Technical Support

1.) RCD mA sensitivity for main/GRID supply vs LOAD/UPD/essentials? Are there any recommendations about RCD mA ratings to avoid nuisance tripping? E.g. upstream main RCD of 100mA, and downstream LOAD/backup/essentials circuit branch behind a 30mA RCD (30mA is required for all sockets/plugs by our regulation in South Africa/SA). 2.) DC coupling in grid-tied mode and RCD selectivity between GRID/non-essentials and LOAD/UPS/backup?

Will using 30mA RCDs for both GRID/non-essentials and LOAD/backup/essentials result in an RCD selectivity issue between the Grid and Load circuits? Specifically when in grid-tied mode, as the circuits are not separate/isolated, but coupled via relays?

From the circuit diagram in the install manual (which I may have misinterpreted), it looks like inverter Load/UPS side is the switching type, i.e. line-interactive (and not Online Double Conversion, since that would complicate the design for doing extra AC grid power passthrough when grid-tied, e.g. bumping up LOAD circuit output from 21.7A to 35A for the 5kW model)? I also assume this because I've seen some spec sheets refer to a 10ms transfer time. So when running in grid-tied mode, the L & N of both the GRID/non-essentials input/output and LOAD/UPS/backup output are coupled and so any earth leakage current would affect both RCDs on the LOAD and GRID circuits? 4.) Can resuming grid-tied / grid AC coupling cause a tripping issue if the external Neutral-Earth bond relay is too slow to respond to the ATS signal?

Does the ATS signal to the external relay have a timing risk that the system has 2x neutral earth bonds exist for a short while (some ms)? With one main N-E bond reconnected at grid side when the inverter switches back to the AC coupled grid tied mode, and a 2nd N-E bond still lagging (if the local relay hasn't disconnected yet), that can cause RCDs to trip. For this brief period, the ground/earth wire will now be in parallel with the neutral between the main DB N-E bond and the inverter ATS triggered N-E bond relay/contactor causing some current that should have been on N to flow via E and tripping the supply/mains/grid RCD because N-E was temporarily bonded behind it? For a short while, the rule of having only one N-E bond and the bond upstream of all RCDs is broken if the timing is off.

E.g. if after waiting the 1 minute for supply from grid to be stable, the inverter then switches back to grid in 10ms and sends the ATS signal at the same time, but the contactor/relay on the end of the ATS only responds and opens the contact in 25ms, there's 15ms with 2x N-E bonds?

Excuse the OCD on these details, but again, it seems electricians in SA are not very experienced with hybrid inverter setups where more than one RCD is involved, hence I'm double checking some of the answers I get from mine.

Again, shocking to hear some solar installers just suggest just wiring the inverters grid input/output upstream of/without the RCD to avoid nuisance tripping when grid comes back online and this might be because there's a potential race condition between the neutral-earth bond ATS signal to open the bond and the inverters internal AC coupling back to grid (going out of anti-islanding mode). 4.) Type A or AC RCD? I've noticed some other inverters, e.g. SolarEdge, recommend using Type A RCBs, not just type AC. Is there any recommendation on A vs AC?

As a related post, https://www.sunsynk.org/forum/techical-support/transformer-less-implies-not-galvanically-isolated-and-rcb-type-b-applicable?origin=auto_suggest, it's clear a type B RCD is not necessary. 5.) Any considerations to include for the inverters built-in RCD? As per NRS 097, the inverter will have an integrated RCD as well? Anything to consider in terms of selectivity between this internal RCD and external RCDs? My guess would be that it's more user friendly to trip external RCDs for branch circuits before causing the inverter RCD to trip so that the circuit with the earth leakage is easier to identify, and only that branch is isolated.

More context from wiring standard in SA (not a question, just context for other readers) For anyone else interested in SA related regulation, I skimmed the latest SANS 10142-1 draft for edition 3 which has and RCD as a requirement for "a.c. circuits supplying water heaters." (not sure, could've been in previous editions too). If this requirement includes the need to use a 30mA sensitivity RCD (as is mentioned nearby in a separate clause), then it's a dilemma, since the sane advice seems to be to just use 100mA for mains to provide selectivity, but having the water heater / geyser on the Grid/non-essential circuit might be in violation of the code if the code expects 30mA. If indeed, one must use 30mA for the geyser, and the geyser is on the grid circuit with the main RCD, then the other option might be to ensure the main RCD has a larger delay time compared to the load/UPS circuit RCD, but these delay type RCDs are quite tricky to find, not to mention costly or not available at 30mA, and likely to also have compliance issue since the RCD needs to break the circuit within a short timespan.