diy solar

diy solar

sma sunny island grid feedback question

grgekeys

New Member
Joined
Jul 8, 2021
Messages
2
I recently bought one of those DC Solar auction trailers.

I've connected it to the house grid and enabled the back feed function where it feeds back extra energy from the solar panels not being used by the charger.

the problem is im only seeing half the power expected @about 1kW instead of the 2.5kW that's being generated by the panels.

does anyone know if there is a setting on the sma sunny islands that controls how much is fed into the grid?
 
With DC coupled PV, if SCC charges battery higher than Sunny Island wants then Sunny Island will suck power from the battery and feed it to grid, of so enabled. Perhaps a higher target voltage on Midnight or lower on SI would deliver more.

Are you sure 2.5kW is being generated? Or can be? If it is generated, it is being consumed somewhere. If Midnight is moving off Vmp to reduce output because battery voltage is high enough, then there is an adjustment to be made somewhere. Maybe you're just having a bad production day due to haze, heat, angle? Ten 250W panels for 2500W STC aren't normally going to deliver a full 2.5 kW, but certainly more than 1kW.

A question no one has answered for me yet regarding these trailers ... Is there a battery shunt connected to Sunny Island so it knows how much current Midnight delivers? Or, is there a data cable from Sunny Island to a device MNSICOMM and from there to Midnight?


I don't know if SI is capable of grid export when it is capable of commanding Midnight.
 
yeah it's definitely producing roughly 2.5 at Times according to the midnite classic.

I've attached images of the classic and the two sunny islands, you can see the master reporting the power at 1.8 kW and the slave at .9kW it seems like the slave is the one doing the feed back into the grid.

20210710_124139.jpg
here's the master
20210710_124157.jpg
here's the slave20210710_124205.jpg
 
I think that's one inverter sucking down -0.9 kW from grid and charging batteries, while other is pulling 1.8 kW from batteries to produce AC.

+1.8 kW - 0.9kW = +0.9 kW net AC production from battery.
2131W on Midnight - 900W consumed by SI = 1200W going into battery.

Both have a check mark next to grid connection. Suppose that means both connected.
I think they are set up for 120/240V split phase. Do you have both inputs fed from grid?

Master shows relay #2 closed? Probably default > 30% SoC, no need to shed load.


Does the trailer have a battery shunt, with sense lines to Sunny Island?
Does it have a data cable from Sunny Island to Midnight (probably through a small module to convert protocol)?
 
I have two sunny islands and am purchasing 40-60kW of lifepower4 batteries from signaturesolar. How does the gridfeed mode know to feed battery power back into the grid (e.g. shunt, 3rd party data, etc?) That is what has concerned me is how the si's figure this one out. On a solark it makes sense as all of the current is going through the transfer switch and the solark can measure production vs. usage and balance it out with pushing ac back to the grid to net this one out.

Even though I have 8kw of solar panels and produce at maximum around 6400 watts as seen on the sb7000us, I want to be able to gridfeed export to shave my usage at night and after 4pm. In California we are now at .48c per kwh between 4-9pm and .25c per kwh after 11pm through 7am. The mid-tier between 7am to 3pm is about .32c per kwh. I have a time of use (ETOU-B) setup w/ PG&E and want to stop the $2500-$3000 per year bleed in true-up costs at the end of the year. I am willing to spend $15k to solve this problem - easy math in terms of payback. I own an electric car now as of July and need to charge at night - so I am leaning towards the 60kw system w/ gridfeed mode. Any help as to understand how this is supposed to function would be appreciated. My original purchase in buying a pair of 5048's on the cheap was to just get backup power and use my old generac 16kw w/ the si's to have power for weeks in a situation where PG&E turns off power as we have PSP insanity here at this point.

I produce about 30-35kwh per day in the winter and 55-60kwh per day in the summer on the existing solar. I ended up getting another sb6000us with the sunny islands and have another 2-3kw of panels I will put on my house to maximize production.

Thanks
 
I don't think US model Sunny Island will feed grid from battery (with supported equipment.)
It will let GT PV backfeed grid, if told to do so.
If DC coupled SCC drives battery voltage higher than SI wants, it can feed that to grid.
But if battery is charged and you want to run battery down, I think you would have to configure a lower target battery voltage in SI. (Perhaps people have done that with a gizmo talking to it.)

Other models like Sunny Boy Storage are designed for peak shaving, and I think shifting export.
I would imagine the new European model Sunny Island also offer that.

Lithium batteries - probably best to have a model with BMS that talks to Sunny Island. Otherwise, you have to tell SI it is VRLA and some of the communication enabled functions are unavailable.

You may be able to program SI to disconnect from grid during peak times, connecting only if battery gets low. If nothing else, with a timed relay. Certainly you can use grid as generator, never export.

SMA does have some newer monitoring and control hardware. That can tell Sunny Boys what to do via SpeedWire, but I'm not aware of them commanding Sunny Island.
 
I don't think US model Sunny Island will feed grid from battery (with supported equipment.)
It will let GT PV backfeed grid, if told to do so.
If DC coupled SCC drives battery voltage higher than SI wants, it can feed that to grid.
But if battery is charged and you want to run battery down, I think you would have to configure a lower target battery voltage in SI. (Perhaps people have done that with a gizmo talking to it.)

Other models like Sunny Boy Storage are designed for peak shaving, and I think shifting export.
I would imagine the new European model Sunny Island also offer that.

Lithium batteries - probably best to have a model with BMS that talks to Sunny Island. Otherwise, you have to tell SI it is VRLA and some of the communication enabled functions are unavailable.

You may be able to program SI to disconnect from grid during peak times, connecting only if battery gets low. If nothing else, with a timed relay. Certainly you can use grid as generator, never export.

SMA does have some newer monitoring and control hardware. That can tell Sunny Boys what to do via SpeedWire, but I'm not aware of them commanding Sunny Island.
Hedges

I currently design and manufacture automotive microcontrollers for various projects as a business and have canbus w/ a automotive microcontroller design already in-hand that can handle over 100v input and has protection diodes on all inputs to survive harsh transients under the hood of a vehicle. These can talk 500kbps canbus to SMA-NET. If you follow the pylontech specific documentation, ID's 0x355 need to be reported for SOC/SOH, and ID 0x356 for battery voltage, current and battery temp. Since the sunny island does not read the battery voltage in lithium mode, this data can be fudged to values which might allow for grid-feedback assuming the SI logic is still sound as a ratio of how much overvoltage to grid-feed. Based on how much overvoltage is reported, you could gridfeed back and reverse engineer this to allow 2-4kw potentially. The BMS on the lifepower4 never receives messages from the canbus of the SMA - so my device will be using rs485 and relaying the data from the lifepower4, converting it and transmitting this over canbus to the SI's. This sounds like peakshaving can work on paper - getting data from another device like a victron shunt or other device like a sense to measure current to determine how much current is going to/from the grid is necessary to feed into the microcontroller (e.g. man in the middle) to fudge the voltage parameters. Obviously too low of voltage reported and we have major problems - including possibly turning on the generator connected to the SI from the sunny island. It might be a better idea to command the generator from the MIM (man in the middle) microcontroller potentially in this condition as well in case something happens between the RS485 and Canbus controllers.

Do you happen to have any documentation on the sunny boy storage devices which likely use SMA-NET to accomplish peak shaving? I could see if the SI's might respond to the same ID's on the canbus on the newer software. My Si5048's were just updated to the newest software a few months ago to support lithium w/ homegrid. The homegrid batteries did not report SOC/SOH to the SI's and thus I picked up the SI's cheap.

Id be happy to talk to you on the phone and discuss this further. I can send you my personal in a private msg if you'd want to chat. Thanks - Brock
 
I don't have any coms info for Sunny Boy Storage.
There is YASDI library for Sunny Island.
I haven't tried any coms programming, just cabled things together and used Sunny Web Box or Sunny Explorer.
Some others here I think have talked to SI.

Sunny Boy Storage requires a high-voltage battery, don't know of anyone who has done DIY.
I would like to know if it could respond to frequency shift, not just UL-1741-SA but both charge and discharge, AC coupled to SI.

I also have an interest in Speedwire export limiting of Sunny TriPower
 
Hedges

It looks like this guy here has rev engineered some of the canbus status messages and such using the yasdi library.
--https://github.com/sshoecraft

I am sure w/ your 3 phase setup you would love to find a mutlicluster box, they are $13k. I am sure most of it is just some can relaying and centralizing some switch controls and added relays. I would personally like to have a full 100a on each phase of my service. Maybe more.

This same person (sshoecraft) appears to be able to set configuration settings remotely as well - this might be critical to limit current/loads in a grid feed scenario in real time. It appears that these utilities will be crucial to validating whether the canbus messages sent from the bms side are successfully being set on the SI side for lithium.

The SI manual states for grid feed mode - Section 14
AC feed-in generators on the stand-alone grid side (Sunny Boy) can feed their energy into the utility grid through the internal transfer relay of the Sunny Island; limitations (see Section 14.1.6 "Limits and Power Control", page 126). The following illustration shows the direction of energy flow for the "Net Metering" and the energy consumption from the utility grid. Grid feed-in from the DC side into the utility grid In order to allow electricity to be fed from the DC side into the utility grid, the battery voltage in a charged battery (on the utility grid) must be increased by external DC chargers or the Sunny Island Charger above the nominal charging voltage.

According to fortress battery, they state to manually set the system up initially for lithium as VLA, in the situation where the BMS CAN communication fails these settings are then made effective when using the sunny island. See doc attached. That is a really good starting point for anyone who is going to use an unsupported bms in open loop.
 

Attachments

  • SMA-Integration-Guide-Fortress.pdf
    390 KB · Views: 8
Back
Top