SMA Sunny Island no longer agrees with charge controller(s)

[Trukinbear]

Hi guys, my 5year old system seems out of whack.

My networked to my SI Midnight classic charge controllers at this moment say I'm at 97% SOC, and they're putting a 'trickle' of 20A to the battery, BUT the SI6048 says it's 39% SOC and wants to run the generator and put 120A to the battery. Help

SMA SI6048
SMA network box
2-Midnight Classics
684 AH 48V forklift battery
6kw solar panels
8kw diesel generator
175 gallons (currently) diesel

 

[Ampster]

What voltage are you measuring on your pack? When was the last time you had a full charge? Sometimes it takes a full charge for the Coulomb counter to reset to 100%. Presumably your charge controller is putting the correct current into your batteries. Lead Acid batteries like a long Absorb time so it could not hurt running your generator or rebooting the Inverter if it does not reset.

 

[Trukinbear]

 

[Hedges]

I thought you had Classic on the data bus to SI, so SI told them what voltage to do. (and that required shunt connections to be different from when they don't communicate.)

By the way, Welcome Back!

I'd expect 39% to be such a low voltage that Classic would deliver all it could.

Latest voltage around 60V shouldn't look like 39%.
Shunt issues?

 

[Trukinbear]

What voltage are you measuring on your pack? When was the last time you had a full charge? Sometimes it takes a full charge for the Coulomb counter to reset to 100%. Presumably your charge controller is putting the correct current into your batteries. Lead Acid batteries like a long Absorb time so it could not hurt running your generator or rebooting the Inverter if it does not reset.

I'm showing 53.4 V with the generator disconnected from the system for about an hour now - I was literally boiling the battery. Last full charge 4 days ago, and 80% soc cutoff (rain/generator) since. Charge controller says it's full and in float. SI says 40% and it's mad that it can't get the generator to start

 

[Trukinbear]

Oh, looking back through my charts I see what is happening. Even though the charge controllers, and my SI6048 agreed that the battery was at 88% SOC the SI 6048 suddenly dropped it's version of SOC to 20%, went into Battery Protection mode, and cranked the generator. The generator however is controlled by the SI6048 so it's only actually putting ~20A to the battery because the Si6048 is being told by the Midnight Classic 250's that the battery is at 88%.

This is where my forklift battery shows it's unsuitability for off grid solar, at least with a SI6048. Now at 5 years we're looking at some loss of capacity for sure, but the biggest issue is (according to the engineer at Big Sur) that my battery's 968AH capacity is not at 48V - a forklift can pull these down to 40 and still deliver a half day of operation without damaging the battery.

My SOC reset occured late at night, when the battery is coldest and under a very light constant load of 15-25A, and the heat pump kicks on at 5am the voltage (momentarily) dropped to 46.0v from what looks like around 47.2v - the SI6048 even though it has calculated the SOC using the Whizbang Jr at 88% suddenly saw 46.0v and went into Battery Protection mode.


IF via QCG - new battery - I set the nominal battery voltage below 48, the battery should still be charged as the 48v battery it is because all my cell voltage charge setpoints etc. I'm hoping that if at 48v nominal it cuts off 2v below at 46v, that if I set nominal to 47v it'll cut off at 45v, and not do much else.


Thoughts?

 

[Hedges]

I didn't know SI read SoC that Classic computes with Whiz Bang. Thought Classic with adapter was just emulating SIC-40.

I have an SIC-40, got its adapter and used it on single SI-5840.
(I also have Classic, Whiz Bang Jr, and adapter, but not using yet.)

I thought SI determined SoC by its own measurements or by external analog shunt.

You can set 46V nominal vs. 48V nominal; then pack voltages will be calculated based on VRLA cell voltage settings x23 instead of x24. You would have to increase bulk & float accordingly. People running 16s LiFePO4 open loop I think use 46V nominal VRLA to achieve the desired voltage. I don't know if 46V will let you hit the voltages you want or not. with 48V nominal, can you not reduce low voltage limits as desired? Is there a cell resistance parameter? (I don't see one). Do you have suitable temperature coefficient?

Of course, you'll be using FLA not VRLA.

 

[Trukinbear]

I didn't know SI read SoC that Classic computes with Whiz Bang. Thought Classic with adapter was just emulating SIC-40.

I have an SIC-40, got its adapter and used it on single SI-5840.
(I also have Classic, Whiz Bang Jr, and adapter, but not using yet.)

I thought SI determined SoC by its own measurements or by external analog shunt.

You can set 46V nominal vs. 48V nominal; then pack voltages will be calculated based on VRLA cell voltage settings x23 instead of x24. You would have to increase bulk & float accordingly. People running 16s LiFePO4 open loop I think use 46V nominal VRLA to achieve the desired voltage. I don't know if 46V will let you hit the voltages you want or not. with 48V nominal, can you not reduce low voltage limits as desired? Is there a cell resistance parameter? (I don't see one). Do you have suitable temperature coefficient?

Of course, you'll be using FLA not VRLA.

It figures it all those ways, yes it sees the Midnight Classic 250s as 2 SIC-40 charge controllers, AND the data from the Whizbang is used by the Classics... Usually this all agrees, they all agree on Ah in/Ah out. However the SI also monitors the battery voltage, and 46v triggered an SOC 'recalibration.'

 

[Hedges]

Seems like you need a way to adjust the "recalibration" voltage.
Not all parameters seem to be user-settable.
And some things that ought to be computed off settings are apparently hardwired. For instance, setting it for 50 Hz operation (used in some parts of the Americas), it didn't function as expected, maybe the frequency shift.

Since it supports removing bad cells, I'd think 46V nominal would support operating to lower voltages. Just don't know if that will allow desired charge voltage. (If Classic was not connected by data cable, then Classic would charge as before, only charging by generator would be affected.)

So far I'm 3 years into SunXtender AGM. They only get cycled during grid failures.

 

[daklein]

This does look a little similar to what mine did today. It must be the solstice.

SMA SunnyIsland 6048 lead acid SOC estimation

Wow, that was a big drop in the SOC estimation, from about 75 down to 20. Dropped into the battery protection mode and connected to the grid even though it was still on-peak hours. I just happened to notice looking at the monitoring. If the grid had not been available, it would have shut off...

diysolarforum.com

I'm not sure I'm following the plots and description perfectly, but,
I don't think the Midnite SOC estimation (at that point in time at least) can be too accurate. If the voltage is only about 46v, while being charged at 20A, the SOC can't really be 97%. The much lower SI estimated SOC may be more correct.

I used to have Midnite Kid and Whizbang Jr shunt in a camper, and the SOC estimate always seemed good. Maybe the Midnite & SMA are not seeing the same data for current and/or voltage. But even then, SOC estimation is going to be imprecise sometimes, requiring a correction that won't happen the same way in both devices, so there will be a disagreement at some points.

I would love to understand more about what CAN messages are being sent into the SI and what they do. Are the CAN (or rs485?) messages setting the SI battery charging current limits based on the Midnite estimated SOC? That could be useful! Although, if I get around to it, I'm pretty sure I can change parameters in the SI over rs485 / yasdi to accomplish that. Currently I run out to the garage often to change settings to make it do what I want overnight.

 

[Hedges]

Haven't found IR or Peukert figures in SI menu, but there is battery lead resistance. That could let you program it to expect lower voltage with current.

SMA SunnyIsland 6048 lead acid SOC estimation

Wow, that was a big drop in the SOC estimation, from about 75 down to 20. Dropped into the battery protection mode and connected to the grid even though it was still on-peak hours. I just happened to notice looking at the monitoring. If the grid had not been available, it would have shut off...

diysolarforum.com

 

[daklein]

Haven't found IR or Peukert figures in SI menu, but there is battery lead resistance. That could let you program it to expect lower voltage with current.

Great idea! Increase that a little bit and see what happens to SOC resets.