or tested to handle much disturbance if they're sitting doing not much at light load. Any GT inverter or microinverter has to do some sort of disturbance for anti-islanding, it seems like they'd do some testing and development for it. Or at least they'd be interested when issues come to look into it or offer advice or solutions. And someone else out there must expect to run a 4kw AC unit. I thought the SIs were honey badger inverters. I still like them, but they're not as fierce as I expected.
Have you ever looked at the current output waveform of the Enphase branches? If not, I suggest you check. It seems that you encounter the problem more often in the afternoon? From what I have seen of Enphase micros, the current output can become very jittery during the afternoon hours as the sun fades that I believe is due to MPPT hunting. So, I suspect the MPPT hunting adds additional disturbances, on top of the anti-islanding perturbations, that further destabilizes the island grid during the afternoon hours.

One challenge with having 35 micros vs a single string inverter of the same capacity is that the micros do not synchronize their anti-islanding perturbation or MPPT hunting. The result is almost constant disturbances in an island grid. Most off-grid inverters like the SI don't have the ability to actively reject such disturbances especially at the sub-cycle time scale. Instead they rely on the system inertia to handle such disturbances. Also, I suspect the topology of your micros with multiple branch configs and inverter models further complicates the problem. Not sure you already tried this but you may want to incrementally turn off micros while the problem is happening with the A/C on to see if that helps. Anecdotally, it seems more problems occur when the AC couple ratio between the off-grid inverter and micros rated output capacity is < 2:1.

I'll see what happens with my next load I'm working on: I want to use off-peak off-grid excess for hydronic boiler water heating, but need a variable dimmer sort of control on the heater element to match the available solar. The WH element is 3800w, 16a, and the SCR chopper is going to make very poor sine wave loads. See how that works during AC coupling when it's the micros supplying all the load and the SI are just the lazy hosts in charge of 60Hz and a smooth sine wave.
SCR chopper will likely cause havoc.
 
I think the SI will keep operating into poor PF loads, "honey badger" as you say. Don't know if they have adjustable correction levels.

Why can't you put some PV panels in series? You could pick up a used Sunny Boy, maybe 5kW or 3.5kW (check tables for which support "backup"), and system should be good with some or all PV on them.
Yes the SI keep running, but don't keep the frequency fixed at 60 Hz, they let it oscillate. So I guess they're like the honey badger after being bitten by the cobra, they get a little drowsy.
Sorry I can't stop laughing.

My system in these conditions may be similar to the grid example with solar, as SMA explains here: https://my.sma-service.com/s/article/Low-Power-Factor-with-Solar-in-Operation?language=en_US
My pair of SI inverters is the 'grid', my load is the A/C unit that's relatively big and not a great PF. It all works fine if the SI is the only source (discharging the battery). The PF seen by the SI is something it is programmed to handle, and the SI is also providing significant active power. Now we add the solar (microinverters) and they provide 1.0 PF to the system and they offset all the required active power. There is only out of phase reactive power left and the SI sees maybe an unusual high PF? Or at least the PF at other points in the system could be off too. Somehow this causes the SI frequency control to become unstable. The microinverters turn off because they see grid frequency instability, even with their grid frequency range opened to 57 - 62.5Hz.

I don't care how unusual or aggressive or whatever the Enphase anti-islanding might be, the SIs are a honey badger, they should be keeping the frequency the same, even if the PF they see or the resulting voltage is wacky, they should still be able to decide they want 60.0hz. I'd love to discuss or be educated by the SMA experts who sprechen sie deutsch but they don't care, and they don't care to make their inverters have robust control. Somehow, apparently, the Sol-Ark or other inverters handle this allegedly.

I'd rather keep the main GT array with micros (avoid hvdc on the house, keep the array as is for a utility contract with 9yr remaining not supposed to expand it, not all exactly the same orientation and minor variable shading ). An array on garage roof has 3 micros, and 3x3 series panel strings ~120v DC Midnite Classic charging a second battery bank, and have a plan to put AFCI and rapid shutdown on it, one day.... The third array is the DC solar trailer panel rack, sitting on the ground with micros, and that could be converted to a string inverter, that could make sense. I'm not opposed to SMA SB string inverters, I should start watching for a good deal.

On the garage array and the lithium battery it charges, once that battery is full, relay controlled microinverters discharge that battery (effectively directly inverting the solar DC power immediately), so that adds more AC coupled behind the SIs. Another option would be to use DCDC converters between that lithium bank and the main FLA bank, so the SIs are doing the inverting, and then they are putting out both active and reactive power, and maybe not as bad PF to deal with.

I have a pretty low cost bar to get things under, since I've done pretty well with collecting parts (DC solar trailer, basically free old micros). I really shouldn't be complaining, I got a honey badger inverter setup for pretty low cost, this is only an issue on half a dozen days in the summer, with my current loads.
 
Last I heard SolArk preferred AC coupled PV behind "generator" relay, and DC coupled > AC coupled.
So it can let microinverters run full blast, vary DC coupled, or disconnect microinverters and vary DC coupled.

SI is trying to vary frequency so AC coupled matches load and battery charge matches target. So it probably works fine so long as battery takes 100% of excess AC power.

DC Solar trailer means 2400W of PV coupled through Midnight. It that was put on a Sunny Boy, would give SI some smoothly responding power. May still stumble under some loads, as frequency shift knocks off microinverters. Sunny Boy frequency range could overlap microinverters or be higher. Best I think if the SB array is in full sun same time as microinverters; if sleeping it won't be able to help.

I just ordered Midnight Classic 150 and MNSICOMM (also Wiz Bang Jr and shunt) to use either with SI (UL listed unlike my SIC-40), or for a vehicle mount system.
 
Have you ever looked at the current output waveform of the Enphase branches? If not, I suggest you check. It seems that you encounter the problem more often in the afternoon? From what I have seen of Enphase micros, the current output can become very jittery during the afternoon hours as the sun fades that I believe is due to MPPT hunting. So, I suspect the MPPT hunting adds additional disturbances, on top of the anti-islanding perturbations, that further destabilizes the island grid during the afternoon hours.
I haven't checked with the scope more than just at the SI inverters, but yes, it would be nice to see what various points in the system look like. One of the things I tried to help the problem was adding more ferrite cores on some of the branches of the system. I'm not sure it had an effect, but at one point the cores were preventing good communication with the envoy and I removed some of them. I have noticed this summer that the garden array (furthest away from it's envoy) loses communication as the solar power is higher, the envoy is not getting data from those inverters. It seems like later in the day it recovers and the data is available and shows a reasonable day total. Only my logged data, taken from the envoy every 15 min shows the issue.
One challenge with having 35 micros vs a single string inverter of the same capacity is that the micros do not synchronize their anti-islanding perturbation or MPPT hunting. The result is almost constant disturbances in an island grid. Most off-grid inverters like the SI don't have the ability to actively reject such disturbances especially at the sub-cycle time scale. Instead they rely on the system inertia to handle such disturbances. Also, I suspect the topology of your micros with multiple branch configs and inverter models further complicates the problem. Not sure you already tried this but you may want to incrementally turn off micros while the problem is happening with the A/C on to see if that helps. Anecdotally, it seems more problems occur when the AC couple ratio between the off-grid inverter and micros rated output capacity is < 2:1.
Agree, having more micros on does seem to make it worse. The previous summer, I didn't notice an issue, although it may have been there. Previously, some of the solar was DC coupled to the SI battery, so even at a balanced solar vs. total system load, the SIs were inverting 1-2kw. Now those panels go to another battery and then micros from the battery back to the system. I thought I read a patent from Enphase about how the micros do synchronize their anti-islanding.

SCR chopper will likely cause havoc.
I'm guessing so. Anyone have an idea for a variable load controller for WH element that would be nice? Maybe a motorized variac.... It's a dump load, and heat in the basement is a reasonable loss, so maybe that's ok. It would be more mad science in the basement, so that's a plus.
 
I haven't checked with the scope more than just at the SI inverters, but yes, it would be nice to see what various points in the system look like. One of the things I tried to help the problem was adding more ferrite cores on some of the branches of the system. I'm not sure it had an effect, but at one point the cores were preventing good communication with the envoy and I removed some of them. I have noticed this summer that the garden array (furthest away from it's envoy) loses communication as the solar power is higher, the envoy is not getting data from those inverters. It seems like later in the day it recovers and the data is available and shows a reasonable day total. Only my logged data, taken from the envoy every 15 min shows the issue.
I doubt ferrite cores have any effect on anti-islanding perturbation or MPPT hunting variation as they are low frequency.

Agree, having more micros on does seem to make it worse. The previous summer, I didn't notice an issue, although it may have been there. Previously, some of the solar was DC coupled to the SI battery, so even at a balanced solar vs. total system load, the SIs were inverting 1-2kw. Now those panels go to another battery and then micros from the battery back to the system. I thought I read a patent from Enphase about how the micros do synchronize their anti-islanding.
Patented concepts don't always work in the real world. The micros probably try to synchronize via Envoy comms but in reality I don't really see it.

I'm guessing so. Anyone have an idea for a variable load controller for WH element that would be nice? Maybe a motorized variac.... It's a dump load, and heat in the basement is a reasonable loss, so maybe that's ok. It would be more mad science in the basement, so that's a plus.
Sinusoidal AC power diversion controllers are available. The question is how much are you willing to spend? They are much more expensive than simple SCR choppers.
 
Searching for droop control or Q(V) control finds these SMA documents, and they may be relevant if thinking in general terms about an off-grid microgrid. https://files.sma.de/downloads/EDMx-Q-V-R11-BA-en-12.pdf https://files.sma.de/downloads/IPC-QoD24-7-STP-TI-en-17.pdf

Some inverters, at least the SMA PV solar inverters, can be tuned or set to adjust for power factor of other loads. Maybe the newer microinverters have a setting for this, but I see just 1.0 PF available in my grid profiles and I don't think I can change that.

This ongoing pondering gives me a couple more ideas to try (next summer....):
1) In the SI menus, there's a parameter for how many kw of solar is connected to the system, under 231.09 and 231.10. I think those just to do with the feed-in operation when connected to grid, but maybe they have some other consequence. Does anyone know what they do for sure? Maybe this would matter if the SI were assuming that there's xx kW of AC coupled (SMA SunnyBoys) which would behave in a certain way to maintain the recent average Hz, or the Hz requested over rs485.
2) I could flip some breakers and remove interlocks and change the layout of the system.
This might change the PF at the SIs, and have some effect on behavior? I forget, maybe I tried this already.
Current: GT PV micro array --------- bsmt panel & A/C unit ------------------ garage SIs ------------garage battery micros, and garden micros
Experiment: GT PV micro array ------------------------------ garage SIs ------------garage battery micros, and garden micros
bsmt panel & A/C unit ------------------/
 
I'm not finding 231.09 in my manuals.

Reactive power stuff (probably not the same as inductive reactance I usually think of) mentioned in Data Manager manual I think is for grid support, not off-grid operation. There is also mention of total kW of PV in there.

I'm guessing badly behaved microinverters are your problem.
 
Back
Top