Are smart socket power statistics accurate? I have calibrated the power meter for a year, and the error is all in the low-power standby mode.

Are smart socket power statistics accurate?: High-power load is very accurate, low-power standby is most likely to cause false highs.

I have used several smart sockets with power statistics in my home for more than a year, and I only recently checked the accounts seriously. The reason is simple. The electricity bill is 14 kilowatt-hours more than last month. I first copied the total meter reading, and then added the sub-item power of each device in the app, but it didn't add up. One socket registered 9.7 degrees for the TV box, but I had clearly unplugged that box for less than half a month. Later, I took out the power meter and plugged in all the electrical appliances at home that could be measured for comparison. Only then did I realize that the problem was not a broken socket or a fake app, but rather the fact that its set of statistics was inherently different from one person to another.

Let’s distinguish two things first: Power and electricity are not the same thing

There are usually two numbers displayed in the smart socket App. One is the current power in W. ； The other is today's or this month's electricity consumption, the unit is kWh, which is what we call "degree". The power is the amount used at this moment, and the power is the total amount consumed in the past period of time. If you want to calculate the power of the socket, you must first read the power every once in a while, then multiply it by time and add it up. This "sometimes apart" is the key.

I have three sockets on hand, one is the Mijia Bluetooth gateway version, one is from the Tuya platform with metering, and the other is a no-name brand but the page is pretty similar. The three apps all have power and battery, but the refresh intervals are different. The Mijia unit refreshes its power once every ten seconds, the Tuya unit sometimes doesn't move for a minute, and the off-brand unit simply updates after a switch action. If you only look at the current power, it's fine. Once you multiply the power by time to calculate the battery capacity, the one with rough sampling will start to deviate.

So don’t use the current power of the smart socket as an electricity meter, let alone use it as a basis for electricity bill settlement. It is for you to see trends and find big power consumers, not a billing meter. If you really want to know the real electricity bill, there are only two things you can trust.: The electricity meter from the power supply bureau, and the power meter you plug in yourself.

The comparison I made: Three sockets to one power meter, the error is one order of magnitude

I use a Nortel power meter that costs just over 200 yuan. It has power factor and cumulative power display. It is not connected to the Internet and the readings are much more stable than the mobile app. The measurement method is not complicated either: In the same power strip, plug in the power meter first and then the electrical appliance. After stabilizing for two minutes, record the wattage. ； Then switch the electrical appliance to a smart socket and check the wattage displayed by the App and the power reported the next day. In order to reduce interference, I picked a few high-power pure resistors and a few low-power standby ones.

On the high-power side, the result is better than I expected. The electric kettle is nominally 1500W, the power meter reads 1486W, and the Mijia socket reads 1480W, with an error of less than 1%. ； The hair dryer is rated at 1800W, the power meter reads 1812W, and the Tuya socket reads 1788W, a difference of 1.3%. ； Youting is in two settings, the power meter reads 1980W, and the off-brand socket reads 1925W, a difference of 2.8%. This error is not large in percentage terms. Calculated into monthly electricity bills, the difference is within one or two degrees.

What's really outrageous is the low power. For the router plus optical modem, the power meter reads 8.4W and the Mijia socket reads 12.1W, which is 44% higher. ； When the TV box is in standby, the power meter reads 0.8W, and the off-brand socket reads 2.3W, which is nearly three times. ； The smart speaker is 3.1W, and the Tuya socket is 5.7W, which is 84% higher. The smaller the load, the easier it is for the error to amplify , and it reported higher, not random jumps.

There's a move here that I missed at first: Loads below tens of watts, such as routers and optical modems, must be measured continuously for at least ten minutes before reading, otherwise the power will suddenly jump when running speed tests in the background and connecting to devices. The power meter can capture the average, but the app may record it as a peak at that moment. But even if I measure stably, the error at low power is still more obvious than at high power.

Why is low-power standby most likely to cause false highs? Two technical reasons, don’t blame the socket alone

The first reason is the sampling interval. In order to save costs and not occupy too much server space, smart sockets do not sample at high speeds like professional electricity meters. Many of them record data every ten seconds or even tens of seconds. The load of an electric kettle is about 1500W when powered on, with little change in the process, and the impact of an occasional leak is small. ； If the power of a hair dryer, electric drill, or water pump fluctuates high and low, or if a TV box uses 0.8W most of the time and occasionally jumps to 3W, the sampling just misses the peak, so it may be overestimated or underestimated. My TV box is a typical example. It is on standby at night and wakes up in the background. It may convulse twice a night. If the socket just catches a 3.1W, the whole night is calculated as 3.1W.

The second reason is power factor. For a purely resistive electric kettle, the voltage and current are basically synchronized, the power factor is close to 1, and the measurement is easy to calculate. ； In switching power supplies such as routers, charging heads, and set-top boxes, the current is not a clean AC wave, and the power factor is low. The readings of cheap metering chips will drift under low current. You see that the App reports 12.1W, but the actual power supply bureau will not bill it as 12.1W because the power measurement algorithm is different. This can also explain why high power is accurate and low power is false.

One more thing don’t ignore: Smart sockets also consume power themselves. For those with WiFi, relays, and indicator lights, the standby power may be 0.3 to 1W. This part is usually not deducted separately, so if it is plugged in for a long time, the power will be a little higher than if you measure the appliance separately. If you add up a little, you can get a little more than half a degree to a degree a month, but it's generally not as exaggerated as the previous two reasons.

How to use this string of numbers without cheating yourself? The three scenes fit in perfectly

I’m not saying that the power statistics of smart sockets are useless. On the contrary, I still look at them every day. Just need to know when to trust and when to just use it as a reference. I summarized it into three points for myself, which is easier than memorizing a bunch of parameters.

Catch big power consumers: Can be trusted. For example, if your air-conditioning partner reports that the air conditioner is 5 degrees Celsius a day and the NAS is 0.8 degrees Celsius a day, whoever has more or less will not be able to reverse the verdict. Make a list of power consumption at home. Smart sockets are very useful.
Check specific electricity bills: Don't believe it. The sum of the app sub-items does not equal the total list, and the error lies entirely in dozens of small devices. When the electricity bill goes up, look at the general meter first, and then look at big items like refrigerators and air conditioners.
Look at trends on a single device: Watch a week, not a day. On any day, there are two sudden increase in temperature. It is more meaningful to check whether it was forgotten and whether someone came to stay than to look at the single-day numbers.

There is also a practical scenario, which is to catch "forgot to turn off electrical appliances". I have used it to find that the heated towel rack in the bathroom and the dehumidifier in the basement were not turned off twice. This is more than worth saving a few kilowatt hours of electricity. If the battery suddenly rises, go to the socket and take a look. Most of the time you can find the problem by following the clues. If dozens of smart devices at home are online at the same time, stabilize the router first. I used to How to handle dozens of smart home device routers at home As written in, first clarify the basic network so that sockets and gateways are less likely to go offline.

When buying a socket with power statistics, avoid two pitfalls first

If you haven't bought it yet and are looking for it, my suggestion is not to regard "power statistics" as the decisive selling point, it is just a plus. If you really want to buy it, take a look at two things first: Load specifications and platforms. 10A and 16A are not interchangeable. A 2000-watt oil heater will get hot when plugged into a 10A small socket. I wrote before What is the difference between smart socket 10A and 16A I have stepped on it, so calculate the power before buying.

In terms of platform, it is best to be able to access the same set of apps, so as not to jump back and forth between several apps. I have used Mijia, Tuya, and Aqara. Their background logic is different, but the basic functions are pretty much the same. When choosing, give priority to hard functions such as power-off memory, overload protection, timing and power upper limit. ； The power statistics are ranked behind because its accuracy is not suitable for billing.

The power-off memory is also worth mentioning separately. Some sockets will be powered off by default when power comes back after a power outage. If they are connected to devices such as refrigerators, fish tanks, and NAS that cannot be powered off at will, you must choose one with power-off memory, or change the power-on status to "Keep Powered On" in the settings. Otherwise, once the circuit breaker trips, the refrigerator may stop until you notice. I want to know more about how to restore the equipment at home after a power outage. I previously Smart devices collectively go offline after power outage The processing sequence is written here.

ending: I really want to know whether the electricity consumption at home is good. Don’t reverse the order.

I have now set a check sequence for myself.: First look at the general table of the power supply bureau, then look at the high-power appliances, and then it’s the breakdown of the smart sockets. The general meter is correct. High-power appliances determine the bulk of electricity bills. Smart sockets are responsible for helping you find out "who is running away." Don’t open the app and add up each device as soon as you get started. The more you add, the more anxious you will become, and the wrong addition will happen.

If you also encounter a sudden increase in electricity bills, first make a list of the devices that are turned on 24 hours a day at home, such as optical modems, routers, NAS, surveillance cameras, refrigerators, etc., and then compare them with the overall list. I want to know whether these devices that are powered all year round actually consume electricity. I am here Does the optical modem router consume electricity if it is left on all day long? I have measured them one by one using metering sockets, and the one that consumes the most electricity may not be the one you think. The number on the smart outlet can be a clue, but don’t think of it as the bill itself. It is responsible for saying "there is something going on here". How much is spent depends on the total list.