Other Bubble Be Gone!

With regard to power outages, I think you may be conflating experiences with the two different plugs. While Wemo plugs do loose state following a power outage, Kasa plugs retain their prior state.

1 Like

I just went back and found the email thread, and I think you’re right. I did find a lot of conversation about the fact that the thing didn’t work with computers, and a side gripe about the fact that there is no way to administer, or use the device from a real computer, so even if it had been able to accurately measure computer power use, I probably would have returned it because it also couldn’t be used from a computer. Computers are such a big part of what I do that I just have no patience for something like this that doesn’t work with a real computer. I almost returned the Sense device when I realized that the browser app was crippled, but I was promised improvements, so I foolishly(?) kept it on the strength of that and many other great things yet to come.

I’m wondering if you could be confusing Sense accuracy with smart plug accuracy(?)

In my own limited testing, Sense does a fine job with accuracy, and that is borne out by the fact that it’s pretty close to the power that I get billed for. But both smart plugs were so far off that it was silly. If you have a different experience, then good for you, but that was not my experience, and I did spend a fair amount of time staring at it in disbelief. And I also spent a lot of time trying to get a resolution from both manufacturers. I don’t don’t know what more I could have reasonably done.

The biggest issue I see in comparing results of power monitoring devices like Fluke meter, Kill-A-Watts and Kasa is doing a true an apples-to-apples measurement comparison. The problem is that there are different power measurements as I alluded to before: Real/True Power, Apparent Power and Reactive Power. You’re a bit unclear on which of those you did your comparison.

There are also underlying artifacts of each measurement technique. You can’t really do a good comparison without looking current and voltage samples per second, zero crossing technique, power/energy calculation period, and calculation technique.

Based on what I know, the Kill-A-Watt Real Power is typically very close to the Kasa Real Power. But if you were doing as you suggested, trying to see Reactive Power or Apparent Power, Kasa isn’t your tool and you never should have expected the numbers to be the same except for resistive loads, as you suggested.

Well, I think most people who care about measuring household power consumption one know what a Kill-A-Watt is, and for a cheap meter, P3 did a pretty nice job with accuracy, when using the meter in the way that most people would use it. It’s nothing fancy, but it’s a sort of “standard”.

One of the other meters I was using was my Fluke 179…the specs are here if you’re interested:

I also compared with my UEI DL489:

I also have some other meters and a scope, but the above are the ones I find myself reaching for when making simple comparisons like this. All three of these rather different meters agree closely when measuring computer loads. They also agree closely with the Sense monitor under those same conditions.

The smart plugs were the outliers, often reading ridiculous numbers like 30 watts when the other meters where all reading closer to 300 watts. It was quite dramatic and inconsistent. Sometimes it would read accurately for a while and then I’d look again and it would be off by a factor of 10 or more. Sometimes it didn’t show any power use at all.

If you want to blame it on labels like, "Real/True Power, Apparent Power and Reactive Power, I don’t mind, but in terms of what most people mean when they are measuring household loads, the smart plugs were a failure. I agree that all of those factors you mentioned can affect the measurement accuracy, but we’re just talking about plugging a computer into a smart plug, so we don’t have control over any of those factors. Plug a smart plug into a Kill-A-Watt meter, and both are measuring the same thing. Pass it though a Fluke meter, and you’re still measuring the same load. Put a clamp meter on the Fluke pass-through and now all of the maters are reading the same load in the same way.

And out of that cluster of instruments, all agree except for the smart plug. The smart plug doesn’t even agree with itself, reading 30 watts one minute and 300 watts the next. And then reading nothing at all. I stare at the smart plug saying zero, then I stare at the computer screen where i can see that the test is still running. The computer is telling me how many watts it should be using, and if I back out the PSU inefficiency, that software number matches all of the meters almost exactly.

The smart plugs, however, exist in an alternate universe. In my universe, the computer uses 300 watts, but in the alternate universe, it uses zero watts. Maybe power is somehow traversing the brane between universes, or maybe its drawing power directly from hyperspace, but in any case, it uses no power in the smart plug universe.

It is surreal to see the stupid plug reporting no power use while it is so clear that power is coming from somewhere! I did this over and over again thinking that I must be doing something wrong with the mobile app, or that it was some kind of one-off glitch with the device, but I never was able to make it work in anything like a useful way.

As you say, for the most part, we just want to see it the way the power company sees it, because even if their meter is inaccurate, it’s still what we’re going to get billed for. The smart plugs failed in that regard.

Now I’m confused - neither of those meters have an AC power measurement mode, at least per the spec sheets. They do offer TrueRMS which is nice for getting a reasonable RMS read on the AC current. But if you multiplied the two together to get your power number, your result would be Apparent Power, which is different and larger than Real Power, except when the load is purely resistive (actually technically when the load has a phase angle of 0). You suggest that Real Power and Apparent Power are only labels - in reality, they are very different measurements. And if you understand the difference, plus a little bit more about RMS measurements, you might also understand why the Kasa could be varying widely while the Fluke was offering a stable power value, and both would be correct and accurate for the measurement they were doing. Sorry to be pedantic, but physics has taught me to understand the measurement itself, more than just the measurement reading.

ps: I’m assuming you were using the Kasa mobile app ? I prefer to see my devices on Kasa plugs in the Sense app because you get a reasonable time history of the readings. This is a snapshot of one of my household servers in action.

I’m not a physicist, but I think it’s fair to say that I do have a better understanding of (and more experience with) electricity and power than most people in the world. And I think you know that the power used by a computer should never measure zero, no matter which meter or method is being used to measure it. I’m not saying that it’s not different or that it doesn’t matter in some scenario, but getting bogged down in those technicalities might cause one to ignore the fact that the smart plug is indicating zero while power is still being used, and that fact just can’t be argued away.

Real Power consumption can truly be zero when the load is briefly 100% (or nearly 100%) Reactive. This happens when inductors / transformers are first energized or when a capacitor first starts charging. An AC supply tied directly to a capacitor or an inductor/transformer will show voltage and current waves, but will use almost no energy and dissipate almost no real power. I say almost because there will be a small series resistance from the wiring. Devices like power supplies and switching power supplies can indeed have moments when they are consuming near zero Real Power because they have capacitors and transformers at their inputs that create heavily reactive loads. Those moments of low Real Power are typically short lived and the Real Power can indeed jump around due to the active electronics in the PSU.

BTW - there’s also a chance that zero readings were caused by networking issues. Kasa plugs do experience occasional data dropouts.

1 Like

I’m not qualified to argue the science, but the fact remains that the smart plug results were not even remotely similar to the four other instruments that I was testing with. And it wasn’t a network issue, because it could read a 100 watt incandescent light bulb just fine. I know what I saw, but maybe they fixed it with a firmware or app update since my experience…(?)

1 Like