[Bug]: AC Coupling power/energy sensors read 0 on S6-EH1P-8K-L-PLUS despite real generation (SolisCloud shows non-zero)

[IamMattM]

Inverter Model Details

Brand: Solis
Model Number: S6-EH1P-8K-L-PLUS
Connection Type: Modbus TCP via ESPHome bridge (acting as transparent gateway)
Integration version: 4.1.5
Model option selected: S6-EH1P
Connection option selected: S2_WL_ST
has_ac_coupling: true

🐛 Bug Description
AC Coupling Total Active Power and the associated daily/monthly/yearly generation sensors report 0 continuously, including during strong solar generation. SolisCloud's flow diagram shows real-time non-zero AC-coupled output (2.704 kW live, 12.61 kWh daily yield) at the same moment.

🔄 To Reproduce
Configure solis_modbus with model S6-EH1P, has_ac_coupling: true, connected via ESPHome Modbus TCP bridge
Observe AC Coupling Total Active Power / generation sensors during active solar generation while grid-connected
Compare against SolisCloud's live AC Coupled reading for the same timestamp

Expected behavior
AC Coupling power sensors reflect actual live generation, matching SolisCloud.
Actual behavior
Sensors report 0 at all times tested.
Debug log (raw register read, confirms this is not a decode/multiplier issue):

Group 34496 starting for (192.168.1.180.1)
Read Input Registers: register = 34496, count = 2
block 34496, register 34496 has value 0
block 34496, register 34497 has value 0

All other sensor groups (battery, grid, AC voltage/current) read correctly via the same connection at the same time — isolated to this register group only.
Possible explanation (not yet confirmed)
Per Solis's own documentation on the AC Coupling smart-port feature (SolisCloud Remote Control Settings guide), this feature is described as an off-grid management function, active only when grid and generator are both disconnected. If registers 34496/34497 are tied to that specific off-grid control telemetry rather than general AC-coupled metering (SmartMeter2 , PV inverter , ID:002 in the Solis metering solution below), a 0 reading while grid-connected may be expected behavior — in which case this would be a documentation/labeling issue (the sensor name implies general metering) rather than a data bug. Raising as a bug since the sensor name and lack of any clarifying note led me to expect live data; happy to have this reclassified if that's the case.

[IamMattM]

AC Coupling Total Active Power is scoped to the GEN/Backup smart-port feature, not general AC-coupled PV metering (S6-EH1P-8K-L-PLUS)

Update — found the answer: checking the device's own controls in HA (Settings → Devices → Solis S6-EH1P → entities), AC Coupling Enable is OFF, and the AC Coupling Position toggle only offers two options: "off = GEN port, on = Backup port." There's no grid-tied option at all.

This confirms AC Coupling Total Active Power (and the related registers) are scoped specifically to Solis's smart-port AC-coupling feature — i.e. managing a separate AC-coupled inverter physically wired into the GEN or Backup port, typically for off-grid scenarios. It's not a general "how much is my AC-coupled solar producing" sensor. Since my actual PV is AC-coupled through the conventional path (not via GEN/Backup), this feature is correctly disabled, and the registers correctly read 0 — there's no bug.

Reference: SolisCloud Remote Control Settings guide describes this AC Coupling feature as activating specifically when grid and generator are both disconnected.

Suggestion for the docs/sensor naming: it might be worth a note on the AC Coupling Total Active Power sensor (and siblings) clarifying that this tracks the GEN/Backup smart-port feature specifically, not general AC-coupled PV generation — the current name reads as the latter and is easy to mistake for live solar output.

[IamMattM]

AC couple scenarios 042803 3 (1) (2).pdf

[IamMattM]

Feature Request added : No support for dual-meter (Grid + PV Inverter) configuration — Meter 2 registers never polled
#425

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[pongwitr-cyber]

Same issue here — confirming on different hardware/connection:

Model: S6-EH1P6K-L-PLUS (no-LCD), integration 4.1.5
Connection: direct serial RS485 (USB adapter on HAOS/RPi5), so this is not specific to the ESPHome TCP bridge
On-grid PV inverter AC-coupled on GEN port, AC coupling enabled on the inverter
SolisCloud shows live AC Coupled ≈ 1.06 kW while ac_coupling_total_active_power reads 0 continuously

Happy to provide register dumps if useful.

[Nino6689]

Thanks both — there are two different situations in this thread, so let me separate them.

First, the naming. @IamMattM is right, and this is the real defect. These sensors track the GEN / Backup smart-port AC-coupling feature specifically, not general AC-coupled PV generation. If your AC-coupled inverter feeds in via the conventional path rather than through the GEN or Backup port, these registers correctly read 0 and always will. The current names read as "total AC-coupled solar", which is an easy and expensive misunderstanding. I'll get the names and docs fixed.

@IamMattM — what you actually wanted is now shipped. Your SmartMeter2, PV inverter, ID:002 is covered by the dual-meter support added for #425: enable "Second smart meter installed" and you get the Meter 2 block (33300-33337) — per-phase V/A/W, total power, PF, frequency and lifetime import/export. That landed after your last comment here, so you wouldn't have seen it. It's on master, not yet in a release.

@pongwitr-cyber — yours is the genuinely open case and I don't want to guess at it. You have AC coupling enabled on the GEN port and 34496/34497 still read 0 while SolisCloud shows ~1.06 kW.

My strong suspicion is that your power is showing up in the smart-port registers instead: Smart Port Phase A Active Power (34391). On an S6-EH1P the GEN port is the smart port, and that group is already polled for you. Could you check what that entity reads while SolisCloud shows ~1 kW?

If that's not it, could you post raw dumps of:

• 34440-34460 and 34490-34500
• 33530-33535 (Generator Phase A Active Power group — generator-port telemetry is a plausible alternative home for GEN-port AC-coupled power)
• the state of the switches AC Coupling Enable (43483 bit 1), AC Coupling Position (43365 bit 3) and With Generator (43365 bit 1) — AC Coupling and With Generator are documented as mutually exclusive, so a stuck With Generator bit would make 34496 invalid by design
• your DSP/HMI firmware and Protocol Version (33000 group)

Worth being straight about the history: the 34445-34452 / 34496 register numbers came from a requester's suggestion in #313 and were never verified against a datasheet, so "34496 is the wrong register for the S6-EH1P family" is a real possibility. I'd rather confirm that from a dump than move the mapping and break it for whoever it does work for.

Related: #314 and #408 report the same thing on S6-EH1P hardware.

[pongwitr-cyber]

@Nino6689

Confirmed — your suspicion was right! 🎉

Smart Port Phase A Active Power (34391) is where the AC-coupled power lives on my S6-EH1P. It reads -1860 W while SolisCloud shows the matching AC Coupled generation at the same moment. The value is sign-inverted (negative = power flowing in from the AC-coupled PV inverter on the GEN port). I'm now using -1 × 34391 in a template sensor and it tracks SolisCloud perfectly.

Meanwhile AC Coupling Total Active Power (34496/34497) stays 0 at all times.

Switch states (as requested):

AC Coupling Enable (43483 bit 1): on
AC Coupling Position (43365 bit 3): off (= GEN port, matches my physical wiring)
With Generator (43365 bit 1): off
So no stuck With-Generator bit — the switches are all in the expected state, yet 34496 never reports. This supports the theory that on the S6-EH1P family the GEN-port AC-coupled metering is exposed via the smart-port registers (34391 group) rather than 34496.

Versions: DSP version 8, HMI version 25, Protocol Version 1. Model code (SolisCloud): 3173, grid standard PEA (Thailand), commissioned 2024-08.

Setup recap: S6-EH1P6K-L-PLUS (no LCD), direct serial RS485, integration 4.1.5, has_ac_coupling: true, on-grid PV inverter on GEN port.

Happy to provide the raw register dumps (34440-34460, 34490-34500, 33530-33535) if you still need them — just confirm and I'll capture them.

[Nino6689]

That's a really clean confirmation, thank you — sign-inverted 34391 tracking SolisCloud while 34496/34497 stay flat, with all three switch bits in the expected state, is about as unambiguous as this gets without a protocol document.

I'm going to be careful about turning it into a mapping change, and I'd rather explain why than go quiet on it.

v4.2.0 shipped a setup crash that only appeared on HV-battery inverters (#445, #446, #447) because I added sensors on registers that a model-specific branch treated differently, and I don't own HV hardware to catch it. That was mine, and it took the whole integration down for those users. So the bar for "make a model-specific register change from a single report" has gone up, and it should have been higher already.

Right now the evidence is one inverter, single-phase, PV on the GEN port, Thailand grid standard. What I can't tell from that:

• whether it's the S6-EH1P family or your specific firmware (DSP 8 / HMI 25)
• whether three-phase S6-EH3P units behave the same way, or put it on a different phase register
• whether 34496 is genuinely absent, or present-but-zero because something upstream isn't populating it
• whether the sign inversion is universal or a consequence of the GEN-port wiring

If I map AC Coupling onto 34391 for a whole model family and any of those assumptions is wrong, I break working setups for people who currently have correct readings — which is a worse outcome than your template sensor.

Yes please to the dumps you offered — 34440-34460, 34490-34500 and 33530-33535. Ideally captured twice: once with the AC-coupled inverter actually generating, and once at night with it idle. That tells us whether 34496 is dead silicon or a live register that never gets written, which decides whether the fix is a remap or a firmware-conditional.

You can capture them with the service that's now in v4.2.0:

action: solis_modbus.solis_read_register
data:
  address: 34490
  count: 11
  register_type: input

(If you get an HTTP 500 on any of these, that's #450 — it means the inverter rejected the address, and I've a fix in review to say so properly rather than erroring.)

Also worth knowing: you're not blocked meanwhile. Smart Port Phase A Active Power is already a first-class sensor behind the SMART_PORT feature, so your -1 × template is reading a supported entity rather than something that might vanish — it'll keep working regardless of what we decide here.

And if any other S6-EH1P owner with AC coupling is reading this, a single dump of 34490-34500 while your AC-coupled inverter is generating would help a lot — one more data point is most of what's missing.

[IamMattM]

@Nino6689
Thank you so much for releasing the fix for Meter#2.

The value is now indeed correctly presented after enabling the "second meter option".
The value is a negative number just like like @pongwitr-cyber mentionned for his scenario off the GEN port.

I will as well, add an HA template to convert it to a positive number that the HA energy card can use to show the PV generated power as a positive number.

Thank you for promptly sorting this out.

Much appreciated.

[pongwitr-cyber]

Dumps captured as requested — both conditions, integration updated to v4.2.2 first (the new solis_read_register service works great).

Daytime capture (~13:45 local), AC-coupled inverter GENERATING — Smart Port Phase A = -1850 W at capture time:

34440–34460 (count 21): all zeros (0x0000), clean response, no exception
34490–34500 (count 11): all zeros — including 34496/34497, while ~1.85 kW was flowing on the smart port register at the same moment
33530–33535 (count 6): all zeros

Nighttime capture (~19:40 local), AC-coupled inverter IDLE — Smart Port Phase A = 0 W:

34440–34460: all zeros
34490–34500: all zeros
33530–33535: all zeros

No HTTP 500 / rejection on any range in either condition — the inverter answers all three ranges normally, they just always contain zeros.

So on this unit (S6-EH1P6K-L-PLUS, DSP 8 / HMI 25, Protocol 1, model code 3173, PEA grid standard, AC coupling on GEN port): 34496 looks like a live register that never gets written ("present-but-zero"), identical day and night, while the GEN-port AC-coupled power is reported in the smart-port group (34391, sign-inverted). Same story for the 33530 generator group and the 34440 block.

Happy to capture anything else if useful.