Volta

I've seen a CBI QA17 earth leakage survive being erroneously connected between two lines in a 3-phase installation i.e. seeing 415V - didn't trip, didn't burn. So I'm not sure about the surge theory. If you ask the right CBI rep, they'll tell you they did some upgrades circa 2017 to that product's electronics. Apparently extra protection against overvoltage and more margin to avoid nuisance tripping. Thus the product ion date of the earth leakage might matter. If you have the DB space, I would install SPDs in any case. My board certainly has the two CBI green LEDs next to the main switch.

To make the system work for everyone is certainly a long shot. I'm about 50% there for this exact purpose of optimising solar use, while keeping a healthy reserve, but knowing my own patterns help. Thus the "AI" is me analysing average patters for the last month's logs. The project started out as a rudimentary SoC estimator for a LiFePO pack that doesn't have a BMS comms port. It's an ESP32 reading the RS232 port on cheap Axpert inverter. It uses the axpert's reported voltages and currents to calculate a SoC and then based in this SoC, commands the inverter to switch to bypass or activate the utility charger as needed. At present, I still manually set the ultimate reserve percentage based on the present loadshedding stage. Eskom-se-Push to be integrated next time I'm bored over a weekend. Weather data from a Norwegian site I came across years ago and just look at the percentage cloud cover.

Thanks for the ideas. It might still be under warranty (I'll go dig up the details tomorrow), so that cutting the sticker is not the first choice. If no warranty, then it's open soonest chance I get. My wife might not like the destructive test if that leaves us without power... I'll try get some clamp meters to check efficiency. My early estimates were closer to 100-150W of losses based on the increase in my bill (during a time of no loadshedding, so purely standby consumption). I have power loggers on all branch circuits, so I know the output consumption, but the inverter messes with the phase of the utility input, so that CT doesn't show a reliable reading.

My new OG5.48 seems to work fine, but it makes a hissing noise - not the fans, that's well known - think of letting letting the air out of your car tires. About half the sound level as the fan(at low speed). Is this normal? I've only heard it while being grid connected i.e. in battery mode it's quiet. I've not been able to definitively tie it to active charging, or floating etc. The whitenoise as a nuisance I can live with. But my experience with other electronics is that switchmode noises can indicate problems with a converter, such as a dead capacitor or inductor/transformer that hasn't been potted well or missing EMC filters. System details: OG5.48 with 5kWh Blue nova RacPower LiFePO. No solar yet. I planned adding solar in the coming months, but don't want to proceed if there's something wrong.

I suppose you could make a case for the (proper) dedicated plug, although still a bit of a stretch on the "critical [...] data processing equipment". From the regs, the exception is "socket-outlet that complies with SANS 164-4 and that is intended only fort he connection of an appliance for critical application (such as emergency lighting, a deep-freeze, a burglar alarm, data processing equipment, or life-supporting equipment)". At the end of the day, when a child sticks that little prepaid supply cord in his mouth (you know, after having done the same with a 9V battery), will your short-cut hold up in court?

Did anyone figure out how to stop the fan running in linemode? I recently got a VMIII (Kodak 5kW) and the fan runs 30 minutes on and few minutes off, but all on bypass. Anecdotally, it runs permanently on a sunny day. Baseload in the house is about 1A, so even the passthrough current is negligible. I have it purely as a back-up and the battery is just floating(don't have loadshedding at present). I have no idea why it needs to produce heat in standby, to the point where it needs forced cooling.

The inverter definitely consumes power on its own. There is two losses, the standby losses just to have the inverter on and the charger maintaining the float charge for the batteries. And the "efficiency" losses of energy that is converted (charging/inverting). Standby seems to range between 30-100W, depending on the brand and size of the inverter. And efficiency loss 3-10%, in the inverter. Your battery chemistry will have its own cycle losses. So for a fairly low consumer like @ShanoMac, the standby loss of 100W x 24h works out to 2.4units extra on the 8kWh prior. That's about a 30% increase! That's without any loadshedding charging/discharging at all - just owning the system. For heavier consumers, the standby dilutes away and you are left with the much lower percentage of inefficiencies. These losses are still there if you have solar. But since the panels don't send you a monthly bill, it's not that noticeable 😉

The dedicated CIU plugsocket sounds like a very expensive workaround? I also have all the plugs on inverter. That means the earth leakage can be downstream of the the inverter and work correctly. That upstream single CIU plug would need its own earth leakage to be compliant. I've got the keypad in the garage, where the sub DB is not on essential load, but already has the EL.

Yes. It's not unusual for older earth leakages to trip due to the inrush/surge effects when power is restored. Either through loadshedding or when an inverter switches its bypass relay. I've had two occasions where removing a standard SE RCCB and putting in a new CBI EL (QA17 and a QF14 for a DIN board) solved the nuisance trip. I stand by the CBI stuff. It's SA designed - they understand our conditions better than the imports. Let us know what works out for you.

I can confirm that an Axpert VM III (Kodak branded) can communicate with my brand new BlueNova Racpower's built-in BMS on the Lib setting (the RacPower came with the cable in the box, labelled Axpert at one end - nice touch BlueNova👍). Interestingly, on the Pylontech setting(which was my first guess), it doesn't go to error61, but do a show a battery fault code, so the communication is there, just not correct.

It's not strictly required for the SPD to be rated below the MPPT's max. Remember that SPDs doesn't clamp anywhere near the operating voltage. With a Uc of 600V, you'll find a Up rating on the SPD that is easily 3 to 5 times higher e.g. 2.5kV. That's the actual clamp level during a full surge event. The MPPT must (hopefully) have internal protection to survive the remaining ~2.5kV spike, despite a input rating of e.g. 500V. So it's less about the Uc, and more about the Up. For optimal protection, look for the lowest Up with the highest clamping current In. While making sure Ucpv>= open circuit PV voltage.

Earth leakages doesn't "see each other", but they do see EMC spikes in switching events - and not all ELs are created equal. I'll bet if you replace the GEWIS with a CBI one (like you've got on your input), your problem will be solved. Friend had the same issue - worked for him. Very easy fix.