___

That's actually better than I expected 🙂
But yeah, Hager isn't priced so much higher that you can make a cogent argument for the cheapie. I pretty much won't touch an AC-DC branded breaker anymore.

I have a second-hand story of my own. Just outside Windhoek they were drilling this borehole. The site was right next to an old cement feeding trough used by animals, again the diviner said it had to be there. A drilling operator with a slightly larger than average circumference was working the machine, and the client was standing around with a calculator attempting to figure out how much money he had to continue drilling. At that moment they drilled into an underground cavern and the entire spinning train of drilling machine dropped by a some distance while at the same time water was spraying everywhere. The client had such a fright, he just dropped everything and ran away some distance before turning around and cautiously returning. Water spraying everywhere and the drill operator had gone missing.
After eventually turning off the air pressure, they found the operator. He was lying on his back in the feeding trough, like a tortoise, unable to get back up.

I changed mine to a Thanks. Summoning @DeepBass9 to remove the last sad facey.

Yup. CAN-bus is a BUS. It's two wires running from the one end to the other, with 120Ω terminators at the ends, and many modules can connect to the CAN-H and CAN-L of this bus. There is a whole arbitration and collision avoidance protocol, and everything essentially waits for a gap and then they send their data.
It's a bit like those old 10base2 Co-ax network cable setups (anyone remember those, or is it just me who is getting old?). Or perhaps another analogy, it's like your television RF cable... you can throw a splitter on it and add another tv at te end.
The only thing you have to ensure is that the bus is terminated at both ends. Often there is already a terminator built into the ends (sometimes there is a DIP switch to enable/disable it), and on a short run one missing terminator usually doesn't break the bus... but it's a good idea to ensure you don't break the termination.

10% is 23V, so it could be as low as 207V and still be in spec. If you then allow another 5% under load, you're a tad under 200V. You could also be 10% over or 253V. So the full range then is almost 60V. Just doing the math here... I have no real experience with actual installations :-)

Just one correction to the topic: This is a LiFePO4 battery, not a lithium ion battery 🙂
Lithium Ion is what your phone and your laptop has. Usually less than a thousand cycles before end of life (depending on how hard you work it). LiPo (Lithium Polymer) is what you get in Vape pens and Drones and toys and so on. LiFePO4 is used in some vehicles (it is considered not to have a good enough energy density) but mostly for stationary power applications, aka solar power.

Disclaimer: This info provided on the basis that I think it is accurate. This might be specific to the unit I had in hand, improvements might have been made, etc. Though I doubt it, the adverts on gumtree still show one thing I consider to be a massive red flag. I'm also not an electronics expert, but I know enough to know what I'm looking at. I also looked at a Microcare unit before this came across my path. Here goes.
I was sent a blown unit by someone up north, after I asked him to send it to me. I was curious what it is like and I heard of his problems. He had two of these units, both failed, and the supplier initially repaired them but started to ignore him after the second failure.
So to summarise, in case you want the quick answer: It's rubbish. Don't buy it. If you want to support local industry, buy a Microcare unit. The microcare unit is significantly better quality. The MC unit also uses an opto-coupler to isolate the control circuitry from the big power part, while the WRND simply drives all units directly.
Longer answer. When I first took it from the box, I immediately noticed the screws used to hold the LCD in place. Tapered screws sitting on top of a flat surface. Shoddy. Opened it up, the electronic design is nice. It's a synchronous buck converter, which means it's a bit more modern than the Microcare setup (which is async). One of the FET driver chips blew. It is unclear to me why it did that. The FETs were tied into a heatsink with no mica insulators, the only insulation was the heat compound. Insulation is important: The drain of the FET is connected to the tab on the back and the two FETs are not at the same potential. Split washers were used on some bolts that damaged the circuit boards as it was tightened down. The main current carrying bolts and rails are insulated from the metal case only by a small sliver of double-sided tape. The 120A unit I had had 4 buck converters in parallel with three long rails carrying the current between them. One of the rails was shorter than the others and the difference was made up using a piece of copper wire.
It came across as a good idea badly executed. Almost like a good electronic engineer who really shouldn't do the assembly himself, that sort of thing. I don't know and I always worry that the guy who makes them will find me and yell at me. He lives somewhere in the Northern suburbs.
Edit: A quick search on gumtree, I can see changes (and likely improvements) were made. The screws holding the LCD look better too.

Put them on the back of your bakkie, drive into town. Park somewhere in an area known for alternative shopping. Go and have lunch.

Low voltage (less than 60V). Not required to earth it, and you should certainly not do it unless the inverter maker says so.

So now let me get into the HF vs LF design difference.
Low Frequency designs work like this. You take your 48VDC, and you convert it to 48VAC at 50Hz (a little bit less really, there's some losses in this part). Then you feed this 50Hz low voltage into a big old conventional transformer and on the other side pops out 230VAC. The transformer needs to be big, because the time period t = 1/f is relatively long when f = 50Hz, so you need a nice big store of magnetic energy.
A high frequency design works similar, but it has an extra stage at the end. You again start with your 48VDC, and convert it to 48VAC... but at a MUCH higher frequency (typically 40Khz and above). This also doesn't have to be a sine wave. You then feed this into a transformer again, and convert it to a higher voltage, and then you rectify it back to DC, so that you end up with around 350VDC. This is the so-called high-voltage DC bus that we sometimes talk about, and there is a reason why it needs to be higher than the expected 230V.
You then have a final stage that takes this 350VDC and switches/slices it into a sine wave, and voila, you have 230VAC (RMS).
Because your frequency is much higher, the time constant t = 1 / f is much smaller, and hence a smaller magnetic store is needed.
Also, why 350VDC? Because the 230VAC we are used to is actually an average, an "RMS" value. It's the equivalent DC voltage if you will. Visually, you could think of taking the peaks of the sine wave, slicing them off, and dumping them into the valleys, and it will then level out at 230VDC. The peaks of the sine wave is actually around 325V... and this is why the high voltage DC bus must be at a higher voltage.
OK kids... class dismissed. If I got something wrong, there will be a teacher along to correct me shortly 🙂
 

Put them on the back of your bakkie, drive into town. Park somewhere in an area known for alternative shopping. Go and have lunch.

Aaah late to the party 🙂
The Victron defaults are for 16-series batteries. Pylontech is 15s. So if you have Pylontech or Dyness batteries, you have to adjust them down.
The dynamic cut-off curve is only used while the grid is connected. When running islanded, it only switches off once the low-voltage cut-off is reached, or if the BMS disconnects the DC supply.
The dynamic cut-off curve is more important for lead acid batteries. For lead acid, a single cut-off voltage is not sufficient. A 12V battery that is at 11.5V with no load is almost completely empty while one at 10.5V that is doing a 0.2C discharge may be more than half full. The cut-off curve provides an intelligent way to to have a load-dependent cut-off point. It is not as serious with LiFePO4 since those batteries have much better current delivery capability.
When you have a BMS in the system that will disconnect, or send a DCL=0 request (Discharge Current Limit), you don't really need the cut-off curve. That is why the instructions call for just setting them all to some low voltage. It avoids reaching the cutoff unnecessarily.
It is important to not set the cut-foff too high. If you do that, the inverter will go into sustain mode too easily. For example, it might go into sustain at higher loads and 30% SOC, because the voltage starts drooping a bit too much.
(Sustain mode is when the inverter charges the battery at a low charge current to protect it from damage. Again, more important for lead acid batteries).
So given that you already have DCL control and a BMS that can disconnect the DC bus, it is okay to set the dynamic cut-off nice and low so it does not cause you issues lower down.
Note also that the inverter activates a pre-alarm once you get within 1.2V of the 0.002C value. If you set that value too high, you will get a pre-alarm even though the battery has plenty of juice left. So that is another reason to set it nice and low.

To add some more info to a lot of very informative info that is already here.
One cell will always get full first. That is normal. After that the balancer(s) kick in and start moving charge from that high cell to other cells, slowly pulling the others up until they are all the same voltage. Then when you discharge, one of them will be empty first. That's because the cells are not identical, and again, that is perfectly normal. When the first cell is empty, discharge stops (regardless of whether the others have charge left), and similarly, when the first cell is full, charge stops regardless of whether the other cells are full or not.
If the cells are well chosen to have very similar capacities, then after some time they will be aligned at the top, and at the bottom they will be pretty close (if you get my drift).
In a new pack there is almost always initial imbalances. This can be avoided by fully charging the cells before assembly, but it seems more often than not these days, this step is skipped. It usually takes about two weeks for the balancer(s) to sort it out.
This is where the high end BMS/inverter setups come into their own. The BMS can do a calculation. If one of the cells is at 3.6V, for example, then it can add up the voltages of all the cells, plus a small offset, and instruct the inverter/charger to charge at that voltage. This effectively holds that one cell at 3.6V while the balancer moves charge (slowly!) to the others. As the other cells catch up, the BMS will raise the charge voltage, and over time you'll get to a place where they all even out around 3.5V (which is what you want).
In he absence of such a BMS and/or charger system, you can do this manually. Raise the charge voltage slowly (assuming this is adjustable) while keeping at least one of the cells at 3.6V, and give the balancer time to work. With some batteries (those with passive balancers I think) I found that you have to also cycle the battery slightly, eg a recent BYD battery I worked with had to be cycled between 80% ad 100% a few times before the imbalance started to improve.  I suspect it depends on whether the BMS does active or passive balancing. With an active balancer, you just have to hold the voltage sufficiently high and wait, while with passive balancers it is better to cycle it. At least, that is my experience.

They may not have been charged to exactly the same level. LiFePO4 cells have a very flat voltage curve, so even if they measured similarly before assembly, they may have had widely different states of charge. You'd be amazed how often this question is asked these days. I have a theory, not sure how plausible, that the increased demand means that the battery makers spend less time making sure the cells are balanced, and essentially expect the balancing to happen later after commissioning. As a result it is very common for Pylontech banks to initially report different SOC levels on each module, and it is common for BYD batteries to raise high voltage alarms in the first two weeks.
Some BMSes have impressive balancing capabilities, eg I once saw an active balancer that could pass 2A between cells. The engineer working on it said to me that he thinks it is overkill: The only time that capacity is going to be used is in the first few days... after that it is essentially a waste of money.

Not by the correct definition of the term, no. Although I've had in depth arguments about this with people who insisted that it was (based on a different definition of Hybrid).
A hybrid inverter is like a grid-tied inverter, but it can also do battery storage and work off-grid if there is a failure. It can "mix" power so to speak, so when the battery is low or there isn't enough PV, it would take the difference from the grid. Examples are the Victron Multiplus, the Goodwe inverters, Voltronic InfiniSolar range, Imeon inverters, and so on. The Axpert can't do that.
Some people define Hybrid as "can also charge the batteries and power the loads from the grid". Others interpret Hybrid as "has a solar charger and an inverter in one unit". Hence the arguments.

We did the math in a previous thread. Sending around 3kw worth of power at 100V (approximately, it makes the math easy)  is 30 amps.
Off the top of my head, 25mm^2 has a resistance of 0.9Ω per km. You have to use the total distance (there and back), so for you it is 0.2km, or around 0.2Ω total. 30 amps over that is a 6V drop, or 6%. (Not counting the 6*30 = 180W of heat you'd be dissipating).
Maximum acceptable drop is 3%. So the cables needs about 2 times the capacity, so you're looking at 200m of 35mm^2 cable to even get at a remotely workable number. You don't want to know how much that is going to cost... you can buy a good inverter for the equivalent cost.
Two ways to go about this. The first is to forget about the Axpert and get a Infinisolar with high voltage DC inputs, so you can use thinner cable. The second is to use a grid-tied inverter and pass the power over the existing 230VAC wiring.
Tying all your panels in series will run them around 400V and a much more manageable 8 ampere or so, which again just using the square rule says now you can get away with 10mm^2, about a 1% drop (4V or so out of 400V), and a 30W heat dissipation.
Or buy a 3KW Fronius and with some software hackery, make the Axpert on the other end charge the batteries from AC during the day (so it essentially charges with the power generated by the Fronius).
Or skip the hackery and .... you know... get another inverter that already works well in such a setup :-)

Yes, I've recommended him before and some forum guys here have bought from him. I've bought from him more than once, and everything I bought still works. Unlike some stuff I bought from other suppliers including some "reputable" building suppliers.

So now let me get into the HF vs LF design difference.
Low Frequency designs work like this. You take your 48VDC, and you convert it to 48VAC at 50Hz (a little bit less really, there's some losses in this part). Then you feed this 50Hz low voltage into a big old conventional transformer and on the other side pops out 230VAC. The transformer needs to be big, because the time period t = 1/f is relatively long when f = 50Hz, so you need a nice big store of magnetic energy.
A high frequency design works similar, but it has an extra stage at the end. You again start with your 48VDC, and convert it to 48VAC... but at a MUCH higher frequency (typically 40Khz and above). This also doesn't have to be a sine wave. You then feed this into a transformer again, and convert it to a higher voltage, and then you rectify it back to DC, so that you end up with around 350VDC. This is the so-called high-voltage DC bus that we sometimes talk about, and there is a reason why it needs to be higher than the expected 230V.
You then have a final stage that takes this 350VDC and switches/slices it into a sine wave, and voila, you have 230VAC (RMS).
Because your frequency is much higher, the time constant t = 1 / f is much smaller, and hence a smaller magnetic store is needed.
Also, why 350VDC? Because the 230VAC we are used to is actually an average, an "RMS" value. It's the equivalent DC voltage if you will. Visually, you could think of taking the peaks of the sine wave, slicing them off, and dumping them into the valleys, and it will then level out at 230VDC. The peaks of the sine wave is actually around 325V... and this is why the high voltage DC bus must be at a higher voltage.
OK kids... class dismissed. If I got something wrong, there will be a teacher along to correct me shortly 🙂
 

Correct. Two wires come from the street to the connection point at your house. The neutral is usually already earthed at the transformer, so the neutral serves as both neutral and earth (it is combined in other words, hence the C).
An earth spike is installed nearby and a cable goes from the earth spike to this same connection box on the back/side of your house. Neutral is then bonded to earth at this point, and then from that point onwards earth is a separate conductor (hence the S in the moniker).
So TN (terra neutral), C (neutral and earth combined initially), S (but separate after the bonding point).

New firmware is on its way that can do that. It cannot report individual levels, but it can report the lowest and the highest cell, which actually tells you everything you need to know.
24V batteries already support the low/high diagnostic info.
You're right that the battery stores very little extra above 3.45V per cell. But as Youda determined here earlier in this thread, you need 3.48V per cell to really light up the balancer... plus a little extra. The Pylontech documentation actually says 52.5V to 53.2V.
That's my next suggestion. That might even tell you where the high cell is. You could then also check the temperature of that cell, chances are it runs a little higher than the rest.

That alarm comes directly from the BMS of the battery (communicated via CAN-bus). The Venus-GX is just relaying it. The "Battery monitor (512)" bit tells you the source of the alarm. The BMS on a Victron system almost always has deviceinstance=512.
In some cases I've seen internal alarms caused by a cell imbalance, but that was with a different manufacturer (interestingly, this other manufacturer uses the same CAN message for alarms and warnings, so maybe there is something there). The fact that a short disharge stint (failing the mains) resolves it seems to agree with this observation: Cell imbalance.
The pylontech protocol also doesn't have a specific flag for a cell imbalance (batteries that use the more standard 0x35A message do), so that is another reason to think it might be piggy-backing on this alarm.

Cell imbalances should disappear given enough time. The Pylontech balancers can only pass around 50mA... so it might take a really long time.

Just one thing I want to comment on, and that is the idea that RCDs don't work in non-TN earth systems (TN, terra-neutral, earth and neutral is at the same potential). RCDs are used even in IT systems (no bonding). If there is an imbalance of 30mA caused by some kind of earth fault (probably through a human being), it trips.
There are two reasons for bonding the system. One is to limit the maximum voltage. As an example, it is possible for neutral and live to both be thousands of volts above earth while still only being 230V RMS between L and N. By tying one end to earth you know the highest voltage in the system is 230VRMS and it's on the live wire.
The second reason is to help detect earth faults. With an unbonded system, the first earth fault goes unnoticed, but the second one closes the circuit and causes current to flow. By bonding the system, a single earth fault can be detected.
There are many ways of referencing a system to earth. There is even something in three-phase systems called corner-earthing, where you have a delta transformer (aka three wires, no neutral), you will tie one of the phases to earth. Again, this is to limit the maximum voltage (some part of the system corresponds to 0V) and aid in detection of earth faults.
Now note that even in unbonded systems with an undetected existing fault, an RCD still does the job. If 30mA  of current passes through a human, the RCD will still see it and trip.
Now... what's up with the V-0-V configuration? If you earth the central point, you still have some part of the system referenced to earth, the maximum voltage in the system is now 120V, and if there is a low-enough impedance path to make 30mA flow, your RCDs still trip. So on the upside, there is no reason to consider a V-0-V setup as unsafe in principle.
BUT... South African regulations does not allow you to do this in a residential setting, and your neutral wiring is also not really neutral anymore... it is a second "hot" like it would be in North America.

There is this thing about American law. That's if I even get it. If I sell something to you, and it doesn't work, it's my fault, and you can sue me. If however I sell you the PLANS (blue prints, etc), an it doesn't work... then it is YOUR fault. So every one of these scams work by not selling you a product but instead selling you the plan. The marketing material always have this huge emotional angle to it, or sometimes it will do the old spiel blaming the oil companies (you know, they don't WANT you to know, it's a conspiracy you see!), and then right at the end there will be this thin bit about how it works. Sick of it. HHO and engines running on water and fuel saver devices and charge pumps pretending to generate energy out of thin air...

Not on all models. Some have it built in (they made one of the relays double-throw), others with new anough firmware has the setting as above, and others don't have it at all.

That is true, but the values it sends are practically static. It sends a static 53.2V as a charge voltage (and 52.5V is a better alternative), and the charge current limit (CCL) is 25A per brick, which it lowers to 12A when it gets to 99% full. If you exceed the 12A... nothing happens. The battery is pretty forgiving about everything except voltage: If you exceed 54V, it switches off.
In other words, there is almost no difference between following the instructions of the BMS... and just setting 53.2V and a 25A (per brick) limit in your inverter config. The latter will work just as well as the former in the vast majority of cases.
It is a good thing to connect whatever means of control the BMS has, but what I mean is that the BMS should not have to send CCL=0 to tell me to stop charging, unless it's an emergency. Ideally the BMS should only send CCL=0 when it's absolutely imperative, in other words, when damage will result if I don't, and then it should activate its own self-protection if the inverter/charger does not comply.
A two-signal analog BMS can send a CCL=0 (and DCL=0, discharge current limit) just as effectively without the cost of the CAN-bus electronics.
There are many BMSes that send CCL=0 when the battery is full (and it is my opinion that they should not, they should not intervene in this way). This includes BYD B-Box pro as well as the new 24V Pylontech batteries. If you do that, you get a sawtooth voltage and SOC chart, because the charger stops charging at 100%... then starts again at 99%, causing these little micro-cycles at the top. Voltage control is better. Once you reach the target voltage, current automatically stops flowing because there is no potential difference.
So don't get me wrong, I do like the CAN-bus bmses and the communications. It is a good idea. You get SOC tracking from the horses mouth, you get information about cell temperatures, some BMSes can even tell you the bitmask of the balancer (which cells it's bleeding off), and of course it can communicate alarm conditions. But it's not like you cannot live without it.
On my system I have Victron LiFePO4 batteries with a VE.Bus BMS. This is essentially a two-signal BMS. Not a day's trouble...