A-hah! A new topic. I'll add some pointers back to the other relevant ones:

This question has arisen from the discussion Anyone want to share costs on batteries? where we've been looking at off-grid electricity storage.

@evermore I think we'd appreciate photos of that plastic box design in the process of assembly. And I'll be amazed if steel bars just 1mm thick will increase the likelihood of that populated box being lifted!

When evaluating what current protection mechanisms are required for the LiFePO4 '48v' battery, we also need to consider what each piece of hardware does. @evermore has stated elsewhere that his hybrid solar/grid inverter will be a Growatt SPF5000ES, let's start with that.

Since the inverter is rated 5kW, it can charge and discharge the '48v' battery at around 100A, plus momentary surges. If there's no solar input, the maximum charge rate from the grid is just 80A, but we can ignore that for calculating protection devices.

A BMS would usually be capable of charging a battery at half the rate which it can handle discharging. So to obtain 100A charge, you need a BMS rated at 200A (plus surges).

To protect the inverter and other connected devices against the massive current which would flow in the event of a short-circuit, you need a Type-T fuse. This needs to be rated well clear of the normal charge/discharge currents. So a 250A fuse would be appropriate. The important point is that it must 'blow' open-circuit very fast if 40kA should suddenly flow - that's the Type-T designation!

To isolate the battery manually, or to protect against 'over-current', you need an MCB (Miniature Circuit Breaker). This must be DC-rated to quench arcs when the contacts are opened. Since we're anticipating a max charge/discharge of 100A in normal operation, the MCB can be specified at 125A. Coincidentally this is the largest capacity of trip which can be manufactured so as to fit onto a DIN mounting-rail.

This graphic is probably easier to follow than the text above, although they say the same thing.

To connect these devices together requires wire with a cross-sectional area of 25mm² minimum. Less than that and you risk a fire from cables over-heating.

The comment about Lumberg connectors is valid... but in the wrong topic. 🙂 That's for when we run ELV (Extra Low Voltage) equipment directly off the 48v battery, instead of via the 240v AC inverter.

I'm going to use Growatt SPF5000ES in my system, so this thread is relevant to me. However my plan is for 2 inverters and 2 (possibly 3 in the future) 280Ah battery banks in parallel, which rather complicates things (run before you can walk, and all of that).

The Growatt manual appears to recommend 150A/60V DC breakers; overload power is 5s@≥150% load; 10s@110%~150% load, so I guess 150A is 150% power for those few seconds it can last. Is it OK to rely on the trip curve of the MCB to see us through these peaks?

I found BMS selection a little tricky; ideally the BMS should communicate with the inverter (setting charge/discharge limits and charge voltage, plus propagating warnings/error conditions like "under temperature"). This is my current preference; propagating errors to the power circuitry seems the right thing to do. Get the inverter to manage the currents as the BMS transistors are there as a backup. However it does limit your BMS options.

Other people get away without this communication. Andy from Off-Grid Garage advocates for this approach; he sets the absorption voltage (3.45V per cell), has an absorption time of 2 hours and a float voltage of 3.35V per cell (see his solar controller settings page and linked video). Note that Andy does not particularly like the Daly BMS.

I'm considering going with Seplos BMS (200A, 16S LiFePO4) as they can talk to Growatt SPF series inverters and the BMSes can be paralleled. They also have a pre-charge circuit which activates when turning the BMS on. However they are not cheap and they have low powered passive cell balancing (which should not be a problem with good quality cells which have been top or bottom balanced before forming the battery), and last year they had some quality control issues (which means they've learnt their lesson).

Posted by: @chickenbig

The Growatt manual appears to recommend 150A/60V DC breakers

Those are lot more expensive than using 125A MCBs.  An MCCB (Moulded Case Circuit Breaker) above 125A will cost you about £75 (incl VAT) from China and at least £200 in the UK.

Beware that UK suppliers rarely stock 2-pole units, and that they are invariably AC only. You must have a DC-rated unit in order to quench the arc when the contacts open.

Posted by: @chickenbig

Is it OK to rely on the trip curve of the MCB to see us through these peaks?

Yes. That's what the trip-curves are for.

If you don't trust them, then it suggests you don't trust the manufacturer of that unit. And if that's the case, then don't buy from them in the first place!

Of course, if you do buy a trip which doesn't seem to handle the surges, then please post here and tell us all about it.

Posted by: @chickenbig

ideally the BMS should communicate with the inverter....  Other people get away without this communication. Andy from Off-Grid Garage advocates for this approach

I tend to agree with Andy on this one.

Having the BMS and inverter communicating via CAN-bus or Modbus will make it easier to 'inset' the parameters during initial installation. After that, the only benefit would seem to be delivering status information to you using a single App.

I can't see why you should ever need to re-configure the BMS parameters once you've got that battery bank installed.

If the concept of "inset the parameters" isn't clear, please say so and I'll explain why the settings on the Inverter and BMS are not identical.

Posted by: @chickenbig

my plan is for 2 inverters and 2 (possibly 3 in the future) 280Ah battery banks in parallel

Just have a think as to whether you intend using 16S2P or 2P16S configuration.

Your explanation implies the former, in which each bank of 16-cells has its own BMS. That's what I'm doing.

On YouTube videos you'll also see technically competent commentators putting pairs of (pre-balanced) cells in parallel, and then connecting 16 of those pairs in series. You need a beafier BMS to do that of course, because you'll want it to deliver twice the discharge current!

See this discussion on the DIY Solar Power Forum.

Posted by: @transparent

Those are lot more expensive than using 125A MCBs.  An MCCB (Moulded Case Circuit Breaker) above 125A will cost you about £75 (incl VAT) from China and at least £200 in the UK.

Beware that UK suppliers rarely stock 2-pole units, and that they are invariably AC only. You must have a DC-rated unit in order to quench the arc when the contacts open.

Yes, I was considering Taixi MCCB (TXCM1B or TXCM1Z) from Aliexpress. There is just a hint of doubt in my mind about the provenance and manufacturing standards, since Taixi do not publish "electronic-y" looking datasheets on their website.

Posted by: @transparent

If the concept of "inset the parameters" isn't clear, please say so

I'm not familiar with the phrase; my interpretation of it would be that the settings applied to the inverter are a restriction on what is allowed by the BMS (e.g. inverter stops discharging at 2.9V, BMS protects at 2.8V). But please enlighten me!

Posted by: @transparent

Your explanation implies the former, in which each bank of 16-cells has its own BMS.

Each bank will have its own BMS; each cell is monitored on its own, for improved protection against over-voltage and under-voltage. However it does cost more in terms of wiring and BMS costs.

I'm starting to consider using the Seplos Mason battery box (even if the price has recently gone up); the fuse concerns me, but otherwise it looks a decent package.

I'm afraid I don't have the test equipment, nor the budget, to test the various Chinese MCBs, MCCBs, RCDs and surge-suppressors at full operating conditions. By definition those would be destructive tests!

I haven't (yet) needed to switch above 125A, and therefore don't have an MCCB.

I am deliberately using DC MCBs from a number of different manufacturers, including Taixi, Tomzn, Cenou, Tongou, FEEO and CHZFgold.

As I'm doing a lot of tests with wiring products in different configurations, my trips are used manually more often than most other users. None of them have displayed any sign of damage due to arcing, either in themselves or in connected equipment.

Yes @chickenbig you've understood the general idea of Insetting the Parameters. The BMS is a protective device and therefore shouldn't need to be brought into action under normal operating conditions.

So if the charger/inverter is set to operate between 2.9-3.63v per cell, then you configure the BMS outside of that region, say 2.8-3.7v per cell.

Be careful when you first set up a battery. It's possible that you do have one cell in the batch which exhibits a higher voltage under charge. So if you set the BMS max voltage to be 16x 3.7v (59.2v) you could still see one cell going alarmingly high to 3.8v or beyond, whilst the total battery voltage remained below 59.2v

Over time the BMS will probably sort this out, and the cells will become more even in their potential differences. So when I first set up a battery, I deliberately configure tighter parameters on both the inverter and BMS. A few days later I can relax this once I see the cell voltages are better aligned with each other.

I've just posted more about the Zeeman fuse in the Mason box over here.

I'm late to this thread, but for what it's worth:

I run Batrium on a 16S string of 210aH LiFePo4 cells. Balancing is set to 3.45v/cell, absorption voltage is set to 55v on the inverter charger and on the mppt charger. Float is set to 53.8. No need to ever take LiFePo4 above the knee point of 3.45v/cell.

Loads of people will tell you not to float, but with DC coupled solar you need to float so that you can carry loads directly off the solar array, through the inverter, when the batteries are full at midday and the sun is still shining. If you just disconnect the MPPT from the batteries, you'll end up discharging.

Posted by: @hughf

Loads of people will tell you not to float, but with DC coupled solar you need to float so that you can carry loads directly off the solar array, through the inverter, when the batteries are full at midday and the sun is still shining. If you just disconnect the MPPT from the batteries, you'll end up discharging.

Can you please expand on and clarify this important point @hughf ?

A rough diagram and/or a photo would be wonderful if you're up to it.

The bit about "if you just disconnect the MPPT from the batteries" has a number of possible interpretations. Even though you are recommending that we should not do so, I'm trying to imagine whether you have a switch-isolator in that position, or whether you're telling us readers not to physically undo a cable.

Thanks.

Posted by: @transparent

Posted by: @hughf

Loads of people will tell you not to float, but with DC coupled solar you need to float so that you can carry loads directly off the solar array, through the inverter, when the batteries are full at midday and the sun is still shining. If you just disconnect the MPPT from the batteries, you'll end up discharging.

Can you please expand on and clarify this important point @hughf ?

A rough diagram and/or a photo would be wonderful if you're up to it.

The bit about "if you just disconnect the MPPT from the batteries" has a number of possible interpretations. Even though you are recommending that we should not do so, I'm trying to imagine whether you have a switch-isolator in that position, or whether you're telling us readers not to physically undo a cable.

Thanks.

Sure... Let me explain. I have 60a breakers on the in and out from my MPPT controller, I have a 175a breaker on my 3kW inverter/charger and I have a contactor between the battery back and the main DC bus.

This conctactor is controlled by the BMS and will pop open on various conditions (cell overvolt, cell undervolt, loss of canbus between the master and cell monitor PCBs etc etc). You could just as easily use a shunt-trip breaker and use the BMS to fire the shunt trip, but a contactor was easier for me, and it can be reset under software control, unlike a shunt trip breaker.

Most people who are not needing to carry loads on the their batteries when they are full (they have hybrid inverter/chargers, grid connected ones) will have a charge contactor which sits between the mppt and the battery pack. When the batteries are full, the BMS opens this contactor. The don't float their batteries and will tell you that you can't/shouldn't.

However in my off grid setup, if I did that, the sun would be blazing down and my batteries would be discharging running my daytime loads. So instead I set the float voltage to a value where I don't see any current flow into the cells, in my case that ends up being 3.36v/cell (53.8v float).

It works well for me, we are often in float by midday in the summer and the rest of the day the panels will keep up with the background loads (if they are there). Unfortunately I don't have enough installed capacity to charge my car at any reasonable rate, but I can sometimes manage to get a few hours at 6a charge (the lowest my EVSE can be set to).

Hope that explains it all.

EDIT: One more thing - you must (really really, you must) make sure that the mppt charger and inverter charger you choose can have their charge rate controlled/requested by the BMS over can/modbus. I can't do this and it's a total nightmate getting the settings correct so that you don't end up burning up balance boards by trying to dump all the charge current that your mppt can put out. Batrium will request a reduced charge current over canbus, from supported controllers (unfortunately no support for Outback power systems stuff, as it's too old), when cells start to enter bypass/balance. Victron stuff is a good bet. It's supported by most BMSs (Batrium, REC etc)

Like this @hughf ?

I've depicted all the switches/MCBs as double-pole, and added a mains rotary isolator as stipulated by MCS rules. That may not reflect your actual installation.

I just want to make sure that the diagram is approximately correct before I comment further.

No fuses on the DC side?

No lightning suppressors?

@transparent umm, basically, yes. I’ll do a drawing. A few shunts in there and a common dc bus, and the ac side is a bit different.

no fuses, no lightening suppression (it wasn’t a thing when I put this package together in 2005 from the outback power systems catalog)