Use Solar CTs to detect 240 circuit usage

Has anyone tried purchasing the solar leads and using them to identify the usage on a 240-volt circuit (since no one makes a Smart Plug for 240). I am sure it might screw up some calculation somewhere, but it would also give very concrete information in the short run.

I could see it messing with the total kWh you are showing as buying from the power company. Fortunately, they continue to send me a bill with this info no matter how much I wish they would stop.

It could mess with the total price you expect to pay, but we don’t actually track that number very closely anyway, no tiered pricing, no 30-day billing cycles.

Would it be useful? Perhaps not too. I know there isn’t a way you could tell Sense to make use of the data, but I could. I am relying on Sense to figure out what power is going where and I am just trying to help it guess. If I could actually measure specific circuits it would help me know more about my guesses.

My example is my hot tub. I know it works connected to my tub as before I forced the CTs into my main panel I tested if I had bought a dud on eBay by hooking it up in my subpanel. I got to watch EVERY nuance of the tub for several days. I know that it ALWAYS draws 57 watts. I know that when the heater is on it draws 4kWs.

But since the Hot Tub was identified it seems to think it only draws about 1kW and it puts the other 3kWs into Other. And it TOTALLY hasn’t figured out the 57 watts of Always On, but then I haven’t flipped the breaker on the tub so it really is always on, so no reason for it to see that as part of the tubs draw, per se.

This makes me think there are other circuits that I would like to look at in the same detail, or perhaps examine the tub some more. I forgot to check out the tub with the three pump levels on and the lights on. I will probably get a solar system in the not too distant future, so I would eventually use those CTs for their actual purpose, but in the meantime, any reason not to get them early and use them in this way?

1 Like

My vote would be: get another Sense.

That way you can relax in the hot tub looking at your waveforms (sooner rather than later) knowing precisely how many watts you are using without messing up your panel’s Sense data.

I have 2 Sense’s (as mentioned elsewhere) and I’ve used the “spare” in various configurations (including in a commercial setting to make money from it). Now it’s on my hot water tank tracking long showers.

For a while I had it side-by-side on my panel feeds: they were keeping one another honest. In my case, it seemed the earlier model was giving me different voltage readings and support re-calibrated things. I’m a firm believer in redundancy.

When you get solar and throw in an EV, having 2 will help.

If I had an EV I would definitely add a dedicated Sense to the charger … having a precise longterm track of charging has got to be super useful data when the batteries start their inevitable wane.

Hmmm… $299 vs $50… Also, two Sense monitors cannot be logged in at the same time.

As to tracking the Tesla’s energy usage my understanding was that they were recognized by Sense already (although I haven’t charged my car since I got Sense working so mine isn’t recognized yet). Additionally, the car tracks it’s own charging so all I will learn when it is recognized is my chargers efficiency.

I have seen elsewhere that Ryan has recommended others to use the Solar CTs to track a circuit because he had seen others do it here. I am now shopping for cheap CTs based on this.

Can it be done? Kinda/sorta. It’s absolutely not the designed purpose of those CTs and you’re right that the app UI will not understand what you’re doing. Still, you’re more than welcome to experiment. I certainly don’t recommend it as a way to monitor 240V devices/circuits, but you’re welcome to give it a shot!

I hear you.

Although:

… you can have 2 separate Senses and separate accounts as I do.

1 Like

Didn’t catch that. Indeed you can, you just can’t switch within the app itself.

So, what is the correct current solution for monitoring 240v devices when Sense can’t handle them? Buying multiple Sense units is both expensive and doesn’t lead to an integrated energy story. Hacking HS110’s is a violation of electrical codes, although some Sense users have successfully used them that way. There don’t seem to be any alternative solutions on the market. Too bad TP-Link and Sense can’t get together and offer a solution.

@andy

A couple of solutions out there, all of them requiring CT’s for good reason (accuracy…).

That being said, I for one would purchase a ‘slave’ CT controller that was integrated directly with Sense. Smart plugs aside, there needs to be some sort of CT add on for wired devices.

2 Likes

One of the first things @israndy’s original post made me think of was “somebody must have taken apart a Wemo Insight and tapped the CTs” :grin:

Now where did I put my TA screwdrivers?

I went looking for a cheap Solar CT from Sense, and I didn’t find one. $50 everywhere. Then I thought, I have a micro-fit cable, like on the Sense. Surely all I need to do is find a similar value CT and wire them together. There are a bunch of CTs around here from various other installs. So I got to reading the blog post on CTs and discovered I needed a 3000:1 ratio in the CT to match the ones that Sense sells. I shopped Amazon and eBay again…

BOY those things are expensive. I am not upset paying just $50 for a set.

The new Clip-On sensors from Neur.io are $89. I had hoped that if I was unable to get the original CTs into my breaker box I would be able to use the ones from Neur.io but there isn’t much info on the specifications for them. I thought I would get them just to find out, but the $50 ones from Sense are just so cheap.

Well you persuaded me to play around with the Solar CT input as well: I just ordered one.
[I’d actually missed that Sense is offering them as a separate item]

Also ordered another Wemo Insight to fiddle with.

[BTW: The Sense CTs compare very favorably with my Fluke i200 (which would be about $280 for 2) – if anything I’d say they are more precise]

I did the same. I was gonna try to order with Amazon, but why pay for prime if you cannot have it the SAME DAY. So I ordered from Sense directly, help them out, give them the full profit. Would have gotten it from Amazon on Friday, wonder when it will arrive.

1 Like

I already have separate monitoring for some parts of my system, using welserver.com and add on power metering. What I am hoping for is an integrated with Sense solution, as you mention below. I would buy at least two, probably three, of such a thing…think “HS110 like” but wired-in module for 240v. I could probably do without the control aspect, though that might come in handy at times.

Or, more nodes connected to an upgraded Sense, like the Solar but for consumption not generation. Two/three more would make a HUGE difference in the value of Sense. The idea of discovering devices by their waveforms is cute, but by now it’s pretty clear that there are other needs too.

1 Like

I am hopeful that Sense can get to a point where you tell it what it is looking at on the HS110s and it will match that to what it is seeing in the full waveform to learn from the HS110. It can say “I just need 10 more power cycles” or “Just 20 more days” and then we can move the connection to the next mystery device. I know there are going to be things like the power washer that you bring out once every Spring that will take a bit of time to identify, but being able to give it more data should be able to help it learn faster.

Perhaps they could add that same extra data process to the Solar CTs for those of us w/o panels. Buy the CTs, then clip them to your Oven. After everything on that leg has been identified, it will prompt you to move the CTs to the next legs on your panel. So like buying TED, but just one leg at a time.

Then when everything is as identified as it’s gonna get, hook up to your new Solar Panel system.

I also got three more HS110s in the mail today and a 330 is on the way. Hooked up my wife’s Mitsubishi iMiev it was never going to figure that out. And the 330 for my stereo. This is becoming quite a complex system.

While it would be nice if Sense could learn better from devices hooked to other monitoring systems, I’m not optimistic about that….totally different time scales. Your idea of them using the Solar CT’s as a learning tool is a particularly interesting one, as is the prospect of suing it for a single impossible to detect 240v device…Sense Engineering, are you listening?

My solar is already so well monitored by the Enphase Energy application and by my welserver system, that I could readily remove Sense’s Solar CT’s and use them for a different device or combination of devices.

Where things like the HS110 and HS330 really add value to the overall monitoring is for things Sense simply isn’t going to be able to monitor. Those include “always on” and devices whose waveforms are simply beyond Sense’s capabilities…apparently like my well pump when a USA 240 v version of the TP-Link becomes available.

2 Likes

Sounds like you’ve ended up purchasing ours, but for anyone else reading this…

Aftermarket CTs will likely introduce errors into your monitoring. All of our CTs, including solar CTs, are picked for a particular calibration constant.

1 Like

Agreed!
There are of course many references to these kinds of devices throughout the forum.
Apart from the obvious difficult signatures like washer/dryers with complex cycles – which I admit could be learned and registered reasonably accurately – getting an OLED TV made me realize that a device like an isolated Sense actually nailing the energy use for some things really is impossible. It doesn’t matter that Sense might know through NDI that the TV is on … without knowing what images are actually on the screen it’s not going to ever be able to distinguish the waveform with any great degree of accuracy because the usage is essentially randomly varying by tens of watts.

Some of us have unrealistic expectations of what Sense-alone can achieve. Of course the more virtual data you can feed it (normal washer waveform from the last time the Smart Plug was active; the TV channel I’m watching!; temperature in the room etc) the more it’s going to be able to learn but just like a human often you’ll need hard data in the form of a Smart Plug stream if you expect accurate wattage tracking.

On a macro level: Sense should always be within 1 or 2% of overall actual usage.
On a micro level: 1-20% seems like a better expectation.

1 Like

I certainly understand that it would be difficult to write software that would be able to recognize the waveforms out there, but I also am a fan of AI and the great strides that have been made. Elon’s great example of the AIs that are sat in front of a video game and given no rules and in a very short time it’s improving exponentially, it’s become unbeatable.

Assuming that this is the strategy that Sense is using I actually do have high hopes for the future of recognition. It will eventually have a finite list of devices that it can expect to see. It can then recognize a pattern and assign a probability to the patterns. Additionally, those patterns can be compared with others who have the same individual items, but in a completely different mix.

Updating this thread with a post made by @kevin1 that I missed.

A clear answer to some of my speculations.
in general that thread answers some questions posed here.

1 Like