What generator works with SMA sunny island?

[callmeburton]

TLDR:
Duromax XP13000EH works (confirmed with own offgrid setup)

Original Post
I have a briggs and stratton storm protector 5500 which the SI reads as in spec but gets stuck at "warm" status till it finally errors out many minutes later with a backfeed error, default is 100w and 30sec.

Does anyone have a list of approved generators for the SI?

The SI is configured as offgrid with gen option. I have tried to change many things and test other settings and I am getting no where but poorer for burning fuel ?

 

[toms]

I have a briggs and stratton storm protector 5500 which the SI reads as in spec but gets stuck at "warm" status till it finally errors out many minutes later with a backfeed error, default is 100w and 30sec.

Does anyone have a list of approved generators for the SI?

The SI is configured as offgrid with gen option. I have tried to change many things and test other settings and I am getting no where but poorer for burning fuel ?

I use a Kubota J108. The SMA generator thresholds are configurable - check your settings.

 

[callmeburton]

I use a Kubota J108. The SMA generator thresholds are configurable - check your settings.

I have default settings which are pretty wide. The SI should "connect" after two minuets but it doesn't though it shows in spec and in warm up phase. Then ten minutes later or so it fails with a backend error.

I looked at the generators manual but couldn't find specs on back feed or AVR

 

[RCinFLA]

Phase acquisition and phase tracking has a low bandwidth on hybrid inverters. This means the AC input has to be relatively stable in frequency and phase. About 0.3 Hz/second of frequency wobble is about the limit of tracking bandwidth. This is entirely different than the frequency range limits setup in the inverter. Just because a generator output is within frequency range does not mean the frequency output of generator is stable. You will not be able measure slight frequency wobble on a frequency counter on DVM. The count gate time averages out the wobble. Smooth sound of generator engine speed is a better indicator. You don't want to hear a 'wha-wha' sound pitch change.

What you are calling 'warmup' is phase locking search. The inverter will slowly change its AC frequency, scanning up and down in frequency in an attempt to get within capture range of phase locking and tracking by the inverter. On most units, the wider the frequency acceptance range span setup in inverter the longer it will take to scan for phase lock.

It does not start the frequency scanning until the inverter sees AC voltage present on AC input. The pass-through connect relay is open during this process. If the voltage sensing circuit is defective it will not detect the AC input voltage and will not attempt a connection. It is common for inverter to have a small step-down transformer that takes AC input voltage down to about 12vac to be detected by inverter control processor. Some cheap inverters just use a resistor divider network, sometimes with an optical coupler to help protect low voltage processor.

If you see an indication that inverter has detected an AC input, then issue is either:

1) Too much wobble in AC input phase/frequency. Common issue on synchronous generators is due to unstable engine rpm governor control.
2) Frequency out of acquisition range set in inverter. Inverter will not scan beyond its setup acceptance limits of frequency range.
3) AC input voltage out of acceptance limits of inverter.
4) Possible defective pass-through relay in inverter. Pass-through relay contacts take a beating from grid glitches. When grid glitches occur, it causes a momentary overload on inverter causing pass-through relay to release from AC input under high disconnect current, arcing relay contacts and wearing their contact surfaces.
5) Although rare, it is possible for generator to have an unsymmetrical waveform for positive and negative half cycles of AC output. Inverter will not likely lock to an unsymmetrical generator waveform where center crossing duty cycle is not close to 50-50% between positive and negative AC swings.

Inverter does not initially test for generator distortion. After it connects to generator, a poor AC input waveform will not be a good match to inverter's sinewave causing current spikes across the AC waveform cycle. Inverter may quickly disconnect if the peak current spikes get too great. Generator distortion may not get bad until there is significant load current on generator.

 

[callmeburton]

Tried another green today, generac by briggs and stratton, wheel house 5500. No dice. Voltage was pretty high but in range, 130 per leg. Bounced between 126 and 130 range.

Also tried to configure as grid and grid Gen and neither worked.

There is a Honda with digital Avr i can rent and might try that.

 

[callmeburton]

Update, and an odd one.
I am seeing 200VAC on S1's AC2 on occasion with nothing hooked up to it.

Also S1 has a asterisk after it which normally means AC2 input is in spec.

 

[callmeburton]

Behavior mentioned previously. I have turned off breakers at main panel and it didn't change. Also turned off, though not fully, the main unit and started, again no change.

Haven't placed a load on the voltage source yet.

 

[callmeburton]

So after reading the white paper I realized none of the generators likely fall into the range of something that would work with the SI less it mimicked what a synchronous generator outputs :/

I did try to switch to frequency / current today to see if it would accept the gens I have on site but it didn't work.

So what I need to track down is a "old fashion" brushless synchronous generator and see if that "fixes" the issue ... or find someone in the US with a reasonably priced unit that works with their SI to tests against. But givn no one in the US has provided anything it seems hopeless 0__0

 

[callmeburton]

Did a dive last night into more generator research and other peoples experience with SMA ... it is a mess for sure.

I found one US guy with a Kubota GL7000 generator that "worked" only if he turned the governor down on the generator ... but that is a $7k generator so I can't easily test that one. (details here)

Another US guy went through 5 generators with SMA support, with support blaming each unit, and the original generator, non of these named, "worked" after SMA replaced the whole unit. His problem was a little different though as the relay would close then open after a couple seconds if I recall correctly. (Details here)

Another guy, same forum but can't find the post, mentioned SMA had him change one setting and his generator "worked" ... no mention of the generator or what setting was changed. People suspected he was set to grid only not grid/gen but idk.

All of this leads me to think I need to rewire my system with an external battery charger ... I am not sure how this would interface with the BMS or SI but I know it will have to through my already crowded shunt / battery terminals.

 

[RCinFLA]

Once connected to AC input source, detecting an open circuit at AC input and re-open pass-through relay is somewhat difficult for hybrid inverter as zero AC input current is a legit condition, not requiring pass-through relay to be released (opened).

Important thing to remember, when disconnected from AC input, inverter is its own master for freq/phase/voltage. When connected and sync'd to AC input, the inverter is slave to AC input freq/phase/voltage. Pass-through relay open = inverter is its own master. Pass-through relay closed = inverter is slave to AC input.

Common method inverter's use to detect AC input going open circuit, like opening AC input circuit breaker, is inverter phase tracking is biased to gradually raise its frequency when there is no phase correction updates on zero crossings of AC sinewave voltage. Without AC input to keep correcting the phase, the inverter will gradually rise in frequency until it reaches its high frequency range limit at which point it will release pass-through relay. This can take up to a minute or two before inverter goes out of frequency range in slave mode causing pass-through relay to open, during which time there is inverter generated AC voltage on AC input terminals of inverter.

Another method to detect open AC input is counting the phase correction pulses. Too long a period without seeing some phase corrections is an indication that AC input has been disconnected.

When you have an AC output load on inverter it may detect the sudden drop in AC input current when AC input is opened and pick up AC output load on battery/inverter power. This will cause inverter to immediately release its pass-through relay.

Widening the inverter AC input acceptance frequency limits can increase the AC input release time when AC input circuit breaker is opened.

You should never disconnect AC input, then reconnect it before inverter's pass-through relay has been released. This can cause a phasing mis-match that causes a high surge current on inverter and AC source. This is also why an automatic transfer switch should not be used on AC input of inverter unless transfer switch has a totally open crossover time delay long enough to ensure inverter's pass-through relay has released before connecting to another live AC source.

 

[callmeburton]

Good tips but I don't think it will resolve my issue where the SMA sees the inverter, knows it is "in range," and tries to sync with it but can't after ten minutes then locks it out throwing an error relating to current feedback over time into the generator.

Update, and an odd one.
I am seeing 200VAC on S1's AC2 on occasion with nothing hooked up to it.

Also S1 has a asterisk after it which normally means AC2 input is in spec.

Update on this though.

After dark I turned off all the loads side of the SI, voltage was still present. I put a traditional lightbulb on the leads with the voltage and the only thing I noticed was the SPD connected to AC2 would change in intensity which is interesting. I wonder if the SPD is feeding a voltage to AC2 ... I will remove it later to check this case as well.

 

[RCinFLA]

You also have to be careful of too much path resistance in AC path with generator.

If path loss causes an AC voltage drop at or below inverter's AC low voltage limit the inverter will release from generator.

Within voltage limits of inverter, voltage drop on generator AC line path due to inverter ACout load current and AC sourced charging load will cause the inverter to reduce its output voltage to follow the generator voltage at connection within inverter. If load is suddenly turned off or reduced, the inverter gets a momentary current back push surge current until it reacts and corrects its AC voltage output to match the higher AC voltage. This back surge is pushed to battery on a LF inverter and when battery is fully charged it may cause inverter to release from generator.

Generator also has a reaction response time lag to regulate its output AC voltage due to load changes.

 

[callmeburton]

If path loss causes an AC voltage drop at or below inverter's AC low voltage limit the inverter will release from generator.

Voltage measured at the generator matches the voltage at AC2 inputs, the run is only 4' right now.
I have de-rated the amps in the inverter of the generator down to 15 amps, despite the gen being rated at 30 as well for my tests.

The way the SI is suppose to work is it sees input on AC2 and decides if it is in range. If in range of allowed results it "warms up," during this time (typically 2 minutes) it is trying to sync with the generator. When it does finally sync it connects to the gen. At the same time I tells the SB's to stop feeding the loads side of the panel (AC1) to protect the generator if I recall correctly.

 

[callmeburton]

I am seeing 200VAC on S1's AC2 on occasion with nothing hooked up to it.

Update, I disconnected the SPD and the phantom voltage disappeared. Looking at the SPD it is a 300VDC not AC SPD by MN solar so that could be cause.

Re running all gen experiments now with current generator on site but nothing seems to work yet. Currently getting GnFailLock, no reasons provided, so I assume it failed to sync. This is with a load applied to the gen before "starting" generator, and without.

 

[Hedges]

After dark I turned off all the loads side of the SI, voltage was still present. I put a traditional lightbulb on the leads with the voltage and the only thing I noticed was the SPD connected to AC2 would change in intensity which is interesting. I wonder if the SPD is feeding a voltage to AC2 ... I will remove it later to check this case as well.

Update, I disconnected the SPD and the phantom voltage disappeared. Looking at the SPD it is a 300VDC not AC SPD by MN solar so that could be cause.

SPD is MOV, which has some capacitance. Maybe this completes a resonant circuit with other parasitic elements.

What you are calling 'warmup' is phase locking search. The inverter will slowly change its AC frequency, scanning up and down in frequency in an attempt to get within capture range of phase locking and tracking by the inverter.

Sunny Island has a programmed warm-up period before connecting, and cool-down before shutting generator off.

The frequency synchronization is performed when it wants to connect. I've observed that with grid, have never used generator.

According to documentation, it first raises frequency high enough to knock Sunny Boy offline so it doesn't have to deal with unwanted power while it is down around 60 Hz for synchronization. Sunny Boy is returned to UL-1741 by RS-485 signaling (if it is an older model, not newer with Rule-21). It remains offline for 5 minutes before reconnecting.

Although I haven't seen SI lose the race, I did observe that SI raises frequency to 65 Hz and knocks SB offline the moment it has qualified input AC as good. It then drops to 59 Hz (used to correct mechanical clocks) and waits its own 5 minutes before synchronizing (to grid, don't know delay for generator.)

With old UL-1741 Sunny Boy, 59 Hz is also out of spec, so it doesn't start its 5 minute countdown. With UL-1741-SA Sunny Boy, 58.7 Hz is lower limit of spec, so Sunny Boy and Sunny Island are both on a 5 minute timed race to connect. I've been concerned that Sunny Boy would connect first, but didn't see that happen. In the event of unstable generator rather than stable grid, the 5 minute timer might elapse for both.

I've widened my Sunny Boy grid frequency range to the 59.7 ... 60.5 Hz range of UL-1741-SA and have not seen them reconnect too soon.

Re running all gen experiments now with current generator on site but nothing seems to work yet. Currently getting GnFailLock, no reasons provided, so I assume it failed to sync. This is with a load applied to the gen before "starting" generator, and without.

OK, you've tried loading the generator during the time Sunny Island is waiting to connect?
I thought that might make RPM more stable, but apparently didn't fix the issue.

I've thought about transformer-isolating AC (from grid or generator) rectifying it, and feeding PV input of another Sunny Boy (with frequency-watts ramp from 60.5 Hz to 61.0 Hz, assuming others are older ones with off-grid 61.0 Hz to 62.0 Hz ramp.) An AC coupled Chargeverter. Be careful about capacitor inrush current; Sunny Boy is intended for relatively low current sources like PV or turbine.

 

[callmeburton]

I will have to observe the SI when it gets passed my 30 sec "warm" period I have programmed in right now.

I believe I lowered all times required to test the gen on the SI side as well to make it faster for me to test ... forget which settings those were though.

I haven't tried to cut off the SB completely and then mount the gen on freqcur and cur (current setting) so that might be worth a try as well as there wouldn't be a race condition.

 

[callmeburton]

Small update:
Was looking for YT videos with people who are using the SI's with a generator and found the portable light trailers you can find all over the US with the SI's are using a kubota low boy.
Another video a guy from a company demonstrating the autostart features was using a all house generator from generac.

 

[callmeburton]

research update:
Found a company selling trailers again, SI's installed, and a Duromax / Durostar generator (13000eh) used as the backup power. The seller indicated this is an auto start feature but the duromax has a push button start (hold for 1 second) and a push button stop (hold for 3 seconds) feature. RPM's seem correct though, 3600, and it has a AVR ...

Given the cheaper cost of this I am debating getting one then rigging up some relays to simulate the on off signals this thing expects.

 

[callmeburton]

Update on this:
Had to break down and get a gen, went with the previously mentioned xp13000eh since I have seen it on a light trailer for sale with SMA SI's in there ... I think it gets further than my previous experiments which would just lock the gen out as the SI looses sync before it can even put a load on it.

Tried the basic of adding a load the gen to help stabilizing the output, light loads still sync loss, heavy loads gen lock.

Gen is pretty nice looking but will likely have to go with an external charger and wire it up so the shunt can see it.

 

[Hedges]

Is your Sunny Island configured to expect tightly controlled utility grid, or sloppy generator?

 

[callmeburton]

#231 is set to Gen
#234 settings which are currently not stock
GnOpTmMin ) 1m
GnStpTmMin ) 1m
GnCoolTm ) 1m
GnErrStpTm ) 0h
GnWarmTm ) 10s
I think GnAlSns is Medium, not sure as gen isn't running

Doing manual start tests. If I go to the compact meters (pretty much only use this now) I can see the External source voltage and frequency there and it is in range. I am semi optimistic about this gen working if I change some settings as it did give a sync lost error while as before it only gave genlock errors on other generators.

 

[Hedges]

#231 is set to Gen
...

Doing manual start tests. If I go to the compact meters (pretty much only use this now) I can see the External source voltage and frequency there and it is in range. I am semi optimistic about this gen working if I change some settings as it did give a sync lost error while as before it only gave genlock errors on other generators.

Is that 231.05 = Gen?

If so, that should support wide voltage and frequency range, 80 ~ 150V, 54 ~ 66 Hz.

Update on this:
... I think it gets further than my previous experiments which would just lock the gen out as the SI looses sync before it can even put a load on it.

Tried the basic of adding a load the gen to help stabilizing the output, light loads still sync loss, heavy loads gen lock.

If load needed to get SI to connect, I would think that meant it was out of range under no-load.
In range according to compact meters, you say? Likely slow update, try DMM. I have one that includes Hz.

What voltage and frequency unloaded? Under heavy load?
Maybe it is running a bit high, and set lower would work OK. SI should tolerate lower voltage and frequency so adjusting might work.

From manual, I see 231.06 in a paragraph, but may be typo. Table has 231.05

%%
An auxiliary contact is used because the Sunny Island must "know" whether it is connected to the utility
grid or whether it must manage a diesel generator.
To enable such an operation, you must set parameter "231.06 ExtSrc" to "GenGrid" (see Section 8.2
"Starting the Quick Configuration Guide (QCG)", page 69).
%%

%%
19.2.3 External Settings (230#)
231# Ext General

No. Name Description Value Explanation Default
value
01 PvFeedTmStr Start feed-in operation 04:00:00
02 PvFeedTmStp Stop feed-in operation 22:00:00
03 ExtLkTm Lock time after reverse
power or relay
protection
0 min to 60 min 20 min
05 ExtSrc generator and grid
operating mode
PvOnly PV only PvOnly
Gen Generator
Grid Utility grid
GenGrid Generator/
utility grid
12 ChpEna Combined heat and
power plant
Disable Deactivated Disable
Enable Activated
%%

 

[callmeburton]

I will need to run gen tomorrow most likely so can get some more data. I don't have a DMM with frequency unfortunately.

Is that 231.05 = Gen?

Yup

If load needed to get SI to connect, I would think that meant it was out of range under no-load.
In range according to compact meters, you say? Likely slow update

I am not sure it was required to get it to connect ... still trying to recall the quickest way to know if it fails to connect and for what reason. Typically right now I listen for relay changes and look at the compact meters and the failure history logs and specifically the time stamp to know what likely happened.

What voltage and frequency unloaded? Under heavy load?

IIRC the gen will say 247v 61hz with no load (it has its own meters on the dash) ... and the SI will say 121 and the Hz will bounce a bit from 60-62? unloaded.
With a light load the bouncing isn't as much and likely only in a 0.8 range.
I don't recall the change with the heavy load (my meanwell charger which should have been pulling 2kW of power at the time) but I recall the gen following a rhythmic up / down in ranges but all where "in range" and nothing was really close to out of range. This was subtle and likely hard to notice less you were watching the gauges on the SI for Ext.

My current non-default changes were previously set to help me test gen failure faster. Are there any I should revert that are fighting me or was my direction the correct one?

 

[Hedges]

I don't know, haven't tried generator.
I could imagine frequency changing rapidly it wouldn't like.
When I was debugging SI/SB interaction, DMM with Hz scale was handy, could watch it swing.

The fact loading helps is a clue. Maybe governor on your generator needs to be tweaked. Or maybe it is not able to be stable.

 

[bolek_bolek]

Not sure if it helps, but was it mentioned if the generator used is an old school synchronous type, or an inverter style? They will act quite differently depending on what the load does.
In first case the "load" which is the SMA will be pushing energy back into its coils at slightly higher frequency and increase the RPM, which the generator's regulator might consider load drop and react so to reduce field current and fuel . Even if it's only reactive power travelling back and forth , the EMF is still there and affects the synchronism inside the machine. Then the SMA will see a different AC frequency and act on it, and the two loops (mechanical governor and PLL) will chase each other. A current measurement on the wire between the gen and the inverter might bring some light on what's going on, even if the inverter has watts displayed we are talking VARs here