Refoss EM16P review: local circuit monitoring that tracks the utility meter

Verdict

The Refoss EM16P is a great whole-home energy monitor that does exactly what it claims. I believe it is more accurate than my previous Sense SM3 with Flex sensors and will ultimately provide much better monitoring, so I would recommend it and buy it again.

Scorecard

Each criterion is scored from 1 to 5: poor, limited, adequate, good, excellent. Not scored means there is not enough evidence to assign a rating. Scores are not averaged.

Setup

3 out of 5

Each clamp took a couple of minutes. The traps are real: the monitor only detects its voltage references when it starts, readings are unusable for a minute or two after a restart, and a backwards clamp can only be fixed on the device's own web page. A full install is a long job because every circuit has to be installed, identified and configured up front. Only the two mains and one branch are wired so far.

Local control

Not scored

Local control is why I bought it, and Home Assistant talks to it through an integration that declares local push. I have not tested it without internet access or the Refoss cloud, so this stays unscored.

Reliability

4 out of 5

In use since September 25, 2026. Its local minute history for November 1 to 7 had no missing minutes. Against the utility meter's readings, the aggregate difference was 1.2 percent over three October days and 0.9 percent over five November days. That is weeks of evidence, not months.

Privacy

Not scored

The device's Cloud switch is on and I still use the Refoss app. The switch can be turned off on the device, but I have not checked what it sends or what still works without it, so this stays unscored.

Value

4 out of 5

I paid $165.75 in April 2026 from a third-party seller, with the mains and branch clamps included. Most of those clamps are not installed yet, so the full value depends on finishing the install.

Why I bought it

I bought the EM16P on purpose for local control. It is the Pro version of Refoss's 16-circuit monitor; the listing I bought from called it the EM16 Pro.

It replaced a Sense SM3 with Flex sensors. The Sense was damaged by rain because of my own poor choice of installation location. The Sense team was responsive and sent me to support, and support later told me hardware sales had ended and replacements would no longer be available once the remaining Amazon inventory sold out. I was disappointed to lose a product that had been so easy to install and that I had spent years integrating into my home, including custom Home Assistant code and the emulated_kasa integration.

What I tested

The monitor is on a 120/240 V split-phase service. So far only the two mains and one branch circuit are wired; my panel is cramped and I have not yet cleaned it up to fit the rest of the clamps. Everything in this review comes from that partial install.

Home Assistant reads it through the Refoss RPC (Official Maintenance) custom integration from HACS. For accuracy I compared its mains with the utility meter's readings from an Emporia Utility Connect, first from Home Assistant's statistics in October and then from both devices' own exports in November. The details and limits are in the linked test note.

Setup

Fitting each clamp was easy and took a couple of minutes. The Sense was easier to install and configure at first. The Refoss asks for a much longer setup overall, because each circuit has to be installed, identified and configured before it is useful.

Three things cost me time. A voltage reference connected while the monitor was running was not picked up until I rebooted it. Readings for the first minute or two after a restart were nonsense. And a clamp on backwards shows as negative power and export, which can only be corrected with the circuit factor setting on the monitor's own web page.

The monitor's web page overview adds the branch channels on top of the mains, so it double counts. Its own total line is right, and in Home Assistant I add only the two mains.

Accuracy

Against the utility meter's readings, the Refoss mains read high. The aggregate difference was 1.2 percent over three October days and 0.9 percent over five complete November days. The single October days ranged from 0.7 to 2.0 percent high. Across the 159 complete hours in the November 1 to 7 export, 8 in 10 hours were between 0.4 and 1.5 percent high.

I believe it is more accurate than the Sense was, but I never ran the two side by side, so that is my impression rather than a measurement.

Local control and the cloud

The Home Assistant integration declares local push, and the monitor has its own web page with an export of its minute history and an MQTT option. It keeps only about a week of minute history locally.

Its Cloud switch is on in my setup, and its storage mode shows Cloud. The switch is described as allowing the device to connect to the Refoss cloud, and it can be turned off on the device. I have not done that yet, so I cannot say what keeps working without it.

One Home Assistant tip

The integration creates a voltage sensor for every channel. In my setup those 18 sensors were about 22 percent of everything Home Assistant's recorder wrote. Excluding all but the two mains from the recorder cut total state rows by about 19 percent, with the live values still available.

Limits of this review

This is a partial install, with two mains and one branch. The accuracy figures cover days, not months, and the Emporia path to the utility meter adds its own small differences. Local control and privacy are unscored because I have not tested the monitor without the cloud.

I expect the additional setup to be worthwhile because of the greater accuracy and more detailed monitoring.

Test notes

Read the method, results and failures behind this review.