Investigate: Constant Flow setpoint readback via get_mode() may be unreliable/mis-scaled

[eman]

Origin

Companion issue to eman/esphome-alpha-hwr#44, filed against this repo because the code path involved is shared with the ESPHome port (which cites this file's line numbers directly in its own comments).

Problem

Note: the "pump runs at a fixed RPM regardless of the Constant Flow setpoint" symptom reported in esphome-alpha-hwr#23/#44 is the same root cause as #27 (start()/set_mode() hardcoding a default 3671.0 suffix) and doesn't need a separate fix here — set_constant_flow() already passes its value_m3h argument through explicitly (services/control.py:646-649), so it does not hit the default-suffix bug.

This issue is narrower: whether get_mode()'s setpoint readback is reliable for Constant Flow mode specifically. The ESPHome-side report observed, after switching to Constant Flow, a setpoint readout roughly three orders of magnitude off from a plausible value (0.003056 gal/min displayed setpoint vs 3.940 gal/min measured flow rate). Since the ESPHome port's get_mode_async() is a direct port of this client's get_mode(), the same potential issue may exist here.

Research findings

• get_mode() (services/control.py:370-556) reads the setpoint for every control mode from the same shared Class 10 register, Object 86 / Sub 6 (the "overall operation local request" object), and applies a Pa→m conversion only for pressure-family modes (services/control.py:441-450). Speed (RPM) and flow (m³/h) setpoints are passed through with no conversion, which is consistent with how they're written by set_constant_speed()/set_constant_flow() (also no conversion) — so no unit-mismatch bug is evident from static review.
• However, the pump's measured flow rate is read from a different register: Class 93 (0x5D) / SubID 290, TelemetryObject.FLOW_PRESSURE = (0x005D, 0x0122) (src/alpha_hwr/constants.py:72), documented in resources/GRUNDFOS_APP_BLE_DATA_MAPPING.md:198-199, 467-470 as "Verified: Extended Flow (Flow, Head, Pressure)" from the decompiled Grundfos GO app. This raises the open question of whether Object 86/Sub 6 reliably reflects a valid absolute flow setpoint in Constant Flow mode the way it clearly does for pressure/speed, or whether it can return a stale/differently-scaled value in this mode specifically.
• This has not been confirmed against a real pump from this repo's side — it's a plausible, code-supported hypothesis surfaced by the esphome-side bug report, not an independently reproduced bug here.

Suggested next step

Bench-verify with a real pump: call set_constant_flow(x) for a known x, then dump the raw bytes from get_mode()'s Object 86/Sub 6 read (services/control.py:407-419, already logged at debug level via logger.debug(f"Raw setpoint data: {data.hex()} ...")) and compare the decoded setpoint against both the commanded value and the Class 93/290 measured flow reading, to confirm whether the register is trustworthy for flow setpoints.

References

• eman/esphome-alpha-hwr#23 (original report)
• eman/esphome-alpha-hwr#44 (companion issue in the ESPHome component)
• start()/set_mode() ignore setpoint, forcing hardcoded ~3671 RPM default suffix #27 (root cause of the "stuck at fixed RPM" half of this symptom)